Top Banner
1 Underwater Optical Wireless Communications, Networking, and Localization: A Survey Nasir Saeed, Member, IEEE, Abdulkadir Celik, Member, IEEE, Tareq Y. Al-Naffouri, Member, IEEE, Mohamed-Slim Alouini, Fellow, IEEE Abstract—Underwater wireless communications can be carried out through acoustic, radio frequency (RF), and optical waves. Compared to its bandwidth limited acoustic and RF counter- parts, underwater optical wireless communications (UOWCs) can support higher data rates at low latency levels. However, severe aquatic channel conditions (e.g., absorption, scattering, turbu- lence, etc.) pose great challenges for UOWCs and significantly reduce the attainable communication ranges, which necessitates efficient networking and localization solutions. Therefore, we provide a comprehensive survey on the challenges, advances, and prospects of underwater optical wireless networks (UOWNs) from a layer by layer perspective which includes: 1) Potential network architectures; 2) Physical layer issues including propagation char- acteristics, channel modeling, and modulation techniques 3) Data link layer problems covering link configurations, link budgets, performance metrics, and multiple access schemes; 4) Network layer topics containing relaying techniques and potential routing algorithms; 5) Transport layer subjects such as connectivity, reliability, flow and congestion control; 6) Application layer goals and state-of-the-art UOWN applications, and 7) Localization and its impacts on UOWN layers. Finally, we outline the open research challenges and point out the future directions for underwater optical wireless communications, networking, and localization research. Index Terms—Underwater sensor networks, optical wireless, communication, networking, localization, cross-layer, channel modeling, link budgets, connectivity, optical wireless link layer, optical wireless transport layer, flow control, congestion control, pointing, acquisition, tracking. I. I NTRODUCTION According to a recent survey by the United States national oceanic and atmospheric administration, about 97 percent of the Earth’s water covers the surface of the earth in the form of oceans [1]. The early study of oceans (oceanography), extends back to tens of thousands of years, which includes acquiring the knowledge of ocean tides, currents, and waves. However, it was not until late 18th century that the British government announced an expedition to conduct appropriate oceans scientific investigation. The results of this expedition were published in 1882 as “Report Of The Scientific Results of the Exploring Voyage of H.M.S. Challenger during the years 1873-76 [2].” After this expedition number of books have been published on modern oceanography which includes “ Geography of the oceans [3]”, “Handbuch der Ozeanographie [4]”, “The depths of the oceans [5]”, “The Oceans [6]”, “The Sea [7]”, and “Encyclopedia of Oceanography [8]”. More The authors are with the Department of Electrical Engineering, Com- puter Electrical and Mathematical Sciences & Engineering (CEMSE) Divi- sion, King Abdullah University of Science and Technology (KAUST), Thuwal, Makkah Province, Kingdom of Saudi Arabia, 23955-6900. recently, there has been a growing interest to explore the underwater environment for numerous applications such as climate change, the study of oceanic animals, monitoring of oil rigs, surveillance, and unmanned operations. All of these applications require a medium to communicate in the under- water environment and to the outside world. In recent past years, the study of underwater wireless media has attracted much attention for underwater communications. Today, underwater wireless communications (UWCs) are implemented using communication systems based on acous- tic waves, radio frequency (RF) waves, and optical waves. Underwater acoustic wireless communications (UAWCs) have been one of the most used UWC technology as it provides communication over very long distances. In 1995, an UAWC system was proposed in [9] with the data rate of 40 kbps. In 1996, an 8 kbps UAWC system was developed for a depth of 20 m and horizontal distance of 13 km [10]. In 2005, a more high-speed UAWC system was proposed in [11] which records a data rate of 125 kbps using 32 quadrature amplitude modulation technique (QAM) with symbol error rate of 10 -4 . Furthermore, a 60 kbps UAWC system was demonstrated in [12] using 32 QAM which can support communication over depth of 100 m and horizontal distance of 3 km. However, acoustic waves still have many drawbacks including scattering, high delay due to the low propagation speeds, high attenuation, low bandwidth, and bad impacts on the underwater mammals and fishes. To alleviate the insufficient data rate of UAWC systems, research has been carried out in the past to use low frequency RF waves, e.g., the authors in [13] proposed microwaves based wireless communication system over the surface of the ocean water which can transmit data over tens of kilometers. An underwater microwaves based wireless communication system was employed in [14], which can communicate over a horizontal distance of 85 m. A similar approach was followed in [15] with the data rate of 500 kbps over a horizontal distance of 90 m. The authors in [16] have improved the capacity of underwater microwaves based wireless communication system further to 10 Mbps over the distance of 100 m. However, RF waves including microwaves suffer from serious attenuation in water, e.g., the attenuation in the ocean is about 169 dB/m for 2.4 GHz band while the attenuation in freshwater is much higher, i.e., 189 dB/m [17]. Moreover, RF based UWC requires huge antennas and is limited to the shallow areas of the sea. On the other hand, operating at ultra-low frequencies yields reduced attenuation levels, at the expense of high hardware costs and low data rates. arXiv:1803.02442v1 [cs.NI] 28 Feb 2018
40

Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

Jul 05, 2020

Download

Documents

dariahiddleston
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

1

Underwater Optical Wireless Communications,Networking, and Localization: A Survey

Nasir Saeed, Member, IEEE, Abdulkadir Celik, Member, IEEE, Tareq Y. Al-Naffouri, Member, IEEE,Mohamed-Slim Alouini, Fellow, IEEE

Abstract—Underwater wireless communications can be carriedout through acoustic, radio frequency (RF), and optical waves.Compared to its bandwidth limited acoustic and RF counter-parts, underwater optical wireless communications (UOWCs) cansupport higher data rates at low latency levels. However, severeaquatic channel conditions (e.g., absorption, scattering, turbu-lence, etc.) pose great challenges for UOWCs and significantlyreduce the attainable communication ranges, which necessitatesefficient networking and localization solutions. Therefore, weprovide a comprehensive survey on the challenges, advances, andprospects of underwater optical wireless networks (UOWNs) froma layer by layer perspective which includes: 1) Potential networkarchitectures; 2) Physical layer issues including propagation char-acteristics, channel modeling, and modulation techniques 3) Datalink layer problems covering link configurations, link budgets,performance metrics, and multiple access schemes; 4) Networklayer topics containing relaying techniques and potential routingalgorithms; 5) Transport layer subjects such as connectivity,reliability, flow and congestion control; 6) Application layer goalsand state-of-the-art UOWN applications, and 7) Localization andits impacts on UOWN layers. Finally, we outline the open researchchallenges and point out the future directions for underwateroptical wireless communications, networking, and localizationresearch.

Index Terms—Underwater sensor networks, optical wireless,communication, networking, localization, cross-layer, channelmodeling, link budgets, connectivity, optical wireless link layer,optical wireless transport layer, flow control, congestion control,pointing, acquisition, tracking.

I. INTRODUCTION

According to a recent survey by the United States nationaloceanic and atmospheric administration, about 97 percent ofthe Earth’s water covers the surface of the earth in the formof oceans [1]. The early study of oceans (oceanography),extends back to tens of thousands of years, which includesacquiring the knowledge of ocean tides, currents, and waves.However, it was not until late 18th century that the Britishgovernment announced an expedition to conduct appropriateoceans scientific investigation. The results of this expeditionwere published in 1882 as “Report Of The Scientific Results ofthe Exploring Voyage of H.M.S. Challenger during the years1873-76 [2].” After this expedition number of books havebeen published on modern oceanography which includes “Geography of the oceans [3]”, “Handbuch der Ozeanographie[4]”, “The depths of the oceans [5]”, “The Oceans [6]”, “TheSea [7]”, and “Encyclopedia of Oceanography [8]”. More

The authors are with the Department of Electrical Engineering, Com-puter Electrical and Mathematical Sciences & Engineering (CEMSE) Divi-sion, King Abdullah University of Science and Technology (KAUST), Thuwal,Makkah Province, Kingdom of Saudi Arabia, 23955-6900.

recently, there has been a growing interest to explore theunderwater environment for numerous applications such asclimate change, the study of oceanic animals, monitoring ofoil rigs, surveillance, and unmanned operations. All of theseapplications require a medium to communicate in the under-water environment and to the outside world. In recent pastyears, the study of underwater wireless media has attractedmuch attention for underwater communications.

Today, underwater wireless communications (UWCs) areimplemented using communication systems based on acous-tic waves, radio frequency (RF) waves, and optical waves.Underwater acoustic wireless communications (UAWCs) havebeen one of the most used UWC technology as it providescommunication over very long distances. In 1995, an UAWCsystem was proposed in [9] with the data rate of 40 kbps. In1996, an 8 kbps UAWC system was developed for a depthof 20 m and horizontal distance of 13 km [10]. In 2005, amore high-speed UAWC system was proposed in [11] whichrecords a data rate of 125 kbps using 32 quadrature amplitudemodulation technique (QAM) with symbol error rate of 10−4.Furthermore, a 60 kbps UAWC system was demonstrated in[12] using 32 QAM which can support communication overdepth of 100 m and horizontal distance of 3 km. However,acoustic waves still have many drawbacks including scattering,high delay due to the low propagation speeds, high attenuation,low bandwidth, and bad impacts on the underwater mammalsand fishes.

To alleviate the insufficient data rate of UAWC systems,research has been carried out in the past to use low frequencyRF waves, e.g., the authors in [13] proposed microwavesbased wireless communication system over the surface of theocean water which can transmit data over tens of kilometers.An underwater microwaves based wireless communicationsystem was employed in [14], which can communicate over ahorizontal distance of 85 m. A similar approach was followedin [15] with the data rate of 500 kbps over a horizontal distanceof 90 m. The authors in [16] have improved the capacity ofunderwater microwaves based wireless communication systemfurther to 10 Mbps over the distance of 100 m. However, RFwaves including microwaves suffer from serious attenuationin water, e.g., the attenuation in the ocean is about 169 dB/mfor 2.4 GHz band while the attenuation in freshwater is muchhigher, i.e., 189 dB/m [17]. Moreover, RF based UWC requireshuge antennas and is limited to the shallow areas of the sea.On the other hand, operating at ultra-low frequencies yieldsreduced attenuation levels, at the expense of high hardwarecosts and low data rates.

arX

iv:1

803.

0244

2v1

[cs

.NI]

28

Feb

2018

Page 2: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

2

Fig. 1: Attenuation of optical waves in aquatic medium.

Due to the limitations of low bandwidth and low data rate ofunderwater acoustic and RF waves, an alternative approach isto use optical waves which can provide high-speed underwateroptical wireless communication (UOWC) at low latencies inreturn for a limited communication range. Underwater propa-gation of optical waves also exhibits distinctive characteristicsin different wavelengths as shown in Fig. 1. In 1963, theauthors in [18] found that attenuation within the range of450-550 nm wavelengths (blue and green lights) is muchsmaller compared to the other wavelengths. In 1966, Gilbertet al. [19] experimentally confirmed this behavior of opticalwaves, which provided the foundation of UOWC systems.The research on UOWC is mainly focused on increasing thetransmission range and data rate of UOWC systems. The trendto improve the data rate of UOWC systems by using lightemitting diodes (LEDs) has been followed in [20]–[26].

All of these LED-based UOWC systems have insufficientbandwidth, thus have low achievable data rate and low trans-mission distance. Therefore, laser-based UOWC systems wereproposed in [27]–[29] which provide large bandwidth andhigh-speed data rate. A green laser with 532 nm wavelengthwas employed in [20] to provide a UOWC link which coversa distance of 2 m with the data rate of 1 Gbps. In 2015, theauthors in [30] used a blue laser with 405 nm wavelengthto provide a UOWC link with 1.45 Gbps and transmissiondistance of 4.8 m. To further improve the transmission distanceand data rate, the authors in [31] and [32] employed a UOWClink with 2.3 Gbps and 2.488 Gbps over a transmissiondistance of 7 m and 1 m, respectively. Subsequently, theauthors in [33] demonstrated a UOWC system with the datarate of 4.8 Gbps using 16 quadrature amplitude modulation-orthogonal frequency division multiplexing (QAM-OFDM). AUOWC system with the data rate of 4.88 Gbps was proposedin [34] by using 32 QAM-OFDM to get the transmissiondistance of 6 m. Recently, a 7.2 Gbps UOWC system has beenproposed in [35] for 450 nm blue laser with the transmissiondistance of 6 m. Table I summarizes the comparison betweenthree different kinds of underwater wireless communicationsystems.

Optical waves have the advantage of higher data rate,low latency, and power efficiency at the expense of limitedcommunication ranges. Therefore, networking solutions are

crucial for mitigating range related deficiencies in order toemploy optical waves for underwater wireless communicationsapplications. Furthermore, accurate and precise localizationschemes are also essential for developing effective networkingprotocols. We should also note that some application typesheavily depend upon the sensing location since the obtainedmeasurements are meaningful only if it refers to an accuratelocation. Localization in terrestrial wireless networks has beenstudied widely and detailed surveys are presented on this topic[36]–[41]. However, global positioning system (GPS) and RF-based localization schemes cannot work in the underwaterenvironment as a result of hostile aquatic channel conditions.Thus, many researchers developed localization schemes for theunderwater environment based on acoustic waves. Localizationof underwater acoustic networks have also been studied widelyin the past and a number of surveys are written on this subject[42]–[46]. Since the underwater optical wireless channel posesnew challenges, the existing localization techniques used forterrestrial wireless networks and underwater acoustic networksare not directly applicable to underwater optical wirelessnetworks (UOWNs). Therefore, novel time of arrival (ToA)and received signal strength (RSS) based distributed localiza-tion schemes are developed in [47] for UOWNs. Recently,RSS based centralized localization schemes for UOWNs areproposed in [48]–[50].

A. Related Surveys on UOWNs

With the increasing demands for UOWN applications, quitea few brief survey articles have been published to discussphysical layer aspects of UOWNs. A recent survey on UOWCsystems has been proposed in [17], where the authors havediscussed different modulation schemes, channel models, linkmanagement, and coding techniques along with the possiblepractical implementations of UOWC systems. The link per-formance of UOWC systems was evaluated in [21] and theauthors have introduced different challenges associated withlink developments of UOWC systems. In [51], the authors havereviewed UOWC systems in terms of modulation schemes,channel models, and coding schemes. The channel models ofUOWC systems have also been surveyed in [52] and [53],where the authors have considered vector radiative transfertheory, variable water composition, and inherent properties oflight. The inherent features of underwater wireless communi-cations including UOWC have been briefly surveyed in [54].The recent advances in system analysis and channel modelingof UOWC systems have been summarized in [55]. In [56] thefuture vision of UOWC systems and some of its challengeswere presented.

Unlike the surveys above which mainly tackle the physicallayer aspects of UOWNs, this paper provides a comprehen-sive survey on the challenges, advances, and prospects ofUOWNs from a layer by layer perspective which includes:1) Potential network architectures; 2) Physical layer issuesincluding propagation characteristics, channel modeling, andmodulation techniques; 3) Data link layer problems coveringlink configurations, link budgets, performance metrics, andmultiple access schemes; 4) Network layer topics containing

Page 3: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

3

Fig. 2: Illustration of a generic underwater optical wireless network (UOWN) architecture.

relaying techniques and potential routing algorithms; 5) Trans-port layer subjects such as connectivity, reliability, flow andcongestion control; 6) Application layer goals and state-of-the-art UOWNs applications, and 7) Localization and its impactson UOWNs layers. Furthermore, we outline the open researchchallenges and point out the future directions in UOWNsresearch.

B. Survey Organization

The rest of this survey is organized as follows: In SectionII, possible architectures for UOWNs are presented. SectionIII, addresses the physical layer aspects of UOWNs such asunderwater propagation characteristics of optical waves, chan-nel modeling, and UOWC modulation techniques. The datalink layer issues such as fundamental tradeoff between trans-mission angle and range, link configurations, error and datarate performance, and multiple access schemes are coveredin Section IV. Section V discusses network layer problemsincluding relaying techniques and routing protocols. SectionVI covers the transport layers topics including connectivity,reliability, flow control, and congestion control for UOWNs.Application layer goals and a number of UOWN applicationsare presented in Section VII. Different localization techniquesfor UOWNs are presented in Section VIII. Section IX outlinesthe open research challenges and point out the future directionsin UOWNs research. Finally, Section X concludes the surveywith a few remarks.

II. POTENTIAL ARCHITECTURES OF UNDERWATEROPTICAL WIRELESS NETWORKS

UOWNs can either operate in ad hoc or infrastructuremodes: An ad hoc UOWN is a distributed type of wirelessnetwork which does not rely upon pre-installed network equip-ment. Hence, traffic requests are carried out by the partici-pation of nodes along a routing path which is dynamicallydetermined based on network connectivity and may necessitateself-configuration and self-organization skills because of theabsence of a central control unit. Taking the potential con-nectivity challenges due to the directional light propagationwith limited range, realizing a full ad-hoc UOWN is a non-trivial engineering task. On the other hand, infrastructure-based UOWNs may consist of omnidirectional optical accesspoints (OAPs) or optical base stations (OBSs) each of whichcreates an underwater local area network (LAN) by servingand coordinating nodes in its vicinity or cell coverage area,respectively.

Fig. 2 shows a cellular infrastructure based three dimen-sional architecture where the underwater sensor nodes com-municate with each other and with the underwater OBSs byusing optical waves represented by orange links and dark greenlinks respectively. The communication between the OBSs atsame depth is represented by red colored optical links, i.e.,horizontal haul (H-Haul) links, while the information from theOBSs which are at greater depth is relayed to the central OBSat the surface station by the OBSs at low depth, i.e., verticalhaul (V-Haul) links drawn in blue color. It is also shown inFig. 2 that the surface buoys can be operated on solar power

Page 4: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

4

Architectures of UOWNs

ChannelSpatial

CoverageMobility

Acoustic Optical Hybrid 1-D 2-D 3-DStationary Mobile

Fig. 3: Potential architectures of UOWNs .

thus improving the energy efficiency of the network. Further-more, the submarines and autonomous underwater vehicles(AUVs) can also communicate with the OBSs by using theUOWC. Finally, the information gathered at the surface stationcan be transmitted to the onshore station or mobile station byusing terrestrial RF networks. Different possible architecturesof UOWNs can be classified based on three principles, i.e., thespatial coverage, mobility of the sensor nodes, and channel.Based on these three principles, Fig. 3 summarizes the possiblearchitectures for UOWNs, which are discussed in detail below:• Stationary one-Dimensional UOWNs: Static one-

dimensional UOWNs refers to networks in which theoptical sensor nodes form a line where each nodeattached either to the surface buoys or deployed on theseabed. Each optical sensor node in such stand-aloneUOWNs process and transmit the sensed informationdirectly to the surface station [57]. The architecture ofstatic one-dimensional UOWNs follows star topology,where the transmission between the surface station andoptical sensor nodes is single hop [58].

• Mobile one-Dimensional UOWNs: Mobile one-dimensional UOWNs refers to networks in whichthe optical sensor nodes are deployed in an autonomousfashion. Each mobile optical sensor node process andtransmit the sensed information directly to the surfacestation. The node in such stand-alone UOWNs isusually a floating buoy which senses the underwaterenvironment and transmits back the information to thesurface station or it can be a node deployed in theunderwater environment for a specific period of timeto sense the environment and floats back to the surfaceto transmit the sensed information. The architectureof static one-dimensional UOWNs also follows startopology, where the transmission between the surfacestation and optical sensor nodes is single hop [58].

• Stationary two-Dimensional UOWNs: In the stationarytwo-dimensional architecture of UOWNs, a group ofstatic optical sensor nodes is deployed in underwaterenvironment [59]. Each group of nodes (i.e., cluster/cell)

has a cluster head (i.e., OAP/OBS) which collects thesensed data and transmit it to the surface station. In twodimensional architecture, the sensor nodes communicatewith the cluster head using horizontal communicationlink while the cluster head communicates with the sur-face station using vertical communication link. In two-dimensional UOWNs, it is assumed that all the sensornodes are at the same depth. The network topology ofthis architecture depends on the application requirementand it can be a star, ring, cellular, or mesh topology.

• Mobile two-Dimensional UOWNs: In mobile two-dimensional UOWNs, a group of optical sensor nodesis able to float in the underwater environment. Thetwo-dimensional mobile topology is more dynamic andchallenging [60]. In mobile two-dimensional UOWNs,the cluster head can be an AUV, which can move aroundin the network and collects the sensed information fromdifferent underwater optical sensors.

• Stationary three-Dimensional UOWNs: Generally, thethree-dimensional UOWNs are used to sense and detectspecific underwater phenomena which cannot be observedby seabed sensors [61], [62]. In the stationary three-dimensional architecture of UOWNs, the depth of de-ployed optical sensor nodes is different, where each opti-cal sensor node is floating at different depth [62]. As thenodes are deployed at different depths the communicationin these networks goes beyond two dimensions. In thiscase, the three communication dimensions are given as:1) Communication between nodes at different depth; 2)Communication between nodes and the cluster head (i.e.,OAP/OBS); and 3) Communication between the clusterheads and the surface station.

• Mobile three-Dimensional UOWNs: Mobile three-dimensional UOWNs consists of underwater vehiclessuch as AUVs and remotely operated underwater vehicles(ROVs) which can move in different directions withdifferent depths. Recently, AUVs and ROVs have beenembedded into UOWNs to enhance the performance oftypical underwater networks. In [63], [64] the authors

Page 5: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

5

have proposed a three-dimensional underwater surveil-lance, exploration, and monitoring system called au-tonomous modular optical underwater robot (AMOUR).Extra features such as localization, time division multipleaccess, and remote control have been added to AMOURsystems in [65]. The performance of AMOUR has beenenhanced further in [66] by using cooperative AUVs.

III. PHYSICAL LAYER: ESSENTIALS OF UOWCS

The physical layer establishes the fundamentals of wire-less networks and involves many essential communicationfunctions including channel modeling and estimation, signalprocessing, modulation, coding, etc. Compared to the higherlayers, the physical layer of UOWNs is described well andstudied more in-depth for both terrestrial optical wireless com-munications (TOWC) and UOWC systems. In this section, westart with comparing the virtues and drawbacks of three mainUWC systems: acoustic, radio frequency, and optical. Then, adetailed discussion of underwater propagation characteristicsof optical waves is presented including absorption, scattering,turbulence, pointing, alignment, multipath fading, and delayspread. Thereafter, common modulator types and modulationtechniques are addressed along with the underwater opticalnoise resources.

A. Waves Under the Sea: A Tour of the Underwater Commu-nications

The title of this subsection may evoke many unforgettablechildhood memories since it is an adoption of the title for therenown adventure novel of Jules Verne, “Twenty ThousandLeagues Under the Seas: A Tour of the Underwater World”.As some parts of the underwater world are already full of mys-teries, today’s transportation and communication technologiesstill suffer from tremendous challenges of fulfilling underwaterexploration and observation tasks. In particular, severe aquaticimpairments pose a variety of obstacles to UWC systemsdepending on the nature of the following carrier waves:

1) Acoustic Waves: The involvement of human beings withacoustic waves in the oceans has been greatly motivated bythe intellectual curiosity and necessity to response a possiblethreat. Today, UAWC systems are employed in almost everymilitary and commercial applications of UWC [67]. Account-ing for hostile propagation characteristics of the seawater andenormous size of the still-vast oceans, the most prominentfeature of the acoustic systems is their ability to reach verylong distances up to tens of kilometers [68]. Nevertheless,UAWC systems cannot provide high quality of service due tothe following innate restrictions: 1) The nominal propagationspeed of the underwater acoustic signal is around 1500 m/swhich yields latency in the order of seconds [69]. Hence, delayperformance of acoustic systems are not desirable for real-time communication and control applications, 2) Operationbandwidth of underwater acoustic signals is between tens ofHertz to hundreds of kHz and achievable data rates of acousticlinks are typically in the order of kbps, which is apparentlynot adequate to sustain the transmission of large data volumes[60], 3) Acoustic nodes are power hungry, expensive, and

CommunicationRange[Km]

PHYLayerSecurity

DataRate[Mbps]

PropagationSpeed[m/s]

PowerConsump.[W]

NodeSize[m]

UnitCost[$]

Latency[s]

Acoustic

RF

Optical

Fig. 4: Comparison of acouistic, RF, and optical waves underdifferent performance metrics which are highlighted withgreen and red colors if they favor for high and low values,respectively.

bulky [70]. The cost per acoustic node makes creating a largescale underwater acoustic network economically demanding,energy inefficient which may necessitate battery replacementburden that can be quite a problematic task for nodes placed inthe deep sea, and 4) Acoustic systems can also distress marinemammals such as dolphins and whales [71].

2) Radio Frequency Waves: The exploitation of RF signalsis especially considered to provide a smooth transition be-tween terrestrial and underwater communication systems [14],[72]. Unlike acoustic waves, RF signals are more tolerantto turbulence and turbidity effects of the water, thus, canprovide a faster propagation speed [14]. However, underwaterRF communication is restricted to shallow waters and limitedto the extremely low frequency band (i.e., 30 - 300 Hz) whichyields a limited data rate even at very short communicationranges [73]. Even if low-price terrestrial RF modules can beintegrated into a Penny coin size, underwater RF nodes arecostly, require huge antennas, and high transmission power isrequired to compensate for high antenna losses [72], [74].

3) Optical Waves: In comparison to the acoustic and RFsystems, UOWC can support high data rates on the order ofGbps over distances of tens of meters with very low delayperformance thanks to the propagation velocity almost at thespeed of light (i.e., ≈ 2.25 ×108 m/s) [75]–[77]. These twomain advantages of optical waves can enable many real-time communication and control applications such as large-scale UWSNs and video-surveillance via AUVs. Furthermore,underwater optical wireless transceivers can be built in smallsizes with low-cost and energy-conservative laser and pho-todiodes. Noting that optical communication generally takesplace in a point-to-point fashion, it also provides an enhancedsecurity as eavesdropping is much more difficult than inomnidirectional communications.

Despite all these appealing virtues, there exist many chal-lenges to implement UOWC systems in practice: Firstly,

Page 6: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

6

TABLE I: Comparison of underwater wireless communication systems [54]

Parameters RF Acoustic OpticalTransmission Distance 100 m Upto 20 Km 10-30 mAttenuation Frequency and conductivity dependent Distance and frequency dependent DistanceSpeed 2.255 × 108 m/s 1500 m/s 2.255 × 108 m/sTransmit power Hundreds of Watts Few tens of Watts Few WattsCost High High LowData rate Upto 100 Mbps In Kbps Upto GbpsAntenna size 0.5 m 0.1 m 0.1 mLatency Moderate High Low

as it is the case for the free-space optical communication,misalignment of the optical transceivers can cause short-term disconnection which is generally a result of randommovements of the sea surface [78], [79], depth dependedvariations and deep currents [52], and oceanic turbulence [80].Secondly, even if the carrier wavelength of the light beam ischosen to be blue or green in order to mitigate the underwaterattenuation effects [18], [19], [81], light beam propagation stillundergoes absorption, scattering, and thus multipath fadingbecause of the interactions of water molecules and particulateswith the photons [75], [76]. Such kind of impairments causeperformance degradation and reduce the communication rangesignificantly.

Table I compares these three technologies by tabulating theimportant state of the art system parameters. For the sake ofa better visualization, we also draw a radar chart in Fig. 4to highlight the potential of UOWC systems which obviouslyexhibit a good performance in terms of data rate, propagationspeed, power consumption, latency, cost, and size. However,the main limitation is set by the short communication rangeswhich definitely entails range expansion via networking ofoptical nodes in order to operate in a large area of interest.Furthermore, misalignment of optical transceivers is one ofthe most challenging networking and control problems andnecessitate precise alignment algorithms with inherited self-organization and self-configuration features to keep the nodesconnected all the time. Therefore, it is of utmost importanceto gain important insights into the UOWNs from a networkingpoint of view including relaying, routing, deployment, local-ization, energy harvesting, mobility, network lifetime maxi-mization, self-configuration, and self-organization, etc. Beforeproceeding to lower layers of UOWNs, however, we believeit is necessary to briefly discuss the physical layer aspectsfor the sake of completeness of the survey. Accordingly, thefollowing subsections address the propagation characteristics,channel modeling, and modulation schemes of UOWC in somedepth as they are building blocks of UOWNs.

B. Underwater Propagation Characteristics of Optical Waves

Underwater communication channels exhibit quite differentpropagation characteristics varying with physio-chemical na-ture of oceans at different locations and depths. In particular,optical attributes of the aquatic medium is categorized basedon inherent and apparent properties. While the inherent optical

properties include absorption, scattering, and attenuation coef-ficients which heavily depend on the chemical composition ofseawater [82], apparent optical properties consist of radiance,irradiance, and reflectance factors which are determined bygeometric parameters of light beams (e.g., diffusion and col-limation) [83]. In the remainder of this subsection, we coverthese properties in more detail.

a) Absorption & Scattering: Absorption restricts thetransmission range of an underwater optical wireless link bycausing total propagation energy of an emitted light beam tocontinuously decrease. On the other hand, scattering spreadthe photons toward random directions such that some portionof them are not received by the receiver as it has a finiteaperture size whereas the reception of some other portionsmay be delayed due to following different propagation paths.Thus, scattering leads to multi-path fading, time-jitter, andinter-symbol interference phenomena.

Absorption and scattering coefficients can be formulatedbased on the proposed geometric model in [84]. This modelconsiders a scenario where a volume of water ∆V withthickness ∆d is illuminated by a light beam with wavelength λand incident power Pi. While a fraction of the incident poweris absorbed by the water body Pa = α(λ)Pi, another fractionPs = β(λ)Pi is scattered due to the change of direction.The residual light power Pt = γ(λ)Pi continues to propagateon the transmitter trajectory. Fractions α(λ) and β(λ) can beregarded as absorbance factor and scatterance factor respec-tively, and are related to γ(λ) by α(λ) + β(λ) + γ(λ) = 1 asper the law of power conservation. Accordingly, the absorptionand scattering coefficients are obtained by taking the limitas ∆d goes to zero [85], i.e., a(λ) = lim∆d→0

α(λ)∆d and

b(λ) = lim∆d→0β(λ)∆d , respectively.

According to Jerlov [86], the absorption coefficient canalso be modeled as a superposition of absorptions caused bypure seawater [87], colored dissolved organic materials (whichare highly and less absorptive to blue [88] and yellow [89]wavelengths, respectively), photosynthesising of chlorophyllin phytoplankton [90], and detritus (including organic andinorganic particles) [91]. Similarly, scattering coefficient canbe represented as a summation of scattering effects resultingfrom pure sea water, phytoplankton [92], and detritus [93].Scattering mostly depends on the density of the particulatematters rather than the wavelength as in the absorption.

The overall underwater attenuation effects are referred toas extinction coefficient and expressed as the sum of the

Page 7: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

7

V

Δd

)(iP)(aP

),( tP

),( sP

Fig. 5: Geometric model for inherit optical properties [84].

absorption and scattering coefficients, i.e.,

c(λ) = a(λ) + b(λ), (1)

which heavily depends on water types and depths.Based on their influence on inherent optical properties,

oceanic water types can be classified as follows [92]:

• Pure sea water: Pure sea water consists of pure wa-ter molecules (H2O) and dissolved salts (NaCl, MgCl2,Na2SO4, KCl, etc.), whose absorption effect sum mainlydetermines the total absorption in the pure sea water. Asscattering coefficient of the pure sea water is negligible[87], light beam propagates in a straight line with verylimited dispersion. 2) 3)

• Coastal ocean water: Coastal ocean water are highlyconcentrated due to the dissolved particles, thus, displaymore severe absorption and scattering effects.

• Turbid harbor water: Turbid harbor water shows themost hostile absorption and scattering levels as it hasthe highest concentration of suspended and dissolvedparticles.

Water depths are conceptualized by dividing oceans into vari-ous vertical zones based on the presence or absence of sunlight[94]: The layer near the sea surface is called photic zone whichgoes deep till which the sunlight penetrates. The uppermoststratum of the photic zone (0-200 m) is referred to as euphoticzone and has sufficient light to support photosynthesis. Thelower layer (200-1000 m) is known as dysphotic zone (a.k.atwilight zone) which cannot support the adequate light forthe photosynthesis. The water depths below the photic zoneis called as aphotic zone which is an abyssal region ofpitch darkness. Noting that the average depth of the oceanis around 4.3 km, the photic zone covers only a thin layerbut still poses the greatest biomass of all oceans. Extendingfrom the sea surface to the bottom, the chlorophyll variationcurve is observed to follow a skewed Gaussian profile [53].Accordingly, the attenuation coefficient has shown to startfrom 0.05 m−1 and reaches the peak record 0.1 m−1 around100 m depth, which starts decreasing for deeper waters [52].

b) Oceanic turbulence: Oceanic turbulence is defined asthe rapid variations in the refraction index due to fluctuationsin the aquatic medium parameters such as pressure, density,salinity, temperature, etc. [53]. This phenomenon provokes

inconstant light intensity reception that is referred to as scin-tillation and yields significant performance degradation.

c) Pointing & Alignment: Pointing and alignment arecritical engineering tasks to maintain a constant reliable linkbetween the optical transceivers. The pointing errors andmisalignment are generally considered to be a result of thebore-sight and jitter [95]. The bore-sight is defined as a fixeddisplacement between the transmitter trajectory (i.e., beamcenter) and center of the receiver aperture which may becaused by the inaccurate receiver location information. On theother hand, the jitter is random dislocations between the light-beam and aperture center as a result of oceanic turbulence[80], depth depended variations and deep currents [52], andrandom movements of the sea surface [78], [79]. Even thoughbore-sight can be mitigated by effective pointing and preciselocation information, jitter is still a problem as random natureof the oceanic environment cannot be controlled. We shouldnote that as the scattering effects become more significant(i.e., coastal and turbid waters), tight pointing and alignmentrequirements are relaxed due to the high dispersion of thelight-beam [96].

d) Multipath Fading and Delay Spread: Due to thescattering and reflection effects, some portions of the emit-ted light-beam may follow different propagation paths withvarious traveling distance and reaches to the receiver apertureat different time instants, which yields time dispersion (i.e.,delay spread) and inter-symbol interference (ISI). Unlike theUAWC where delay spread and ISI is quite considerable dueto very long distances and low propagation velocity, thesephenomena have not received much attention as a result ofthe high signal speed and limited communication ranges ofUOWC. Multipath fading can be more significant in shallowwaters because of the reflections from the sea surface, seabed,and obstacles in the vicinity. Impacts of spatial diversity on ISIwas investigated in [97] where high data rates were observedto suffer more from ISI phenomenon. In order to quantifythe time spread, authors of [20] have investigated the impactsof system design parameters such as divergence angle andreceiver aperture size. In [98], time spread analysis deducesthat ISI is significant at 50 m for a polarized light beamwith 1 Gbps data rate. However, a Monte-Carlo simulationbased channel characterization concludes that time spread isnegligible over short distances [99].

C. Underwater Optical Wireless Channel Modeling

In this section different UOWC channel models are brieflydiscussed. Firstly, the underwater optical wireless attenuation,absorption, and scattering models are presented which in-cludes Beer-Lambert law, volume scattering function, radiativetransfer equation, and Monte-Carlo methods. Secondly, theoceanic turbulence models are discussed for UOWC channelsand finally, the models for pointing and misalignment arepresented.

1) Underwater Optical Attenuation Models:a) Beer-Lambert Law: The simplest and thus most

widely used model to describe the UOWC channel attenuation

Page 8: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

8

is Beert-Lambert Law which expresses the received signalpower at the receiver as

Pr(λ, d) = Pte−c(λ)d, (2)

where Pt is the transmission power of transmitter, c(λ) isthe extinction coefficient given in (1) and d is the Euclideandistance between the transceivers. As previously discussedin detail, c(λ) changes for different water types and depths[100], [101]. For practical values of c(λ), we refer interestedreaders to the works in [20], [52], [53], [84], [85], [92], [102],[103]. Assuming a perfect pointing between the transceivers,Beer-Lambert Law presumes all scattered photons are lostby ignoring the multipath arrival of the scattered photons.To overcome this deficiency, more sophisticated models wereproposed, which are introduced in the following subsections.

b) Volume Scattering Function: Volume scattering func-tion (VSF) can be interpreted as the scattered intensity per unitincident irradiance per unit volume of water and expressed as[85]

ϑ(λ, φ) = lim∆d→0

lim∆ω→0

Ps(λ, φ)

∆d∆ω, (3)

where Ps(λ, φ) is the power of scattered light beam intoa solid angle which is centered on φ as shown in Fig. 5.Hence, scattering coefficient can be obtained by integratingthe VSF over all directions, i.e., b(λ) =

∫ϑ(λ, φ)dω. Fur-

thermore, scattering phase function (SPF) can be expressedby normalizing the VSF by the scattering coefficient [85], i.e.,ϑ(λ, φ) = ϑ(λ,φ)

b(λ) which is commonly represented by Henyey-Greenstein function [75], [77], [99].

c) Radiative Transfer Equation: Even though VSF is animportant inherent optical property to characterize the scatter-ing effects, it is not easy to measure in practice [104] and notsuitable for a large number of photons as it only considersthe scattering of a single photon [52]. To mitigate thesedrawbacks, radiative transfer equation (RTE) was proposed asan alternative and it can describe the energy conservation ofa light beam passing through a steady medium [105]. RTE isexpressed as [106], [107]

~r∇L(λ,~r, ~) = −cL(λ,~r, ~) +

∫2π

ϑ(λ,~r, ~r ′)L(λ,~r, ~)d~r ′

+ E(λ,~r, ~) (4)

where ~r is the direction vector, ∇ is the divergence operator,L(λ,~r, ~) represents the optical radiance at position ~ towardsdirection ~r, ϑ(λ,~r, ~r′) is the VSF, and E(λ,~r, ~) denotes thesource radiance. By taking light polarization and multiple scat-tering into consideration, an analytic solution was developedin [98] by using Stokes vector. Another analytical solutionwas devised in [103], [108] where derivation was simplifiedby small angle approximation. Since it is very hard to findan exact analytical solution of RTE [107] which is generallyobtained by making assumptions and simplifications [109],numerical solutions of RTE have gained more attention.

d) Monte-Carlo Methods: Monte-Carlo simulation is aprobabilistic numerical solver which mimics the underwaterlight propagation by emitting and tracking the large amountof photons [99]. It has gained popularity due to its desirable

features of accurate results, easy programming, and highflexibility [110]. However, it sill suffers from time complexity,efficiency, and statistical errors [111]. A robust Monte-Carlobased model was designed in [112] by the U.S. Naval Re-search Laboratory. Recent research efforts on characterizingthe UOWC channels by solving the RTE with Monte-Carlosimulations can be found in [99], [113]–[115].

2) Oceanic Turbulence Modeling: Although UOWC chan-nel modeling studies are mostly concentrated on obtaininga precise characterization of the absorption and scatteringeffects, the impacts of oceanic turbulence on the systemperformance has not received the attention it deserves. Asphysical mechanisms of atmospheric and oceanic turbulenceshare some similar features, several oceanic turbulence mod-eling studies employed traditional free-space optical (FSO)turbulence models. For example, the classical spectrum modelof Kolmogorov was adopted for UOWC channels in [116].Inspired by [116], a generic channel model was proposed in[117] by considering absorption, scattering, and turbulencewhich directly applies the well-known lognormal turbulencemodel, i.e.,

fI(I) =1

I√

2πσexp

(− (ln(I)− µ)

2

), (5)

where I is the received light intensity, µ is the mean logarith-mic light intensity, and σ is the scintillation index.

Impacts of oceanic turbulence and depth on the underwaterimaging were analyzed in [118], [119]. Adaptive optics wereproposed in [120] to mitigate the negative effects of turbulencefor UWOC and underwater imaging. In [121], authors havederived the power spectrum of refractive index fluctuationsin turbulent sea water. Gaussian light-beam propagation inturbulent sea water was studied in [122]–[124]. In the weakoceanic turbulence case, an aperture averaging method wasanalyzed and shown to improve system performance by reduc-ing the scintillation index [80]. In [125], the average speed ofmoving oceanic turbulence has been shown to have a majorimpact on the temporal correlation of the irradiance whereasthe link distance has minor effects. Using the Rytov method,scintillation indices of different optical waves was evaluatedin turbulent aquatic medium [126].

3) Pointing Errors and Misalignment Modeling: Neglectingthe pointing errors caused by jitter, misalignment is modeledusing the following beam spread function (BSF) [103], [127]

BSF(λ, d, r) = Pr(λ, d)E(d, r) +

∫ ∞0

Pr(λ, d)E(d, x)

×

[exp

(∫ d

0

b(λ)ϑ (x(d− y)) dy

)− 1

]J0(yr)ydy, (6)

where E(d, r) and E(d, x) are the irradiance distributions ofthe laser source in spatial coordinates and spatial frequencydomain, respectively; d is the distance between transceivers;r is the distance between the center points of aperture andthe received light-beam; ϑ(·) is the SPF. Using this model,the authors evaluated the BER performance of UOWC undermisalignment condition. In [96], pointing error performancewas investigated as a function of BSF under different water

Page 9: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

9

types. Effects of random movements of the sea surface onthe jitter of transceivers were studied in [78] where PDF ofsea surface movements are considered as a two-dimensionalGaussian distribution. Impacts of transmitter parameters suchas divergence and elevation angles were also analyzed andsimulated using Monte-Carlo method [79]. In [99], misalign-ment of point-to-point (P2P) communication was studied byusing Monte-Carlo simulations and verified with water tankexperiments. Given sufficiently large transmission power, nu-merical results showed that a small misalignment does notyield a significant performance loss for any water type [28].

D. Optical Wireless Modulation Techniques

Optical wireless modulation schemes can be categorizedinto two main class: intensity modulation (IM) (a.k.a. non-coherent modulation) and coherent modulation (CM), whichcan be implemented either by a direct or an external modula-tor.

1) Modulator Types: Direct modulators use the light sourcecurrent by switching the light-source ON and OFF to transmit“1” and “0”, respectively. Even though it has very low com-plexity and price, direct modulators are limited by commu-nication range and achievable data rates due to the chirpingeffect. In external modulators, on the other hand, the lightsource is kept always on to transmit a continuous light-beamwhose intensity or phase is modulated by an external device topass or block the light-source to transmit the desired message.External modulators can provide very high data rates andlong link ranges thanks to their switching speed and constanttransmission power. However, they are not efficient in termsof power, cost, and complexity.

2) Intensity Modulation: The IM is carried out by modu-lating the intensity of the light source by a direct or externalmodulator. If the receiver demodulates the received light usinga direct detector (DD), the overall system is referred to asIM/DD modulation. IM/DD is the most prominent modulationscheme due to its low cost and simplicity, as there is no needfor the phase information. In what follows, we present andcompare common IM schemes from the spectrum, power, cost,and monetary efficiency perspectives.

a) ON/OFF Keying (OOK): The simplest form of theIM modulation is the OOK scheme where the “1” and “0” arerepresented by the presence and absence of light. OOK em-ploys return-to-zero (RZ) or non-return-to-zero (NRZ) pulseformats. The NRZ format occupies the entire bit duration torepresent “1” while the RZ format only occupies part of thebit duration. The performance of OOK severely degrades withthe channel variations, thus, a dynamic threshold mechanismcan improve the overall performance by updating the detectionthreshold according to the channel state estimation [128]. Thelow power consumption, bandwidth efficiency, and simplicitymakes OOK a popular and practical scheme which is theoret-ically and experimentally studied for UOWC in [97], [129],[130].

b) Pulse Position Modulation (PPM): PPM is one ofthe most widely used techniques which modulates each ofM transmitted bits as a pulse within 2M time slots whose

position corresponds to the message sent. PPM provides higherpower and spectral efficiency in return for a more complicatedtransceiver. Even though it does not need a dynamic thresholdmechanism, tight synchronization requirements cause signifi-cant performance loss due to jitter effects. Conventional PPMwas improved by its variants such as differential PPM [131],digital pulse interval PPM [132], differential amplitude PPMand multilevel digital pulse interval modulation [133]. Analyticand experimental studies on the PPM can be found for UOWCin [134]–[141].

c) Pulse Width Modulation (PWM): In an M -ary PWM,pulses only appear in the first M time slots where M is equalto the decimal of the transmitted bits. PWM reduces the peaktransmission power by spreading the total power to M timeslots, which yields higher average power with the increase inM [142]. PWM is especially advantageous with its spectralefficiency and immunity to ISI effects [128].

d) Digital Pulse Interval Modulation (DPIM): In DPIM,“ON” pulses are followed by “OFF” time slots whose numberis equal to the decimal value of transmitted data symbol [143].Unlike PPM and PWM, DPIM is an asynchronous modulationtechnique which can also support variable symbol lengths.Even if it provides higher power and spectral efficiency, theDPIM suffers from error propagation during the demodula-tion process. We refer interested readers to [144]–[146] forapplications of DPIM in UOWC.

3) Coherent Modulation: To the contrary of IM schemes,the CM employs both amplitude and phase information toencode the desired message. At the receiver side, a localoscillator converts the optical carrier down to baseband orRF intermediate frequency which is referred to as homodyneand heterodyne detection, respectively. The CM can providehigher receiver sensitivity, spectral efficiency, and resistance tobackground noise, but with extra cost and complexity. We referinterested readers to [136], [147]–[149] and references thereinfor CM schemes including, phase shift keying and polarizationshift keying.

4) Receiver Noise Sources: The optical receiver is affectedby many noise sources including the photo-diode (PD) darkcurrent, transmitter noise, shot noise, thermal noise, and back-ground noise [150]. Noting that PD dark current is negligiblein practice [128], the transmitted light is affected by the trans-mitter noise caused by the fluctuations of the light intensity.Transmitter noise is generally modeled by laser relative noise[151] which is shown to have a minor effect on the receiverperformance [152]. Thermal noise is generally modeled as azero-mean Gaussian random process which results from thebehavior of electronic circuitry, especially the load resistor.Shot noise (a.k.a. quantum noise) is modeled as a Poissonprocess which originate from random fluctuations of the PDcurrent. If the received number of photons is large, the Poissonprocess can be approximated by a Gaussian process for bothPIN diode based receivers [151] and avalanche PD basedreceivers [153].

The background noise is highly dependent upon water types,depths, and optical carrier wavelength. In the euphotic zone,the solar interference can be regarded as the main contributor

Page 10: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

10

of the background noise, whose variation is given by [101]

σsol = A(πΦ)2∆TfL, (7)

where A is the aperture size, Φ is the receiver’s field of view(FoV), ∆ is the optical filter bandwidth, Tf is the transmis-sivity, L =

ERLfe−kh

π is the solar radiance [W/m2], E isthe down-welling irradiance [W/m2], R is the reflectance ofthe down-welling irradiance, Lf is the diectional dependenceof radiance, k is the diffuse attenuation coefficient, and his the water depth. For deeper regions, another contributoris the bioluminescence (a.k.a. blackbody radiation) which isgenerally focused on blue-green wavelengths and given as[101]

σbb =2hc2aA(πΦ)2∆TaTf

λ5(exp

(hcλkT

)− 1) , (8)

where c = 2.25257 × 108 is the speed of light in theaquatic medium, h is the Plank’s constant, κ is the Boltzmannconstant, Ta = e−τo is the transmission in water, T is thesymbol duration, a = 0.5 is the radiant absorption factor,and λ is the optical carrier wavelength. Accordingly, theoverall background noise can be represented as a summationof solar and blackbody interference which is given as [98],σbg = σsol + σbb.

IV. DATA LINK LAYER: LINK CONFIGURATIONS ANDMULTIPLE ACCESS SCHEMES

Data link layer is the protocol layer that convey databetween the neighbor network entities (i.e., single-hop ormulti-hop connections) and may provide functions to detectand correct possible physical layer errors. Regardless of usersultimate destination, data link layer undertakes the task ofarbitrating among the users, who compete for the same net-work resources such as time, frequency, space, wavelength,etc., in order to prevent frame collisions and specify protocolsto detect and recover from such collisions.

Accordingly, this section first compares wide-beam short-range and narrow-beam long-range transmission schemes andcall attention to the fundamental tradeoff between divergenceangle (i.e., coverage span) and communication range (or thereceived power for a given range). Then, the power budgetof three main UOWC link configurations is presented: lineof sight (LoS), non line of sight (NLoS), and retro-reflectivelinks. Even though the contents of first two subsections arenot solely related to the data link layer, covering them in thissection is especially important to provide valuable insights intothe cross-layer optimization of the first two layers. Followingthe error and data rate performance for UOWNs, poten-tial multiple access schemes are addressed including time-division multiple access (TDMA), frequency-division multi-ple access (FDMA), code-division multiple access (CDMA),non-orthogonal multiple access (NOMA), wavelength-divisionmultiple access (WDMA), and space-division multiple access(SDMA).

A. Narrow Beam vs. Wide Beam Light SourcesDepending upon the divergence angle specifications, light

sources can be classified into two broad categories: wide-beam

and narrow beam sources such as light emitting diodes (LEDs)and laser diodes, respectively. Let us first cover these twotransmitter types in the realm of TOWC systems as they teachvaluable lessons for UOWC applications. The visible lightcommunication (VLC) operates on LEDs to combine their twomain advantages: energy efficient indoor/outdoor illumination[154] and high-speed data delivery [155], which is alreadybeing commercialized by many startups, e.g., Light Fidelity(Li-Fi) [156]. Since the VLC targets to serve multiple usersconcurrently, ongoing research efforts mostly concentrated onefficient resource sharing and multiple access schemes [157].However, FSO communication focuses more on long-rangeand high data rate outdoor TOWC applications such as wire-less X-hauling [158]. Unlike the VLC, pointing, acquisitions,and tracking (PAT) functionality plays an important role forFSO communication systems to maintain a continuous systemperformance [159] since they are employed for P2P long-rangeoutdoor links.

Atmospheric link losses are generally dominated by a beamspreading factor, d−2, where d is the communication distance.In the aquatic medium, however, extinction loss, e−c(λ)d, ofnearly collimated light beams (e.g., lasers) dominates the beamspreading factor. On the other hand, beam spreading factor d−2

is the primary source of loss in the link budget calculationsof light sources with broad divergence angles (e.g., LEDs).Hence, wide-beam light sources can communicate with nearbyreceivers scanned in a broad angle circular sector whilenarrow-beam light sources can reach distant receivers withina tight circular sector, as shown in Fig. 8a. In other words,there is a fundamental tradeoff between divergence angle (i.e.,spanned coverage area) and transmission range (or receivedpower for a fixed range). It must also be noted that even if thetransmitter has a very tight divergence angle, the receivers canobserve a slightly diffused light beam because of the aquaticmedium, which is more significant in water types with severescattering nature, e.g., turbid water.

Accordingly, narrow-beam light sources have the followingadvantages [160]: (i) Higher power reception and longercommunication ranges; (ii) reduced time spread due to therelatively high ratio of “ballistic” photons which propagateswithout scattering. Monte Carlo simulations show that 90%of photons arrive within 10 ns and 2 ns for a wide-beam andnarrow-beam transmissions, respectively. The arrival time caneven be reduced to 90 ps if the narrow-beam transmission isreceived by a receiver with 0.1 mrad FoV; and (iii) improvedspectral and spatial filtering options are available since thereceiver FoV can be reduced significantly due to the limitedlight diffusion at the receiver. Albeit these advantages, narrow-beam transmission requires accurate PAT mechanisms whichare addressed in Section V-A.

For a Gaussian light beam with 1 cm radius beam waist,Fig. 6 shows the photon density [dB] vs. radial displacement[m] and propagation distance [m] over 20 extinction lengthsin clear ocean conditions. Fletcher et. al. considers a narrow-beam laser communication system over 20 extinction length(around 132 m for an extinction length of 6.6 m) with 100mW transmission and 2 cm aperture size [160]. 16-ary PPMwith 1/2-rate forward error correction (FEC) achieves 1 Gbps

Page 11: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

11

Fig. 6: Photon density vs. radial displacement and propagationdistance over 20 extinction lengths in clear ocean conditions[160].

Fig. 7: Comparison of underwater system demonstrations andpotential narrow-beam system [160].

capacity at a wavelength of 515 nm where the attenuationloss is 87 dB and noiseless sensitivity is 2.9 b/photon. Forcomparison purposes, the same set up was also consideredfor a wide-beam transmitter which achieves only 3.5 kbps.Underwater demonstrations in [20], [31], [161], [162] werecompared with the considered potential narrow-beam systemin Fig. 7 where the authors have assumed sensitive receivers,FEC, PAT, and photon-efficient modulation techniques.

B. Aquatic Optical Link Configurations

In this section, we consider three main link configurationsfor UOWNs: 1) LoS Links, 2) NLoS Links, and 3) Retro-Reflective Links.

1) LoS Links: LoS communication is the most straightfor-ward form of optical links where transceivers communicateover an unobscured link which can either happen in a diffusedor a P2P fashion as illustrated in Fig. 8a. Even implementingthe P2P LoS links for stationary transceivers is a trivial taskin clear ocean, it may require sophisticated PAT mechanismsto keep transceivers bore-sighted in the case of mobility.

Long-Range

Narrow-Beam

(P2P LoS)

nb2

Short-Range

Wide-Beam

(Diffused LoS)

wb2

ij

i

ji

wbRnbR

k

j

l

(a) LOS

min

hx

max

i

kj

(b) NLoS (Reflective)

i2 j

ji

i

RR2

(c) Retro Reflective

Fig. 8: Underwater optical link configurations: a) LoS, b)NLoS (Reflective), and c) Retro Reflective.

For a generic optical transmitter node i and receiver nodej, propagation loss factor is given based on Beer Lambert’sLaw as [163]

Lij(λ, dij) = exp {−c(λ)dij} , (9)

where dij is the Euclidian distance between the transceiversand ϕji is the angle between the receiver plane and the trans-mitter trajectory. Likewise, geometric gain (a.k.a. telescopegain) of the LoS link is given as [21]

GLoSij =

Aj

d2ij

cos(ϕji )

2π[1−cos(θi)],−π/2 ≤ ϕji ≤ π/2

0, otherwise, (10)

where Aj is the receiver aperture area of node j and θi isthe beam divergence angle of transmitter node i. In order to

Page 12: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

12

concentrate the transmitted energy on receiver aperture, thedivergence angle of laser-diodes are generally designed to bea few milliradians or less [164] whereas typical LEDs canhave divergence angles less than 140 milliradians to diffuselight to wide angles [165]. Accordingly, received power canbe formulated as a product of transmission power, transceivers’efficiency, telescope gain, and path loss factor, i.e.,

P jr = P it ηitηrjG

LoSij χ(ψji )Lij

(c(λ),

dij

cos(ϕji )

), (11)

where P it is the transmission power, ηit and ηrj are transmitterand receiver efficiency, respectively; χ(ψji ) is the concentratorgain [166], which is defined for non-imaging concentrators as[167]

χ(ψji ) =

{n2

sin2(Ψj), 0 ≤ ψji ≤ Ψj

0, ψji > Ψj

, (12)

ψji is the angle of incidence w.r.t. the receiver axis, Ψj isthe concentrator FoV which can be π/2 and down to π/6for the hemisphere and parabolic concentrators, respectively;and n is the internal refractive index. Notice that the receivergain increases as the FoV decreases. Hemispherical lens arecommon nonimaging concentrators [168] which can achieveΨj ≈ π/2 and χ(ϕji ) ≈ n2 over its entire FoV. The compoundparabolic concentrator [167] is another type of nonimagingconcentrators and can obtain a much higher gain in returnfor a narrower FoV, which is especially more desirable forP2P-LoS links. Since it is easy to implement, most of theexperimental studies considered LoS links under differentwater characteristics and modulation schemes using a varietyof transmitter hardware [20], [31], [147], [148], [161], [162],[169].

2) NLoS Links: LoS links may not always be available dueto the obstructions within the underwater topology, PAT errors,mobility, and random orientations of the transceivers, etc. Insuch cases, a diffused light beam which is reflected over seasurface (or alternatively a mirror located in an appropriatelocation) can be beneficial to facilitate a point-to-multipoint(P2M) (a.k.a. multicasting) transmission to reach obscured re-ceivers, as depicted in Fig. 8b. Assuming that the transceiversare oriented vertically upward, the transmitted light beamis characterized by inner and outer angles θmin and θmax,respectively. As per the Fresnel’s law, propagating light ispartially refracted and partially reflected at interfaces betweenthe mediums with different refractive indices. Therefore, thelight beam transmitted from depth h is partially reflected fromthe sea surface and illuminate an annular surface Aann at depthx with equal power density. Aann is given by

Aann = 2π(h+ x)2 [cos(θmin)− cos(θmax)] , (14)

which defines an annular area taken from a sphere of radiush+ x [163]. Assuming that sea surface is modeled as smooth(i.e., incident angle is equal to the perpendicular angle betweenthe receiver plane and the transmitter-receiver trajectory, i.e.,ϕji ), the telescope gain of the NLoS links is given in (13)where θt is the angle of transmission, θc , sin−1

(nA

nW

)is the

critical angle (i.e., the angle of incidence above that the total

internal reflection (TIR) occurs), nA is the refraction indexof air, and nW is the refraction index of water. Accordingly,received power at node j is expressed as follows

P jr = P it ηitηrjG

NLoSij χ(ψji )Lij

(c(λ),

h+ x

cos(ϕji )

). (15)

LoS and NLoS links have been compared by Jasman et.al. in [170] where they have demonstrated that 100 MHzbandwidth availability of LoS links is reduced to 20 MHz incase of NLoS even in clear water conditions. Indeed, such areduction is not a surprise due to the reflection losses at the seasurface and diffusion of the reflected light beam. Furthermore,multi-scattering effect of NLoS links was addressed in [171]and [172].

3) Retro-Reflective Links: Similar to backscatter communi-cation in RF systems, retro-reflective communication consistsof a light source and a reflector. While the light source couldbe a sophisticated system with high transmission power, thereflector behaves as an interrogator as it lacks the abilityto fulfill transceiver operations due to its simple architecturewith low power availability. Therefore, the continuous lightbeam emitted from the source is modulated and reflectedback to the receiver. Retro-reflective communications can beconsidered in two cases [51]: photon limited case and contrastlimited case which take place in clear and turbid water,respectively. In the former case, absorption is the dominanteffect which reduces the number of photons received bythe reflector. Furthermore, the accuracy of PAT mechanismsat both sides plays a significant role in receiving enoughinformation-bearing photons. In the latter case, scattering is thedominant factor which mainly determines retro-reflective linkrange and capacity. Contrast limitation is especially importantfor underwater imaging applications as a reduction in photonquantity directly reduces the image contrast, which can be con-siderably improved by exploiting polarization discrimination[173], [174]. If the receiver has enough power resource, thereflector can even amplify the modulated light beam in orderto achieve a better performance both in photon and contrastlimited scenarios [175].

Based on the geometric gain of LoS links in (10), telescopegain of the retro-reflective links is expressed as [21]

GRRji =

Aj

d2ij

cos(ϕji )

2π[1−cos(θi)]

ARR cos(ϕij)

π[dij tan(θRR)]2,−π/2 ≤ ϕij ≤ π/2

0, otherwise(16)

where ARR is the aperture area of the reflector, θRR is thedivergence angle of the reflector, and ϕji is the angle betweenreceiver trajectory of the source and the reflector trajectory.Accordingly, reflected light beam is received back by thesource node i as follows

P ir = P it ηitηri ηRRj GRRij χ(ψij)Lij

(c(λ),

2dij

cos(ϕji )

), (17)

where ηRRj is the retroreflector efficiency.

Page 13: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

13

GNLoSij =

Ajcos(ϕ

ji )

2Aann

([tan(θt−ϕj

i )

tan(θt+ϕji )

]2 [sin(θt−ϕj

i )

sin(θt+ϕji )

]2), θmin ≤ ϕji ≤ θc

Ajcos(ϕji )

2Aann, θc ≤ ϕji ≤ θmax

0 , otherwise

(13)

Fig. 9: Demonstration of photon arrival rate for an LoS link[175].

C. Error and Data Rate Performance

Before proceeding to the medium access schemes, it isimportant to quantify the error and data rate performance ofthese link configurations based on a common and straightfor-ward detection technique. Therefore, the authors in [176] con-sidered IM/DD OOK with silicon photo-multipliers (SiPMs)based photon counter detectors. Photon arrivals are generallyassumed to be a Poisson distributed function, therefore, thephoton arrival rate within a slot duration T is given by

pji =P rj η

jD

TRji}c, (18)

where P rj is the received power, ηjD is the detector efficiency,Rji is the data rate, } is the Plank constant, and c is the speed oflight. Fig. 9 shows the photon arrival rate w.r.t. xy-coordinatesfor a sensor fixed at origin and pointing in positive x-axisdirection while the receiver is located at different location andits receiver directed to the origin. Assuming a large number ofphoton reception, then according to the central limit theorem,Poisson distributed photon arrivals can be approximated by aGaussian distribution and the bit error rate (BER) is given by

BERji =1

2erfc

[√T

2

(√p1ij −

√p0ij

)], (19)

where erfc(·) is the complementary error function, p0i,j =

pbg+pdc and p1i,j = pji +p0

i,j are the photon arrival rates whenbinary 1 and binary 0 are transmitted, respectively; pbg and pdcare the background illumination noise and additive noise due todark counts, respectively. For a given BER BER

j

i , achievabledata rate can be obtained from (19) as

Rji =P ji η

jDλ

T}c[(

erfc−1(

2BERj

i

)√2T +

√p0ij

)2

− p0ij

] .(20)

Fig. 10: Illustration of optical polyhedron transceivers (SOTs)and its implementation [169], [178].

Since hard decision forward error correction (HD-FEC) cansuccessfully identify and correct all bit errors below anFEC-BER threshold, one can set BER

j

i ≤ 3.8 × 10−4 asrecommended by the International Telecommunication UnionStandardization Sector (ITU-T) [177].

D. Multiple Access Schemes

For infrastructure based UOWNs, many researchers con-ceptualized omnidirectional OAPs/OBSs by designing themin multi-faceted spherical shape which has single or multi-ple transceivers at each face [169], [178]–[180]. Therefore,underwater OBSs can be designed as geodesic polyhedra asshown in Fig. 10 along with its implementation [169], [178].Geodesic polyhedra approximates spheres with triangles andcan be a good solution against underwater pressure as thegeodesic domes are known to withstand heavy structure loadsby distributing the structural stress over its rigid triangularbuilding blocks [181]. Notice that as the number of faces (pen-tagonal/hexagonal shapes in Fig. 10) increases, it is possibleto employ narrower transmitter divergence and receiver FoVangles which naturally yields longer transmission range andhigher receiver gain (Fig. 10), respectively. In addition to theirspatial reuse and angular diversity advantages [179], OBSs canalso provide flexibility as each LED on a face can be exploitedto serve for fulfilling a specific task.

In OBS based cellular UOWNs, there exists two maininterference scenarios: intercell interference (ICI) and intracellinterference which is also referred to as multiple access inter-ference (MAI). While the former happens when a user receivessignals from other users using the same network resourceswithin the adjacent cells, the latter occurs when a user observesinterfering signals from users sharing the same cell resources.Compared to VLC systems, intercell interference expected to

Page 14: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

14

be at low levels due to the severe aquatic channel impairmentsand can even be further reduced by intelligent OBS deploy-ment strategies. Nonetheless, intracell interference still stays asa first and foremost research challenge for both downlink (DL)and uplink (UL) transmission. Hence, in addition to efficientresource allocation strategies, OBSs necessitate multicarriertransmission schemes and multiple access protocols to serveseveral users simultaneously.

As diagramed in Fig. 11, multiple access schemes canbe categorized into electrical and optical multiplexing sub-categories. Electrical multiplexing schemes consist of timedivision multiple access (TDMA), frequency-division multipleaccess (FDMA), code-division multiple access (CDMA), andnon-orthogonal multiple access (NOMA) whereas optical mul-tiplexing schemes contain wavelength-division multiple access(WDMA) and space-division multiple access (SDMA). To thebest of authors’ knowledge, there is no research efforts onUOWN multiple access schemes excluding the optical CDMA[129], [182], [183]. In what follows, we present multicarriertransmission techniques along with corresponding multipleaccess schemes for UOWNs.

1) Time Division Multiple Access: TDMA is a synchronouschannel access scheme where non-overlapping time slots areassigned to different users as per the requested QoS levels.Hence, TDMA does not allow nodes to transmit simultane-ously and independently. In UAWC systems, TDMA providesa limited bandwidth efficiency because low propagation speedrequires long time guards to prevent packet collisions of theadjacent time slots [68], which may not be the case for UOWCsystems thanks to low propagation delays. TDMA can supporthigh energy efficiency in return for reduced capacity per user[166]. Nevertheless, TDMA requires efficient scheduling tech-niques in order to overcome the MAI. A potential schedulingscheme could be based on users rather than LEDs embeddedon OBSs as they can be much larger than the number ofusers. Even though TDMA has not attracted the attention forUOWNs yet, it can be motivated by research efforts on TDMAbased VLC systems: As a potential solution, each LED isorthogonally allocated to a time slot in [184] and a blockencoding TDM is exploited in [185] where one LED fromeach LED group is allowed to transmit. In [186], TDMA isconsidered for UL transmission where each user has certaintime slots to transmit such that identity of the transmittingusers can be recognized as per the scheduling policy.

2) Frequency Division Multiple Access: FDMA schemepermits multiple users to transmit momentarily over non-overlapping frequencies/subcarriers within a cell area. Notingthat FDMA is not suitable for acoustic systems due to thelimited bandwidth availability [60], it offers high spectralefficiency and robustness again intersymbol interference (ISI)[187] for optical wireless communications. However, it lacksenergy efficiency which deteriorates with the increasing num-ber of subcarriers [188]. Orthogonal FDMA (OFDMA) andinterleaved FDMA (IFDMA) are two well-known schemesstudied extensively for OWCs [157].

OFDMA allocates each user with several time slots andfrequency blocks which spans a number of orthogonal fre-quency division multiplexing (OFDM) subcarriers. Because

Multiple Access Techniques

Optical Multiplexing

Electrical Multiplexing

NOMA OCDMA TDMA WDMA SDMAFDMA

IFDMA

O-OFDMA

Fig. 11: Classification of multiple access schemes.

of the real and unipolar valued signal requirements of theIM, conventional OFDM cannot be directly applied to opticalOFDM (O-OFDM) systems. In return for losing half of thebandwidth, reality constraint can be satisfied by applyingHermitian symmetry on inverse fast Fourier transform inputs.Positivity of the signals can be achieved either by direct currentbiased optical OFDM (DCO-OFDM) [189] or asymmetricallyclipped optical OFDM (ACO-OFDM) [114]. The former addsa DC bias before transmission which may cause overheatingand high signal distortion. At the expense of BER performancedegradation and increased complexity [190], several peak-to-average power ratio (PAPR) reduction techniques wereproposed to overcome these problems [191], [192]. In or-der to obtain unipolarity, ACO-OFDM technique clips thesignal at zero level [193] and transmits only the positivepart of the signal. Even if it is more energy efficient thanthe DCO-OFDM, bandwidth utilization is quite low becauseof using only half of the subcarriers for data transmission.Optical OFDMA (O-OFDMA) was proposed in [194] whichhas a lower decoding complexity and power efficiency incomparison with O-OFDM based interleave division multipleaccess (IDMA). In [195], the authors have considered twohandover schemes for users within the intersection area of twooptical transmitters: In the first scheme, the user combines thesignal of both transmitters, while in the second scheme theeach transmitter use a dedicated band for the user. IFDMAwas proposed in [196] to mitigate the high PAPR effectsof O-OFDMA where it was shown that IFDMA have lowercomputational complexity than O-OFDMA and it reduces thesynchronization errors.

3) Code Division Multiple Access: Optical code divisionmultiplexing (OCDM) is a multiplexing scheme where com-munication channels are distinguished by optical orthogonalcodes in addition to time and wavelength [197], [198]. Asshown in Fig. 12, the data stream is multiplied by a codesequence either in the time domain, wavelength domain, oreven as a combination of both (i.e., 2D coding) [199]. In thetime domain, a bit duration is divided into smaller time slotswhich are called chips. Bipolar time-encoding is a coherenttechnique that manipulates the phase of the optical signal andneeds phase accuracy. As an alternative, positive time encodingis non-coherent which manipulates the power of the opticalsignal without requiring any phase information [200]. On the

Page 15: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

15

Fig. 12: Illustration of OCDMA dimensions [199].

other hand, a wavelength-encoded signal consists of a uniquesubset of wavelengths in order to form the code. Finally,2D coding combines both time spreading and wavelengthassignment such that a data stream is constituted as successivechips of different wavelengths. In the receiver side, decoding isperformed by applying the reverse operations of the encoding.

Accordingly, optical CDMA (OCDMA) employs OCDMtechnique to mediate multiple asynchronous nodes in shar-ing common network resources. Thanks to its high spectralefficiency, distributive, and asynchronous nature; OCDMAhas received much attention to be employed in UOWNs[129], [182], [183]. In [129], the authors have addressed thestructures, principles, and performance analysis of OCDMAbased cellular UOWNs where OBSs are connected to a centraloptical network controller. In [182], the performance of relay-assisted OCDMA networks was characterized by the turbulentchannels. Finally, potential and challenges (e.g., mobility, celledge coverage, blockage avoidance, power control, etc.) ofOCDMA networks were presented in [183].

4) Non-Orthogonal Multiple Access: NOMA is also re-ferred to as power domain multiple access where user signalsare superposed in a way that each signal is allocated to adistinct power level depending upon the channel conditions.While NOMA allocates more power to users with bad channelsconditions compared to those with good channel condition.Employing successive interference cancellation, the user allo-

cated with high powers can cancel the interference of the userswith the low power allocation. Thus, all users can occupy theavailable entire time-frequency resources and increase overallsystem performance significantly [157]. Even though NOMAhas attracted attention for VLC systems [201], [202], there isno study targeting NOMA for UOWCs.

5) Wavelength Division Multiple Access: WDMA facili-tate the multi-user access harnessing the wavelength divisionmultiplexing (WDM) such that each user has a dedicatedwavelength along with an optical tunable reception filter inorder to operate on assigned wavelength. WDM multiplexes anumber of optical signals at different wavelengths (i.e., color)into a single one. Coarse and dense WDM are two standardtypes which are named based on the available number ofchannels and their spacing. Even if WDMA reduces the signalprocessing complexity to a grate extent, it may significantlyincrease the hardware complexity and cost [203]. Since un-derwater operational wavelength is different from TOWCs,it is necessary to standardize the WDM channels and theirspacing for blue-green wavelengths. It is also important todevelop efficient wavelength assignment policy as the nodes inUOWNs can observe different channel conditions at differentwavelengths because of varying water types and depths.

6) Space Division Multiple Access: SDMA harnesses thespatial distribution of the users and directivity of the lightbeam propagation to permit parallel transmission on thesame network resources which can either be in time, fre-quency/wavelength, or code domains. In [204], random group-ing and optimal grouping approaches were proposed for anSDMA based VLC system and obtained results have shownthat SDMA can offer 10 times higher throughput than theconventional TDMA scheme. Notice that SDMA is a potentialtechnique to be employed for underwater OBSs as they canbenefit from both spatial and angular diversity.

V. NETWORK LAYER: RELAYING TECHNIQUES ANDROUTING PROTOCOLS

Due to the communication range limitations of UOWCs,relay-assisted UOWC is a key enabler technique to realizeUOWNs by expanding coverage area, extending the commu-nication range, enhancing energy efficiency, providing coop-erative diversity, and improving the end-to-end system per-formance [128]. However, the full benefit of relay-assistedUOWCs can be obtained by effective routing algorithms takingthe underwater propagation characteristics of light beams intoaccount. Therefore, this section first covers serial relayingand parallel relaying techniques using decode-and-forward(DF), amplify-and-forward (AF) methods, and Bit-detect-and-forward (BDF). Thereafter, potential routing protocolsfor UOWNs are surveyed including location-based routing,source-based routing, hop-by-hop routing, cross-layer routing,clustered routing, and reinforcement learning based routing.

A. Relaying TechniquesAs depicted in Fig 13, relaying can be implemented by

involving either a single node or multiple nodes at each hop,which are referred to as serial and parallel transmission,respectively.

Page 16: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

16

i

s

j

l

kn

d

Parallel

Relaying

i

s

l

kj

dSerial

Relaying

m

n

m

i

s

l

kn

d

j

d’

m

s’

Multiple

Access

Interference

Fig. 13: Illustration of serial and parallel relaying techniques between a source and destination pair along with an MAIinterference scenario.

1) Serial Relaying and PAT Mechanisms: Serial trans-mission (a.k.a. multihop transmission) employs the relayingnodes in a serial configuration along a certain routing path[97], [175], [182], which is especially beneficial to extendthe communication range and expand the cell coverage inad hoc and cellular UOWNs. In [182], authors exploited theserial relaying to expand the coverage area of OCDMA basedUOWNs. They evaluated the end-to-end performance of theproposed relay-assisted OCDMA network under absorption,scattering, and turbulence effects. In [97], end-to-end BERperformance of a multi-hop transmission was analyticallyevaluated by using single-hop BER expression as a buildingblock. Authors in [97], [182] have applied Gauss Hermitequadrature formula and derived the closed-form BER solutionunder the lognormal fading channel. In [97], end-to-end BERperformance is obtained by assuming that each hop experiencethe same error of probability, which may not be the case inreality. Therefore, an end-to-end BER performance analysiswas considered in [175] where we have distinguished the errorprobability of each transmission hop.

The key point in multi-hop transmission is to employnarrow-beam light sources in order to concentrate the re-ceived signal power at the receiver aperture area. Althoughnarrow-beam transmission significantly enhances the systemperformance at each hop, it requires highly directional beamsand rapid PAT mechanisms which accounts for beam wanderand jitters due to aquatic turbulence and random motionpatterns (roll, pitch, and yaw) of the transceiver platforms[164]. Furthermore, the precision of the localization algorithmsis quite decisive for positioning the complementary node inits FoV during the acquisition [205]. Lastly, a fast closed-loop tracking and wavefront control is necessary to sustain aconstant link [206]. To the best of our knowledge, there is nostudy addressing the PAT mechanisms for UOWNs yet. Whenthe location accuracy is low, pointing errors and misalignmentcould be mitigated by ensuring a certain diffusion area (pro-portional to the localization error) rather than directly pointingto the estimated receiver location. Therefore, it is essential

to develop robust and adaptive divergence and power controlschemes [207].

2) Parallel Relaying and Relay Selection Protocols: Paral-lel transmission (a.k.a. cooperative transmission) is an alter-native relay-assisted transmission scheme and basically builtupon the idea that the source node may be overheard by anumber of neighboring nodes which can act cooperatively torelay traffic request of the source node. In other words, a setof transmitting nodes (probably each with a single opticaltransmitter) jointly process and transmit the traffic requestby creating a virtual antenna array [208]. This cooperationnaturally increases the degree of diversity and provides op-portunities to mitigate multipath fading effects. Even thoughparallel relaying has received quite an attention in TOWCs(please see [128] and references therein), there is no UOWCwork addressing the virtue and benefits of the cooperativerelaying.

Relay selection is an interesting research topic for cooper-ative relaying schemes because involving all the neighbor intransmission may always not yield the desired results [209].This is mainly because of the fundamental tradeoff betweenthe divergence angle and received transmission power (or thecommunication range for a fixed power reception). In Fig. 13,for instance, relay node ` does not participate in relaying asit does not provide a better performance than involving relaynodes m and n only. MAI raises another issue when a relaynode is incorporated with relaying to convey two different datastreams as shown in Fig. 13 where node k is not able to servedata streams s → d and s′ → d′ at the same time unlessit employs an efficient multiple access scheme. Notice thatnode k constitutes the bottleneck of these two data streamsand such critical nodes mainly determine the overall networkperformance. It is important to develop adaptive divergenceand power control schemes [207] for employing efficient relayselection strategies in order to sustain and improve the networkperformance.

3) Traffic Forwarding Methods: Inspired by the methods inthe well-known TOWC parts, several signaling strategies canbe employed for UOWCs:

Page 17: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

17

B

C

D

A

AB B

C

AD

BD

CD

AC

Fig. 14: Illustration of focus beam routing protocol [212].

a) DF Relaying: In DF, the received optical signal ateach hop is converted into electrical signal, then decoded,and finally re-encoded before retransmission for the next hop.Although DF greatly improves performance as it limits back-ground noise propagation, it may introduce significant powerconsumption and encoding/decoding delay to the system [175].

b) AF Relaying: AF is conventionally realized by ex-ecuting optical-electrical-optical (OEO) conversion at eachnode, amplifying the received signal electrically, and thenretransmitting the amplified signal for the next hop. However,actual merits of AF relaying over the DF counterpart emergesonly if OEO conversion is eliminated. Alternatively, all-opticalAF relaying process received signal in the optical domain andrequires only low-speed and low-power electronic circuitry toadjust the amplifier gain [210]. The main drawback of the AFtransmission is the propagation of noise added at each node,which is amplified and accumulated through the path [175].

c) BDF Relaying: Different from the DF method, therelay node detects each transmitted bit of the source andforwards it to the next relay without applying any errorcorrection [211].

B. Underwater Routing Techniques

Routing holds a significant place in order to keep theUOWNs connected by discovering and maintaining the trans-mission routes. The physical layer issues for UOWNs are wellstudied in the recent past but the research on network layerissues such as routing is still in its infancy. A number ofrouting protocols for underwater acoustic wireless networkshave been highlighted in [213]–[217] some of which can bewell adapted for UOWNs. The key point in adapting theexisting routing protocols is that designers should take theangular sector shaped coverage region of optical nodes alongwith the fundamental tradeoff between the angle and radius ofthis sector. In what follows, we highlight some of the routingalgorithms proposed for underwater acoustic networks, whichcan also be adopted to apply for UOWNs:

1) Location based routing: The location information ofunderwater sensors is used in location-based routing strategy todiscover the best route from the source to the destination node.In location-based routing, every node should be aware of itslocation, the target area, and neighbors’ locations. The data isforwarded in accordance with the location information. AUVbased routing protocols were proposed in [218], [219] whichintegrate localization and routing. An energy efficient andreliable routing protocol was introduced in [220], where thetransmission from the source node starts with local floodingand then an adaptive mechanism is established to find theoptimal route with minimum energy consumption. Directionalflooding protocol were proposed in [221], [222] where thesource node knows its own location, the sink location, andthe location of its neighbors. The flooding region in [221],[222] was defined by the link qualities among the neighbors.The flooding phenomena can burden the network therefore, in[212] the authors have proposed a routing protocol based onfocused beam. It is assumed in [212] that every node knowsits location and location of the destination node, where thedecision about the next hop is made at each intermediate node.Focus beam routing is a good candidate for UOWNs due to itsdirective nature from source to destination. Fig. 14 shows thedata forwarding scheme used in focus beam routing, wherenode “A” is the sender node and node “D” is the destinationnode, the intermediate nodes are selected based on the coneangle θ (which can be considered as twice of the divergenceangle). Nodes which lies within the cone angle ±θ/2 of thesender node, are selected as relay nodes for forwarding thedata.

A geographical reflection enabled routing protocol wasintroduced in [223] which tries to find the stable route betweenthe source node and the destination node. Directional antennaswere used in [223] to consider both LoS and NLoS linksbetween the neighbor nodes. Fig. 15 shows the differentscenarios for the proposed routing protocol in [223], where itcan be seen from Fig. 15b and Fig. 15c that the directive andNLoS communication can help in simultaneous transmissionsrespectively, thus improving the throughput of the network.The proposed routing protocol in [223] was designed foracoustic underwater networks which can also be well adoptedfor UOWNs. Sector based routing protocols were designedin [224], [225] with the location prediction of destination.The network topology in [224], [225] is fully mobile whereeach node moves along a pre-defined route. Comparative studyof location-based routing protocols for underwater acousticnetworks was carried out in [226]. In all of the location-basedrouting protocols, it is assumed that the underwater sensornodes find its location by using GPS or by using underwaterlocal positioning systems. However, GPS cannot work in theunderwater environment and the underwater local positioningtechniques have large localization error due to the hostileunderwater environment.

2) Source based routing: A simple and energy efficientsource based routing protocol was introduced in [227]. Theprotocol in [227] selects the route with minimum transmissiondelay from source to the sink node. Once the route is defined,the nodes along the route can also transmit the data to the

Page 18: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

18

A B CE

D

F

(a) Multihop omnidirectional optical links.

F

B

E

A C

D

(b) Multihop directional optical links.

F

B

E

A C

D

(c) Multihop directional optical links with reflection.

Fig. 15: Multihop underwater routing protocols: a) Omnidi-rectional, b) LoS directive, and c) NLoS directive (Reflective)[223].

sink node. The average end to end delay, average energy con-sumption, and packet delivery ratio of the proposed protocol in[227] outperforms other traditional routing protocols. Anothersource based routing protocol for small size UAWC networkswas proposed in [228] where each node just share informationwith its single-hop neighbor nodes and find a minimum costpath from source to the destination. Source-based routingprotocols are good for reducing the energy consumption ofrouting protocols.

3) Hop-by-hop routing: In hop-by-hop routing, the inter-mediate nodes (or relay nodes) selects the next hop by itself.Hop-by-hop routing provides flexibility and scalability to thenetwork but the route selection may always not be optimal.Channel aware hop-by-hop routing protocols were introducedin [229], [230] where the speed of acoustic waves in differentdepth were taken into consideration for the relay nodes toreduce the end to end transmission delay. In [225], the authorshave proposed a hop-by-hop routing protocol based on beam-

width and direction of the intermediate nodes. Adaptive depthbased routing protocols were introduced in [231], [232] whichtakes into account the speed of acoustic waves at differentdepth levels, depth of sink node, and distance to sink node.A MIMO-OFDM based routing protocol was introduced in[233] to take the advantage of multiplexing and diversity gainadaptively. The proposed cross-layer design in [233] adaptsitself to the noise and interference for underwater acousticchannels and selects a suitable transmission mode for thesubcarriers. An energy efficient and network topology awaregreedy routing protocol was proposed in [234] which assignsadaptive weights to the highly connected nodes. For under-water delay tolerant networks a redundancy-based routingprotocol was designed in [235] which adopts a tree-basedforwarding method to replicate packets.

4) Cross Layer routing: The cross-layer routing protocolstake the information available from different layers into ac-count and provide a solution to several networking issuessuch as scheduling, defining routing policy, and power control.Cross-layer routing protocols can also select the next hop fortransmission by considering the transmission delay, distance tosink, channel conditions, and buffer size of the candidate node.Cross-layer strategy increases the overall network performanceand minimizes the energy consumption of the network. Cross-layer protocols for the 3D underwater environment were in-vestigated in [236], [237] which utilizes the channel efficientlyand sets the optimal packet size for transmission. Multipathpower control routing protocols were proposed in [238], [239]which combine multipath routing and power control at thesink node. Channel aware cross-layer routing protocol are alsoinvestigated in [240] which exploits the link quality for therelay selection.

5) Clustered routing: Cluster based routing is especiallysuited for infrastructure based UOWNs as shown in Fig. 2. Incluster based routing, the network is divided into a number ofclusters/cells based on the geographical location of the nodes.Once the network is divided into clusters, the cluster head(i.e., OAP/OBS) is selected for each cluster by using anycluster head selection strategy. The cluster head is used asa gateway to communicate between the clusters and to thesink node. A location-based routing protocol was introducedin [241] which divides the network into clusters and the datafrom the nodes are gathered by the cluster heads. A distributedclustering based protocol was proposed in [242] where thecommunication between the cluster head and the sensor nodewas single hop. Location unaware cluster based multihoprouting protocol was proposed in [243] where the sensor nodesdo not know their location and location of the cluster head.The interested readers are referred to [244] where a number ofcluster based routing protocols are highlighted for underwaterwireless sensor networks.

6) Reinforcement learning based routing: The routingprotocols based on reinforcement learning uses Q-learningmethod for the network states and adapts itself to the topologychange. The node analyzes its remaining energy and energyof its neighbor nodes, applies a reinforcement function, andthen selects the optimal node to forward the data [245]. Therouting problem in [245] is fully distributed and formulated as

Page 19: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

19

a Markov decision process where the state space consists ofall the nodes. A machine learning based routing protocol wasproposed in [246] which is energy efficient and improves thelifetime of the network. A layer structured based routing pro-tocol was introduced in [247] for hybrid acoustic and opticalarchitecture where the upper layer cluster heads supervise therouting in lower layer by using the Q-learning function.

VI. TRANSPORT LAYER: CONNECTIVITY, RELIABILITY,AND FLOW/CONGESTION CONTROL

Unlike the first two lower layers, transport layer of UOWNsis still in a primitive stage and remains totally unexplored.Therefore, this section discusses the fundamental challengesfor developing an efficient transport layer including connec-tivity, reliability, flow control, and congestion control aspectsof UOWNs.

A. Connectivity analysis of UOWNs

Connectivity of UOWNs is the most critical component oftransport layer as other network functions heavily depend ona connected network assumption. It is also used as a metricfor different performance parameters such as survivability,robustness, and fault tolerance [248]. Connectivity is measuredby number of links in the network and a network is referredto be as connected if there exists at least one connecting pathbetween any two nodes in the network. In strongly connectednetworks bidirectional links exist between any pair of nodeswhile in a directed network the links are usually unidirectionaluntil and unless both nodes are in the beam scanning angleof each other [249]. The problem of network connectivity isaddressed in [250] for omnidirectional networks such that nonode is obscured for RF wireless sensor networks. The exactclosed-form analytical expression of connectivity in multihopwireless networks for physical layer parameters still remainsas an open research problem. The connectivity parameters ofUOWNs depends on the transmission range of optical sensornodes, number of optical sensor nodes, number of descendantsand antecedents, node orientation, and the beam width (seeFig. 16b).

Range limitation of UWOC can be augmented with multi-hop UOWNs where nodes can share information for longdistances through intermediate nodes. Indeed, multi-hop co-operative communications have been extensively studied forRF networks [208], underwater acoustic networks [251], andTOWNs [252]. Due to the omnidirectional communicationcapability of RF and acoustic signals, wireless sensor networksare traditionally modeled as geometric random graphs [253]where two sensor nodes ni and nj are generally assumed toestablish a bidirectional communication link (i.e., ni � nj).On the contrary, such a model is not suitable for UOWNs be-cause a node can only reach to the nodes within a certain beamscanning angle around their transmission trajectory, that is,optical wireless nodes are connected via unidirectional links.Directed communication networks are generally modeled byrandom scaled sector graphs [254] where a unidirectionalcommunication link from node ni to nj (i.e., ni → nj) isestablished if and only if nj is positioned within the beam

scanning angle of ni. Notice that a directed reverse path ispossible (i.e., nj → ni) if ni is in the beam-width of njor through other multi-hop path as illustrated in Fig. 16a. Aconnectivity framework for multihop UOWNs was discussedin [255] where the authors have assumed bidirectional linksbetween every pair of optical sensor nodes. In [256], we haveanalyzed the connectivity of UOWNs by using random sectorgraphs where we have considered unidirectional links betweenunderwater optical sensor nodes.

In order to define a random sector graph consider that thetotal number of optical nodes are m, the scanning sector (cov-erage area) of ni, 1 ≤ i ≤ m, is defined as a tuple of randomorientation ζi, scanning angle φi, communication range Ri,and sensor node coordinates ci, i.e., Si = (ζi, φi, Ri, ci)which is illustrated in Fig. 16b. Accordingly, UOWNs canbe defined as a random sector directed graph G(V, E) whereV = {c1, . . . , ci, . . . , cM} represent the set of vertices andE ∈ {0, 1}M is the set of links which is primarily characterizedS = S1, . . . ,Si, . . . ,SM . Notice that Ei,j = 1 only if ni → njholds. Random sector directed graphs and random geometricgraphs are identical in case of φ = 2π [248], [249], [254],[275]. Notice that two nodes i and j are connected when thedistance between them is less than R in random geometricgraphs, however, the connectivity of random directed sectorgraphs also depends on the beam scanning angle and itsorientation. Fig. 17a and Fig. 17b shows two different randomdirected sector graphs with scanning angles of φ = π

3 andφ = 2π, respectively. It is obvious that increasing the scanningangle for each node from φ = π

3 to φ = 2π, increasesthe number of links in the graph. These asymmetric anddirectional characteristics of the random directed sector graphsrequire us to define descendant and antecedent neighborsfor every node. The descendants of node ni are defined asDi , {nj | ∀ j : Ei,j = 1}, i.e., the set of nodes who lieswithin the coverage region of ni, antecedents of ni are definedas Ai , {nj | ∀ j : Ej,i = 1} the set of nodes who can reachto ni. In Fig. 16b, the set of descendants and antecedents ofni are shown as {nj , nk, nl} and {ng, nh, nf}, respectively.

In order to find out the probability of a connected UOWN,we have considered networks of m = 100 and m =500 optical sensor nodes randomly deployed in underwater100 m× 100 m square area respectively. The probability of aconnected network is evaluated when each node is connectedto at least one node (k = 1) and when each node is connectedto at least two other nodes (k = 2). The transmission rangeR varies from 1 to 20 meters and we set the beam scanningangles of the nodes with different widths of φ = 2π

9 ,π2 ,

3π4 ,

and 2π to see the impact of scanning angles on the probabilityof a connected network. Fig. 18a - Fig. 18d shows thatincrease in the beam scanning angles, number of nodes, andtransmission range results in high probability of a connectednetwork. Table II summarizes the literature on connectivityanalysis of different wireless networks.

B. Reliability

Packet losses may occur during the transport as a resultof the hostile underwater channel impairments and network

Page 20: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

20

j

i

k

l

mn

Bidirectional Link Multi-hop Directive Links

ji

(a) Connection types in random sector directed graphs.

ji

i

i

j

i x-axis

Transmitter

Trajectory

of Node iiS

i

k l

jh

f

g

i

Angular sector of node iDescendants

of node iAntecedents

of node i

iR

(b) Connectivity parameters, descendants, and antecedents of ni.

Fig. 16: Depiction of connection types, connection parameters, descendants, and antecedents for random sector directed graphs.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

(a) m = 100, φ = π3

, and R = 0.2m. (b) m = 100, φ = 2π, and R = 0.2m.

Fig. 17: Illustration of random directed graph scenarios for different parameters.

TABLE II: Comparison of connectivity analysis of wireless communication systems

Literature Channel model Link type Graph model[257]–[264] Terrestrial RF Bidirectional Random graphs[61], [265]–[268] Underwater Acoustic Bidirectional Random graphs[269]–[274] Terrestrial Optical Bidirectional/Unidirectional Random graphs/Random sector graphs[48], [51], [255], [256] Underwater Optical Bidirectional/Unidirectional Random sector graphs

congestion. Hence, transport protocol can check the data cor-ruptions by means of error correction codes and verify the cor-rect receipt by the ACK/NACK messages to the source node.Considering the relation between a node and its antecedentsand descendants as described above, optical sensor nodes mayalways not be able to convey ACK/NACK messages to itsantecedents. That is, operation of such a mechanism requiresa fully connected network such that there is always anothercommunication path to deliver ACK/NACK messages to thesource node. Hence, it is essential to handle shadow zoneswhere temporary connectivity loss and high bit error ratesoccur [69]. Transmission control protocol (TCP) is the best-known connection-oriented transport layer protocol which as-sumes congestion as the only cause of packet loss and reducesthe rate if packet losses occur. However, obstruction, pointingand misalignment events are quite common in UOWCs and

an efficient UOWN transport layer protocol must distinguishbetween packet losses due to the congestion and channelimpairments. Alternatively, user datagram protocol (UDP) isa connection-less transport layer protocol which may suitethe UOWN better for very simple transmission applications.Rather than traditional end-to-end approaches, reliability canalso be characterized in a hop-by-hop fashion. However, ahop-by-hop based reliability may not guarantee an end-to-endreliable network. Therefore, UOWNs paradigm necessitatesnovel transmission protocols which ensures the reliabilityby accounting for the underwater channel impairments andlimited connectivity of UOWNs.

C. Congestion and Flow ControlCongestion control is needed in order to avoid from being

congested due to oversubscription of many traffic flows which

Page 21: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

21

2 4 6 8 10 12 14 16 18 20Tranmission Range (m)

0

0.2

0.4

0.6

0.8

1P

om=100, k=1 (Ana)m=100, k=1 (Sim)m=100, k=2 (Ana)m=100, k=2 (Sim)m=500, k=1 (Ana)m=500, k=1 (Sim)m=500, k=2 (Ana)m=500, k=2 (Sim)

(a) Probability of connectivity vs. transmission range for φ = 2π9

.

2 4 6 8 10 12 14 16 18 20Tranmission Range (m)

0

0.2

0.4

0.6

0.8

1

Po

m=100, k=1 (Ana)m=100, k=1 (Sim)m=100, k=2 (Ana)m=100, k=2 (Sim)m=500, k=1 (Ana)m=500, k=1 (Sim)m=500, k=2 (Ana)m=500, k=2 (Sim)

(b) Probability of connectivity vs. transmission range for φ = π2

.

2 4 6 8 10 12 14 16 18 20Tranmission Range (m)

0

0.2

0.4

0.6

0.8

1

Po

m=100, k=1 (Ana)m=100, k=1 (Sim)m=100, k=2 (Ana)m=100, k=2 (Sim)m=500, k=1 (Ana)m=500, k=1 (Sim)m=500, k=2 (Ana)m=500, k=2 (Sim)

(c) Probability of connectivity vs. transmission range for φ = 3π4

.

2 4 6 8 10 12 14 16 18 20Tranmission Range (m)

0

0.2

0.4

0.6

0.8

1

Po

m=100, k=1 (Ana)m=100, k=1 (Sim)m=100, k=2 (Ana)m=100, k=2 (Sim)m=500, k=1 (Ana)m=500, k=1 (Sim)m=500, k=2 (Ana)m=500, k=2 (Sim)

(d) Probability of connectivity vs. transmission range for φ = 2π.

Fig. 18: Probability of connectivity vs. different transmission angles and ranges.

may not be affordable by available network capacity whereasflow control is required to manage the sender’s transmissionrate in order to prevent buffer overrun at the receiver. Due tothe window-based mechanism which relies upon the accurateround trip time (RTT), most of the TCP implementationsare unsuited for underwater acoustic networks as they incurend-to-end delay with high mean and variance [69]. Even ifUOWCs provide very high propagation speeds, a potentialtransport protocol still needs to take the link failures becauseof the dynamic topology changes of UOWNs into account.Alternatively, rate-based transport protocols do not depend onwindows-based mechanism and can provide a flexible ratecontrol, however, it requires feedback messages to dynamicallyadapt the transmission rate according to the packet losses. Therate-based scheme is not appropriate for UOWNs due to highmean and variance of feedback delay [60] where some of theUOWN nodes may not receive any feedback messages if thereis not a connecting path from the receiver to the transmitter incase of limited connectivity. Accordingly, proposed congestionand flow control mechanisms should account for such kind

of specific challenges related to UOWNs. It is especiallyimportant to leverage critical information from lower layersto predict and handle shadow zones as connectivity can beregarded as the main delimiter of any potential transport layerprotocol.

VII. APPLICATION LAYER

Even though one can count numerous applications forUOWNs, application layer protocol is a completely unexploredarea of research. The main purpose of a potential applicationlayer protocol is multifold [60]: 1) to provide a mediatinglanguage to query the entire UOWNs, 2) to advertise eventsand assign the tasks, and 3) to provide efficient networkmanagement tools which can see and manipulate the hardwareand software features of the lower layers. Having these func-tionalities in the hand, application layer protocols are neededto be customized according to the QoS requirements of targetapplications, which are summarized in Fig. 19 and surveyedbelow:

Page 22: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

22

Applications of UOWNs

Ocean Sampling

Environmental Monitoring

Navigation SurveillanceMine

Reconnaissance

MilitaryDisaster

Prevention

Underwater Exploration

Underwater Habitat

Water Quality

Fish FarmMarine

LifeCoral Reef

Earthquake, Volcano, and

TsunamiFloods Oil Spills

Fig. 19: Classification of UOWNs applications.

1) Ocean Sampling: Ocean sampling provides an embed-ded ocean research capability with the help of mobile andnetworked sensors. In a comprehensive form, ocean samplingcombines observation tools with efficient modeling to reduceerror in the state estimation for oceans. The network ofunderwater optical sensors and AUVs can perform differenttasks such as synoptic adaptive sampling of three dimen-sional coastal of oceans, measuring the physical properties ofoceans, ecosystem productivity, and other fundamental tests tounderstand the behavior and capabilities of ocean processes.Some of the well-known projects developed on ocean samplingare autonomous ocean sampling network [276], Bermuda bio-optics project [277], Littoral ocean observing and predictionsystem [278], [279], coastal ocean dynamics [280], and oceanresearch interactive observatory networks [281].

2) Environmental Monitoring: Environmental monitoringapplications of UOWNs are specifically related to monitor thephysical underwater environment. Underwater environmentalmonitoring applications can further be categorized into threemajor categories i.e., monitoring of underwater exploration,monitoring of underwater habitat, and monitoring of the waterquality.

• Monitoring of Underwater Exploration: There are abun-dant resources such as oil and gas present in the un-

derwater environment which is required to be explored.Water is covering most part of the Earths surface, the dryparts of the Earth are connected by underwater cablesand pipelines. These underwater cables and pipelinesprovide some of the most basic necessities for instancegas pipelines, oil, and optical fiber. Therefore, UOWNscan be used to monitor these underwater cables andpipelines, as well as UOWNs can be used to explorethe underwater precious resources. In [282], the authorshave developed an underwater monitoring system to findthe manganese crust. AUVs were used to observe themanganese crust on the seabed at Katayama sea-mountand detailed three-dimensional seabed images were takenby the AUVs for investigation. An underwater monitoringsystem was proposed in [283] which combines ROVsand AUVs to discover the underwater mineral resources.Large and deep scale areas can be scanned using thissystem for oceans exploration. A deep ocean monitoringsystem was also proposed in [284] for ocean exploration.The monitoring system was tested in coastal areas bydeploying cameras in the ocean. Detailed literature ondifferent ocean exploration monitoring systems and theirpossible architectures using underwater sensor networkswere presented in [45], [46], [60], [61], [285].

Page 23: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

23

• Monitoring of Underwater Habitat: The study of under-water habitat is one of the most interesting and chal-lenging fields of natural sciences. Underwater habitatmonitoring includes fish farm monitoring, marine lifemonitoring, and study of underwater coral reef and plants.

– Fish Farm Monitoring: Fish farming is considered tobe a good economical resource, but it requires strictmonitoring of the habitat conditions. For the moni-toring purposes, underwater sensors are deployed tomonitor the underwater habitat. In [286] and [287],the authors have developed an underwater sensornetwork which was able to estimate the amount ofresidual food and waste in the underwater habitat.Similarly, an underwater monitoring system wasdeveloped in [288] to maintain the ecosystem fortrout fish by sensing the water quality. A Zigbeebased underwater sensor network has been proposedin [289] to monitor the properties of small fishfarms. The underwater sensors were able to sensethe temperature, pH, and NH4+ of underwater habi-tat and send the information to the central stationusing Zigbee. An underwater monitoring system wasdeveloped in [290] for large lakes and fish farms. Theproposed system was able to measure the depth ofthe farm and acidic level of the water. The depthwas measured by using the optical sensors whileacidic levels of water were measured by using thepH and oxygen sensor. An increased lifetime waterquality monitoring system was proposed in [291] forfish farms and tested at the farming cages located inMediterranean sea, Italy. Recently a semi-automaticfish monitoring system has been developed in [292]which monitors the migration of fish by using thevideos taken by the AUVs.

– Marine Life Monitoring: The applications of marinelife monitoring include overseeing various speciesrelated to oceans. An underwater marine life mon-itoring system was developed in [284] which wasable to monitor the marine life in the underwaterenvironment within its coverage area. An underwatersensor network was deployed in [293] to monitor themarine life at different levels. In the proposed systemthe sensor nodes were divided into small clustersand the cluster head directly sends the sensed in-formation to the surface buoy. The proposed systemwas deployed in Queensland, Australia to monitorunderwater temperature and luminosity. The authorsin [294] claimed a cost-effective monitoring systemcalled smart environmental monitoring and analysistechnology (SEMAT). SEMAT was used for marinelife research and water quality monitoring. Anothermarine life monitoring system was developed in[295] and tested at Menor coast, Spain. An intelligentarchitecture and different protocols for marine lifestudy have been studied in [296]. An interestingseashell monitoring system has been deployed inZhejiang, China [297]. A detailed survey on marine

life research using underwater sensor networks ispresented in [298].

– Coral Reef Monitoring: Coral reefs are one of theimportant and diverse underwater ecosystems whichare built by the microorganisms. In [284] the au-thors have developed an underwater sensor networkfor coral reef monitoring. In the proposed systemsstationary sensor nodes were deployed at the tar-get coral places. AUVs were used to deploy thesensors and collect the data from the sensors. Theprototype of the proposed system was deployed atOkinawa, Japan. An intelligent surface buoy has beendeveloped in [293] to monitor the coral reef. Theproposed system harvest energy from sea waves, thusimproving the lifetime of the monitoring system. In[299] and [300], underwater sensor networks havebeen deployed to monitor coral reef at northeasternAustralia. The proposed system is still in operationfrom last two years and reliably monitoring thelargest coral reef ecosystem on planet Earth.

• Monitoring of Water Quality: Life has started from Waterand it is an essential resource on Earth for living organ-isms. Therefore, it is very important to keep an eye on thequality of water. An application to monitor the quality ofpool water for trout farms was developed in [288], [301].Different parameters of the trout farm such as demandfor oxygen, ammonium nitrogen, electrical conductivity,and pH, were monitored for 270 days. A low cost andeffective water quality monitoring system was presentedin [302] for lakes and ponds which consumes low powerand reliable data transmission. Similarly, an underwatersensor network has been developed in [303] to monitorthe quality of drinking water. Sensors were integratedwith AUVs to collect the water samples and informationwas sent to the surface station for investigation. Tomonitor the quality of water in Indian rivers, a waterquality monitoring system has been deployed in [304].In addition, pollution and wreckage detection techniqueshave been presented to monitor the quality of water [305].A comprehensive survey of water quality monitoringsystems by using sensor networks is presented in [306].

3) Navigation: Underwater habitat is extremely irregularand dark with growing depth. Therefore, navigation in suchan environment is a challenging task and require assistance.The common assistive technologies used for navigating theships, boat, and vessels on the surface of water cannot beused for underwater navigation due to the different mediumfor transmission. Underwater sensor networks are the promis-ing technology to assist the navigation in the underwaterenvironment. In [307] the authors have proposed an assis-tive navigation system using AUVs for underwater sensornetworks. An on-demand underwater technique for locatingthe underwater sensors was proposed in [308]. In short, manyresearchers developed assistive localization schemes for theunderwater environment based on acoustic waves. Navigationand localization of underwater acoustic networks have beenstudied widely in the past and number of surveys are written

Page 24: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

24

on this subject [42]–[46]. Since the UOWC channel posesnew challenges, the existing localization techniques used forterrestrial wireless networks and underwater acoustic networksare not directly applicable to UOWNs. Therefore, novel timeof arrival (ToA) and received signal strength (RSS) baseddistributed localization schemes are developed in [47] forUOWNs. Recently, in [48]–[50] we have proposed RSS basedcentralized localization schemes for UOWNs.

4) Surveillance: Underwater surveillance is quite impor-tant, especially for intruder detection. Underwater sensor net-works can be used for offshore and onshore surveillance.Onshore surveillance applications include battleships detec-tion and logistics arrival. The warfare surveillance system(GLINT10 field trail) has been tested in [309]. AUVs wereused in complete autonomous fashion with signal processingcapabilities to act and share the information about the un-derwater battlefield. The protection of offshore and onshoreequipment’s and infrastructures were investigated in [310] byusing the underwater acoustic network. The proposed systemwas composed of four acoustic stationary nodes, one mobilenode, and two AUVs. The proposed system was deployed inNorway and worked successfully for continuous five days. Anovel layout for underwater surveillance has been proposed in[311] which consists of surface sensors. The surface sensorsare then moved to specific depths to get maximum coverage.Data mining tools were used to classify and detect differentobjects such as submarines and divers. Electromagnetic wavesbased different architectures for underwater surveillance wereproposed in [312]. For the interesting readers, we refer to[313], which is the most recent survey presented on underwatersurveillance systems .

5) Mine Reconnaissance: As the sensors are able to sensedifferent physical parameters, it is rational that sensors candetect underwater mines. Detection of underwater mines canassist the ships for a safe voyage. An underwater minedetection system has been developed in [314] by Naval warfarecenter, Florida which can detect the underwater clutters. Anunderwater mine detection systems have also been proposedin [315]–[319] which considers image processing techniquesto localize the mines. Four different types of AUVs and twodifferent sonars were used in [320] to find out the underwatermines. The proposed system was tested in the Chesapeake Bay,U.S. to detect different kind of underwater mines. A machinelearning and deep learning approaches were used in [321],[322] to classify the underwater mines from other objects.

6) Military: Underwater sensors are also deployed for anumber of military applications. The underwater networkconsists of a number of high-resolution cameras, sonars, andmetal detectors combined with AUVs to detect mines, securesubmarines, and ports. Norwegian defense has developed anAUV suitable for underwater military applications. This AUVwas named as HUGIN 1000 and it was able to work indifferent applications such as localization and classification ofunderwater objects, mine detection, route surveys, and envi-ronment assessment [323]. U.S. military has recently launcheda project of worth 37 million US dollars which involves the de-velopment of intelligent underwater surveillance sensors. Thisproject is named as Ocean of things with two major purposes:

building efficient low-cost underwater sensors and developingdata processing techniques to get the useful information [324].

7) Disaster Prevention: Natural disasters are imminentamong which water-based disasters are deadly and causeenormous destruction. Due to the insufficient resources andchallenging environment of oceans, development of disastermonitoring system for water-based natural disasters is still asignificant challenge. Prediction of a natural disaster using adisaster monitoring system and taking preventive mechanism isvery important. Underwater sensor networks offer broad rangeof disaster monitoring applications from predicting a disasterto the aftermath of a disaster. Water-based disasters includeunderwater volcanic eruptions, floods, underwater earthquakes,and oil spills.• Earthquake, Volcano, and Tsunami: Monitoring of un-

derwater earthquakes and volcanoes is very importantotherwise it can lead to enormous destruction. In [325],an early warning system was proposed by using un-derwater sensor networks for earthquakes and tsunamis.An efficient architecture of underwater sensors to detecttsunamis was presented in [326]. The proposed sys-tem utilized seismic pressure to detect the tsunami andtransmitted the information to the surface station. Tothe best of our knowledge, there is a lot of literaturewhich address underwater natural disasters but very fewpractical underwater disaster monitoring systems exist.

• Floods: The destruction caused by floods and its increasedfrequency require to timely generate flood alerts. Togenerate timely flood alerts deployment of underwaterand on the surface sensors can be used. A flood moni-toring system was proposed in [327] which consists ofan AUV, sensors, and a remote station. The authors in[328] have proposed a flood warning system based onthe underwater sensor network. The proposed systemwas tested and implemented in 650 km2 watershed insouthern Spain. Similar flood monitoring systems havealso been implemented in 15 different flood regions inNigeria [329]. Recently, a cloud computing based floodmonitoring system have been proposed in [330].

• Oil Spills: Large underwater oil spills cause a seriousbiological impact on marine life. Therefore, to monitorthese large oil spills, underwater sensor networks providea good solution. In [305], an efficient underwater sensornetwork was deployed to detect the ocean pollution.The thickness and location of the oil spill have beeninvestigated in [331]–[333] by using underwater sensors.Optical detection methods were used in [334], [335] todetect underwater oil spills with the help of LEDs.

VIII. LOCALIZATION IN UNDERWATER OPTICALWIRELESS NETWORKS

Numerous acoustic based localization techniques have beenwell investigated in the past since it is important for tagging thedata, detection of an underwater object, tracking of underwaternodes, underwater environment monitoring, and surveillance.Nevertheless, due to the challenges discussed in previoussections for each layer of UOWNs, there is a dire need to

Page 25: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

25

Underwater Localization Algorithms

Based on Channel Based on ComputationBased on AnchorsBased on Ranging

Time based

Angle based

Received Power based Acoustic Optical Centralized Distributed

Estimation Based

Prediction Based

Prediction Based

Estimation Based

Based on Mobility

Stationary MobileHybrid

Hybrid

Fig. 20: Taxanomy of underwater localization algorithms.

develop novel localization techniques for UOWNs. Therefore,this section provides the fundamental concepts of underwaterlocalization, state of the art underwater localization systems,and development of localization techniques for UOWNs.

A. Basics of Underwater Localization

Localization of underwater sensors is an important part ofUOWNs for many applications such as resource exploration,surveillance, underwater environment monitoring, and disasterprevention. The large propagation delay of acoustic channelsand high attenuation of RF/optical channels pose significantchallenges for underwater localization. The major challengesfor underwater localization are• Deployment of nodes: Most of the localization algorithms

depends on the distribution of sensor nodes and theanchor nodes to form a network [44], [336].

• Mobility of the nodes: Due to the uncontrollable phe-nomena such as winds, turbulence, and current, the un-derwater sensor nodes inevitably drifts from its position.The location of anchor nodes on the surface buoys canbe accurately measured by using GPS but the location ofthe underwater nodes cannot be precisely measured [43],[44].

• Harsh underwater channel: Variations in the underwaterwireless communication channel is very severe for alltype of carrier waves. The effects of attenuation, ab-sorption, reflection, scattering, and noise do not allowfor accurate range measurements, thus reflecting largelocalization estimation error [45].

• Synchronization: As the GPS signals are not availablein the underwater environment, it is hard to achievethe time synchronization between the sensor nodes.Thus, if the time of arrival based ranging is used, thismiss-synchronization will lead to large localization error[337]–[339].

A large number of underwater localization algorithms havebeen proposed in the past for underwater acoustic wireless

communication systems. All of these localization algorithmsconsider different parameters of the network such as networktopology, range measurement technique, energy requirement,and device capabilities. In addition, the accuracy of localiza-tion algorithms also depends on many other factors whichinclude propagation losses, number of anchor nodes, thelocation of anchor nodes, time synchronization, and scheduling[340]. The underwater localization algorithms can be classifiedbased on different parameters such as range-based/range free,anchor-based/anchor free, acoustic/optical, stationary/mobile,and centralized/distributed. Centralized and distributed algo-rithms can further be classified into estimation based algo-rithms and prediction based algorithms. The taxonomy ofunderwater localization schemes is shown in Fig. 20. Everyunderwater localization algorithm requires distance estimationbetween the nodes or between the node and anchors. Thedistance is estimated by using acoustic ranging or opticalranging for UWC systems.

1) Acoustic Ranging: Underwater acoustic channels sufferfrom two kinds of major losses, i.e., attenuation loss andspreading loss [341]. Attenuation loss is a result of scattering,diffraction, absorption, and leakage from ducts while spreadingloss is a combination of cylindrical and spherical losses [342].Acoustic ranging based localization algorithms can further beclassified as time-based, received signal strength based, andangle based algorithms. Most of the underwater localizationalgorithms are time-based which includes time of arrival (ToA)and time difference of arrival techniques (TDoA).

• ToA based acoustic ranging: ToA based ranging is one ofthe most popular ranging technique used for underwaterdistance estimation. Indeed, ToA is more preferred inunderwater acoustic systems compare to terrestrial sys-tems since the ToA technique for RF signals requireshigh-resolution stable clocks. But as the speed of soundwaves is very slow compared to the speed of RF signal,ToA is best suited for underwater ranging [343]. InToA based underwater acoustic localization systems the

Page 26: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

26

anchor nodes transmit the acoustic signals and eachsensor nodes require to receive the acoustic signal from atleast three anchor nodes to carry out trilateration and findits estimated position. In [344], [345] the authors havedefined the objective function for ToA based underwateracoustic localization as

f(x, y) =

√∑N

i=1(di −

√(x−Xi)2 + (y − Yi)2)2

(21)where x and y is the two-dimensional estimated locationof a sensor node, N is the total number of anchor nodes,di is the estimated distance to anchor node i, and Xi Yiis the two-dimensional location of anchor node i. Theunderwater acoustic ToA distance estimation is mainlyaffected by dispersion of underwater acoustic channels,multipath fading, and Doppler spread [346].

• TDoA based acoustic ranging: In ToA based rangingtechniques the sensor nodes and anchor nodes must besynchronous while in TDoA based ranging the sensornode do not need to be synchronized to the anchor nodes,thus, mitigating the limitation of time synchronizationrequirement of ToA techniques [347]. The authors in[348] have proposed a TDoA based underwater local-ization system called silent positioning which relies onthe range differences collected from four different anchornodes. The problem of three-dimensional underwaterlocalization by using TDoA based acoustic ranging hasbeen studied in [62], [349], [350]. The performance ofTDoA based localization was investigated in [351] forthe reverberant underwater environment and multipathpropagation channel. Localization of underwater acousticnetworks has been investigated in [352] by using Haus-dorff distances for TDoA ranging. Localization in shallowwater using TDoA measurements was achieved in [353]by considering two hydrophones mounted under the boat.In [354], the authors have implemented an accurate andprecise underwater acoustic localization system by usingTDoA measurements with the help of a microphone and aspeaker. Recently, a real-time acoustic ranging techniquehas been proposed in [355] to improve the accuracy ofTDoA measurements. The TDoA ranging can providebetter accuracy compare to ToA ranging at the expenseof more complexity and cost of the system [348].

• RSS based ranging: Due to the problems of attenuationand spreading loss for underwater acoustic channel, RSSis not well suited for underwater ranging. However, dueto the simplicity and nice transmission behavior of RSSat certain depth [356], it was used in [357] for underwatersource localization. Source localization in an underwaterenvironment by using RSS was also studied in [358] byusing Thorp’s propagation model [359].

2) Optical Ranging: Optical light passing through theaquatic medium suffers from widening and attenuation inangular, temporal, and spatial domains [47]. In the literature,only ToA and RSS based localization techniques exist forUOWNs. In [47] the authors have proposed an underwateroptical positioning system, where an OBS was considered as

an anchor node which transmits optical signals. The sensornodes receive the optical signals from multiple anchors andlocate itself using simple linear least square solution. In [163],an RSS based distance estimation technique is developedfor UOWNs for a given data rate. The RSS based distanceestimation strongly depends on different parameters such ascharacteristics of the underwater optical channel, divergenceangle of the transmitter, field of view of the receiver, transmit-ted power, and trajectory angle. For a LoS link and achievabledata rate Rji , the estimated distance dij between node i and jis obtained in [255] from (11) as

dij =2 cosϕjic(λ)

W0

c(λ)

2 cosϕji

√2πThcRj

irj(1−cos θi)

ηjtλPtiηitηjAj cosϕj

i

, (22)

where T is the pulse duration and W0(.) is the real part ofLambert W function [361]. Table III summarizes the literatureon different ranging techniques for underwater localizationsystems.

B. State of the Art Underwater Monitoring and LocalizationSystems

In the past various techniques have been used by theoceanographers for oceans exploration. The most commonmonitoring equipment include ocean floor sensors, floatingsensors, surface buoys and surface stations. Sensed data fromthe sensors on the ocean floor is collected by the surface buoys.The surface buoys are fixed and they can send the collecteddata to the surface station using wired or wireless communica-tions. In case of floating sensors, the sensors do not have fixedlocation and drift with ocean currents. Floating sensors aredynamic in nature and they can sense a reverberant underwaterenvironment. At present, the largest ocean monitoring systemis developed by Global Ocean Data Assimilation Experiment(GODAE) and Oceanview called Argo [360], [362]. Argoconsists of 3800 free drifting floating sensors which measurethe salinity, currents, and temperature of the ocean up to2000 m of depth. The location of Argo float is determinedusing GPS once it is on the surface of the ocean and italso transmits the data to the onshore station using satellitecommunication. In Argo project, the floats do not interactwith each other and work independently. Fig. 21 shows thecurrent distribution of Argo floats in oceans all over the world.China has announced recently a similar project to Argo, tobuild underwater monitoring systems across the south and eastChina seas for intruder detection [363]. In 1980, the U.S. Navyhas developed a large scale network of underwater devicescalled Seaweb [364]. Seaweb consisted of AUVs, surfacebuoys, gliders, repeaters, and surface stations. Seaweb has usedacoustic waves for underwater communication and RF wavesfor terrestrial communication.

Acoustic localization systems for underwater monitoringutilizes two different approaches, namely long baseline (LBL)and short baseline (SBL) [365]. In the LBL approach, theacoustic transponders are installed in the underwater operationarea. Sensor nodes that are in the coverage of these acoustic

Page 27: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

27

Fig. 21: Distribution of floating sensors in Argo [360].

TABLE III: Comparison of different ranging techniques for underwater localization.

Literature Channel model Ranging Technique Accuracy Complexity[343]–[346] Acoustic ToA High Moderate[62], [347]–[353] Acoustic TDoA High High[356]–[358] Acoustic RSS Low Low[47] Optical ToA High Moderate[47] Optical RSS Low Low

transponders respond by using a certain ranging method andlocalize itself either by using triangulation or trilateration[366]. In the SBL approach, the surface station follows theunderwater sensor nodes and transmits short-range acousticsignals for the sensor nodes to localize itself. The SBLunderwater positioning systems have been used by WoodsHole Oceanographic Institution to find out the location of adeep underwater ROV [367]. In addition to the LBL and SBLapproaches, there exists an underwater localization systemcalled GPS intelligent buoy (GIB). The GIB is a commercialsystem in which the surface buoy acts as a relay betweenthe surface station and the seabed sensors. GIB collects thedistance estimation from the sensors to itself and sends itto a central station where the central station finds the globalview of all the seabed sensors. GIB systems have numerousapplications which include weapon testing and training [368],

tracking AUV’s [369], global view of the network [369], andintruder detection [370]. Table IV summarizes some of thewell known commercial underwater localization systems.

C. Localization Techniques for UOWNs

UOWNs localization is one of the major research areasnowadays because of the development of high-speed UOWCsystems. Localization in terrestrial wireless networks has beenstudied widely and detailed surveys are presented on this topic[36]–[41]. Nevertheless, GPS and all of these RF-based local-ization schemes cannot work in the underwater environment.Thus, many researchers developed localization schemes for theunderwater environment based on acoustic waves. Localizationof underwater acoustic networks have also been studied widelyin the past and number of surveys are presented on this subject

Page 28: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

28

TABLE IV: Comparison of different commercial underwater localization systems.

System Company Channeltype

Approach Accuracy Applications

Underwater acoustic LBLpositioning [371].

Evo Logics Acoustic LBL 1.5 cm Offshore positioning, navigation, car-tography, geodesy, and sensors track-ing.

HiPAP - Acoustic underwa-ter positioning and naviga-tion systems [372].

KONGSBERG Acoustic SBL 2 cm Seabed positioning of vessels, sub-seameteorology, and telemetry.

Mini-Ranger 2 UnderwaterPositioning (USBL) system[373].

Sonardyne Acoustic UltraSBL

2 cm Oil and gas exploration, marinerobotics, and marine security.

USBL positioning systems[374]

iXblue Acoustic UltraSBL

6 cm Hydrography, maritime vessels, oceanscience, and defense.

VideoRay ROV PositioningSystems [375]

KCFTechnologies

Acoustic UltraSBL

150 cm Navigation, tracking, search and res-cue, and target detection.

TrackLink Acoustic Track-ing Systems [376]

LinkQuestInc

Acoustic UltraSBL

0.5 cm Navigation, tracking, underwater sur-veys, and oil and gas exploration.

Teledyne Benthos underwa-ter acoustic systems [377]

Teledyne Ma-rine

Acoustic LBL/UltraSBL

5 cm Navigation, tracking, underwater sur-veys, and oil and gas exploration.

Bluecomm UOWC [378] Sonardyne Optical Notdefined

- High speed underwater wireless com-munication.

Anglerfish UOWC [379] STM Optical Notdefined

- High speed underwater wireless com-munication.

[42]–[46]. Since the localization techniques used for terrestrialwireless networks and underwater acoustic networks cannotbe directly applied to UOWNs, novel localization schemeshave recently been developed for UOWNs. We divide theselocalization schemes into two categories as distributed andcentralized schemes. In distributed localization schemes, everyunderwater optical sensor node localizes itself by communi-cating with multiple anchor nodes. In centralized localizationschemes, the underwater optical sensor nodes do not localizethemselves but the location information is sent to them by thesurface buoy or sink node periodically.

1) Distributed Localization Schemes for UOWNs: In thissection, we summarize two distributed UOWNs localizationschemes where one is based on ToA ranging and the other isbased on RSS based ranging.

• ToA Based Scheme: In [47], the authors have proposedfor the first time a ToA based underwater optical wire-less positioning system. The authors considered an OBSplaced in an underwater hexagonal cell and a num-ber of users with transceivers capable of UOWC. EachOBS consists of 60 green LEDs forming an underwaterOCDMA network where the modulation scheme of OOKis considered. For the ToA scheme, first, the distanceis estimated between the users and the OBS by usingthe relationship of the transmission time of an opticalsignal, speed, and the distance. It is assumed that all theOBSs and the users are synchronized, and all the OBSstransmit the beacon signals at τ = τ0. The users receivemultiple beacon signals from multiple OBSs at differenttimes namely τ1, τ2 τ,..., τm, where m are the numberof OBSs. Different underwater channel impairments such

as turbulence, current, and multipath lead to the distanceestimation error for ToA ranging. Once the ToA basedestimated distances are available from at least threeOBSs, the user was able to locate itself in two dimensionsby using linear least square solution.

• RSS based Scheme: As the optical signal from the OBSsto the user passes through the underwater environmentit suffers from attenuation, absorption, and scattering.The underwater user requires the RSS signals from atleast three OBSs in this case as well. RSS scheme haslow cost because every transceiver is able to estimatethe received signal power. However, the RSS based dis-tance estimation requires precise modeling of the channel[182], [380]. The RSS based distance estimation stronglydepends on different parameters such as characteristicsof the underwater optical channel, divergence angle ofthe transmitter, field of view of the receiver, transmittedpower, and trajectory angle. The widening and attenuationof the underwater optical signals are dependent on thewavelength. Monte Carlo simulations were used in [47]to find out the RSS based distances. Once the RSS baseddistances were estimated to at least three OBSs, the userwas able to locate itself in two dimensions by using linearleast square solution.

2) Centralized Localization Schemes for UOWNs: In cen-tralized UOWNs localization schemes, the underwater userdoes not localize itself but the location is sent to the userby the surface buoy or sink node periodically. To the bestof our knowledge, only RSS based centralized localizationschemes are developed for UOWNs. In [48], we have proposeda localization scheme for UOWNs with limited connectivity.

Page 29: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

29

0 10 20 30 40 50Position along X direction

0

10

20

30

40

50P

ositi

on a

long

Y d

irect

ion

AnchorsSensor NodesEstimated locations

(a) Localization performance of ToA based distributed UOWNs [47]

0 10 20 30 40 50Position along X direction

0

10

20

30

40

50

Pos

ition

alo

ng Y

dire

ctio

n

AnchorsSensor NodesEstimated locations

(b) Localization performance of RSS based distributed UOWNs [47]

0 10 20 30 40 50Position along X direction

0

10

20

30

40

50

Pos

ition

alo

ng Y

dire

ctio

n

AnchorsSensor NodesEstimated locations

(c) Localization performance of ToA based centralized UOWNs[48], [50]

0 10 20 30 40 50Position along X direction

0

10

20

30

40

50

Pos

ition

alo

ng Y

dire

ctio

n

AnchorsSensor NodesEstimated locations

(d) Localization performance of RSS based centralized UOWNs[48], [50]

Fig. 22: Localization performance of ToA and RSS based distributed and centralized UOWNs.

TABLE V: Comparison of different UOWNs localization schemes.

Scheme RangingMethod

Computation Architecture

Underwater optical positioning systems [47] ToA Distributed OpticalUnderwater optical positioning systems [47] RSS Distributed OpticalUOWNs localization with limited connectivity [48] RSS Centralized OpticalEnergy harvesting empowered UOWNs localization [50] RSS Centralized OpticalEnergy harvesting hybrid acoustic/optical UOWNs lo-calization [49]

RSS Centralized Hybrid acoustic/optical

As the transmission range of users in UOWNs is limited, amultihop UOWNs is considered and the single neighborhooddistances are computed by using RSS. Using these singleneighborhood distances a novel distance completion strategywas used by the surface station to get the global view ofthe whole network. In [50], we have presented an energyharvesting empowered underwater optical localization schemewhere the underwater sensor nodes are able to harvest theenergy from ambient underwater sources. As the nodes can

harvest energy from the underwater environment, it helps toincrease the localization accuracy and lifetime of the network.Based on the harvested energy availability, the sensor nodescommunicate with its neighbor nodes and computes the RSSranges. A closed form localization technique was developed tofind the location of every optical sensor node in UOWNs. Theproposed localization technique accurately minimizes the errorfunction by partitioning the kernel matrix into smaller blockmatrices. Furthermore, a novel matrix completion strategy

Page 30: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

30

was introduced to complete the missing elements in blockmatrices, which results in better approximation. In [49], wehave proposed an energy harvesting localization scheme forhybrid acoustic and optical underwater wireless communica-tion system. A weighting strategy was used in [49] to givemore preference to accurate measurements.

3) Comparison of Localization Schemes for UOWNs:In order to compare the different localization schemes forUOWNs, we have simulated two different scenarios where theactual locations of the sensor nodes and anchor nodes are keptsame in both scenarios for fair comparison. To evaluate theperformance of distributed ToA and RSS based UWONs lo-calization schemes proposed in [47], we considered 50 opticalsensor nodes deployed randomly in 50 m× 50 m square areaand 4 anchor nodes deployed at each corner of the consideredarea. The optical sensor nodes are able to communicate withat least three anchor nodes directly and localize itself usinglinear least square solution. Fig. 22a and Fig. 22b shows thelocalization performance of the two schemes with root meansquare error of 0.8 m and 1.6 m respectively. To evaluatethe performance of centralized ToA and RSS based UWONslocalization schemes, we have considered the same scenario of50 optical sensor nodes deployed randomly in 50 m × 50 msquare area and 4 anchor nodes deployed at each corner ofthe area. But here the limited transmission range of opticalsensor nodes is taken into account which leads to multi-hopUOWNs. In this case, the internode single hop distances aremeasured by the optical sensor nodes and sent to the surfacestation via surface buoys. The surface station then finds out thelocation of each optical sensor node by using dimensionalityreduction techniques and linear transformations [48], [50].Fig. 22c and Fig. 22d shows the localization performance ofthe two schemes with root mean square error of 0.3 m and 0.9m respectively. Table V summarizes the UOWNs localizationschemes.

IX. FUTURE PERSPECTIVES OF UOWNS

In the following, we will advise some potential futureUOWNs research directions.

A. Future Directions in UOWNs Research

1) UOWC Channel Modeling: To model the UOWC chan-nel, there is still a need to further investigate and analyze newtheoretical models which can either be developed analyticallyor computationally. The analytic models for UOWC channelare quite simple because of simplifying the complex natureof photon propagation but these models are either analyticallyintractable or hard to evaluate computationally. On the otherhand, computational models are complex and their time com-plexity may not be suitable to employ in a large scale network.Therefore, modeling and performance analysis of UOWNsnecessitates accurate and simple UOWC channel models asthey are building blocks of UOWNs.

2) Novel Network Protocols: The current research onUOWNs is highly concentrated on physical layer problems,which tines out toward the higher layers. To the best ofauthor’s knowledge, the networking aspects are studied only

in few papers so far [47]–[49], [97], [175], [182], [183], [255],[256], [381]–[385]. Noting that the limited communicationrange and directivity of UOWCs yield limited network connec-tivity, implementing UOWNs in real life necessitates adequateprotocols and network architectures.

First and foremost, UOWNs requires effective routing al-gorithms in order to increase the network connectivity andperformance by extending the communication range and ex-panding the coverage. Even though some of the potentialrouting protocols are highlighted in Section V-B, there isno sufficient efforts toward UOWN routing techniques except[175] which only considers a centralized routing scenario toshow impact of multihop communication on network perfor-mance. Therefore, it is quite of interest to develop distributedand dynamic routing algorithms which adapt itself accordingto environmental and network changes. Furthermore, a noveltransport layer protocols are required because UOWC channelsare quite different from terrestrial and underwater acousticwireless networks.

3) Cross Layer Design Issues: Even though we have sur-veyed the UOWNs following a strictly layered perspective,which is traditionally employed for wired networks, consid-ering a cross layer design could improve the overall systemperformance in a great extent. Indeed, QoS requirements ofapplication layer can only be satisfied by mapping theminto the lower-layer metrics such as end-to-end data rate,delay, energy efficiency, packet loss, etc. Accordingly, it isinteresting to investigate a cross-layer optimization frameworkwhich adapts physical layer parameters (e.g., divergence angle,transmission power, communication range, etc.) to channelconditions and dynamically change the routing paths to satisfyQoS requirements, avoid congestion, increase the reliability,and maintain the network connectivity.

4) Localization in UOWNs: Localization is of utmost im-portance for UOWNs where it can be used for node tracking,intruder detection, and data tagging. A greater number ofunderwater applications demands for distributed localizationschemes as they can provide online monitoring. Althoughfew research work is carried out to develop distributed [47]and centralized [48]–[50] localization schemes for UOWNs.Due to the severe UOWC channel conditions distributedlocalization schemes for UOWNs are challenging and needsfurther investigation. Limited range of UOWC links and higherenergy consumption of distributed schemes led to the develop-ment of centralized UOWNs localization schemes where thelocalization is performed at the surface station. Centralizedlocalization schemes are good to get the overall global viewof the UOWN. Moreover, the impact of localization schemeson location-based routing and clustering for UOWNs still needto be investigated. Also, the cross-layer schemes such as theimpact of link quality, connectivity, transmission range, andenergy on localization performance are open issues.

5) Practical Implementations of UOWNs: Research on im-plementation of UOWNs is limited and further need to bestudied. There is a dire need to develop underwater opticaltransceivers which can overcome the problem of link mis-alignment, low transmission range, low bandwidth, energyconsumption, and compactness. There is a great potential to

Page 31: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

31

develop more advance low cost and low power underwatertransmission light sources, receiving nodes, and energy har-vesting systems. Testing of the UOWNs also needs to becarried out in the real underwater environment. Hybrid systemscomprising of both acoustic and optical underwater wirelesscommunication system have been introduced in [386]–[388].Authors in [389] have explored a statistical analogy betweenunderwater acoustic and optical wireless links for predictingthe signal to noise ratio of underwater optical links. Theresearch on developing hybrid systems is still in its infancy andneeds proper analysis. Also, the adaptive switching between anacoustic and optical mode for different operations need extraattention.

6) Energy Harvesting for UOWNs: Underwater opticalsensor nodes have limited energy resources, which has asubstantial impact on the network lifetime. Taking the engi-neering hardship and monetary cost of battery replacement intoaccount, an energy self-sufficient UOWN is essential to maxi-mize the network lifetime. In this regard, energy harvesting is apromising solution to collect energy from the ambient sourcesin the aquatic environment and storing it in an energy buffer.As surveyed in [390], ongoing research efforts on terrestrialcommunications have shown that energy harvesting plays asignificant role in enhancing the performance. However, mostof the energy harvesting techniques are designed for outdoorenvironments and not applicable in the aquatic environment.Recently, acoustics resonators are used in [391] to acquireacoustic energy from the underwater environment and harvestto the sensor nodes. A muti-source energy harvesting systemwas proposed in [49], [50] which harvest energy from multipleunderwater sources such as acoustic resonators and microbialfuel cells (MFCs) [392], and harvested to the sensor nodes.Moreover, albeit the notable research body on designingdifferent protocols for underwater communication networks,no significant research is carried out on the energy harvestingmethods for UOWNs.

7) Towards Internet of Underwater Things (IoUTs): Therehas recently been a growing interest in developing internetof underwater things (IoUTs) which can lead to enablingvarious underwater applications [393]. In the recent past,several attempts have been made to develop routing [394],[395], scheduling [396], and data analytic [397] techniquesfor IoUTs. The IoUTs research is still in its infancy and needto be more explored. Multi-hop UOWNs can be a potentialtechnology to implement the IoUTs because of its low powerconsumption and higher data rate.

X. CONCLUSIONS

In this paper, we have presented a comprehensive survey ofunderwater optical wireless networks (UOWNs) research. Thissurvey covers different aspects of cutting-edge UOWNs from alayer by layer perspective. Firstly, each layer of UOWNs suchas physical, data link, networking, transport, and applicationlayers are briefly presented and then localization techniquesfor UOWNs are surveyed. We started with defining differentpossible architectures for UOWNs and then the issues relatedto each layer are thoroughly discussed. Besides providing the

technical background on UOWNs, we have also provided de-tails on the challenges to design a practical UOWN. Addition-ally, localization is an important task where the location of theunderwater optical sensor node can be used for node tracking,intruder detection, and data tagging. Conventional terrestrialand underwater acoustic localization schemes do not meet therequirements of UOWNs where the unfavorable behavior ofUOWC asks for novel localization schemes. Even though wehave surveyed the state of the art localization schemes forUOWNs, the subject still remains open and requires to developaccurate and practical localization schemes. To reach thisgoal, communication, networking, and localization in UOWNsrequire more research efforts. In short, this survey can help thenovice readers to get an insight of each layer and localizationof UOWNs which can lead to the development of practicalUOWNs.

REFERENCES

[1] NOAA, “How much water is in the ocean?” https://oceanservice.noaa.gov/facts/oceanwater.html, online; accessed 10 Dec. 2017.

[2] C. W. T. and S. Journal Murray, Report on the Scientific Results ofthe Voyage of H.M.S. Challenger During the Years 1873-76 Under theCommand of Captain George S. Nares and the Late Captain FrankTourle Thomson. Charleston, South Carolina: Nabu Press, 2012.

[3] A. F. Laloe, The Geography of the Ocean: Knowing the ocean as aspace (Studies in Historical Geography). Howick Place, London:Routledge, 2016.

[4] O. Kruemmel, Handbuch der Ozeanographie. Bremen, Germany:Europaischer Hochschulverlag Gmbh, 2014.

[5] J. H. and Journal S. Murray and Journal Journal Adolf, The Depths ofthe Oceans. New Hampshire, US: Wentworth Press, 2016.

[6] E. J. P. and S. A. Earle, The Oceans. New York, US: McGraw-Hill,2001.

[7] M. N. Hill, The Sea. Cambridge, US: Harvard University Press, 2001.[8] R. W. Fairbridge, The encyclopedia of oceanography. New York, US:

Reinhold Pub., 1966.[9] M. Stojanovic, “Recent advances in high-speed underwater acoustic

communications,” IEEE J. Ocean. Eng., vol. 21, no. 2, pp. 125–136,Apr. 1996.

[10] A. Zielinski, Y. H. Yoon, and L. Wu, “Performance analysis of digitalacoustic communication in a shallow water channel,” IEEE J. Ocean.Eng., vol. 20, no. 4, pp. 293–299, Oct. 1995.

[11] H. Ochi, Y. Watanabe, and T. Shimura, “Basic Study of UnderwaterAcoustic Communication Using 32-Quadrature Amplitude Modula-tion,” Japanese J. of App. Physics, vol. 44, no. 6S, p. 4689, Jun. 2005.

[12] H. C. Song and W. S. Hodgkiss, “Efficient use of bandwidth forunderwater acoustic communication,” The J. of the Acous. Soc. of Am.,vol. 134, no. 2, pp. 905–908, 2013.

[13] R. K. Moore, “Radio communication in the sea,” IEEE Spectrum,vol. 4, no. 11, pp. 42–51, Nov. 1967.

[14] A. I. Al-Shamma’a, A. Shaw, and S. Saman, “Propagation of electro-magnetic waves at MHz frequencies through seawater,” IEEE Trans.Antennas Propag., vol. 52, no. 11, pp. 2843–2849, Nov. 2004.

[15] A. Shaw, A. I. Al-Shamma’a, S. R. Wylie, and D. Toal, “Experimentalinvestigations of electromagnetic wave propagation in seawater,” inEuropean Microwave Conf., Sep. 2006, pp. 572–575.

[16] C. Uribe and W. Grote, “Radio Communication Model for UnderwaterWSN,” in 3rd Int. Conf. on New Technologies, Mobility and Security,Dec. 2009, pp. 1–5.

[17] Z. Zeng, S. Fu, H. Zhang, Y. Dong, and J. Cheng, “A survey ofunderwater optical wireless communications,” IEEE Commun. SurveysTuts., vol. 19, no. 1, pp. 204–238, Oct. 2016.

[18] S. Q. Duntley, “Light in the sea∗,” J. Opt. Soc. Am., vol. 53, no. 2, pp.214–233, Feb. 1963.

[19] G. D. Gilbert, T. R. Stoner, and J. L. Jernigan, “Underwater experimentson the polarization, coherence, and scattering properties of a pulsedblue-green laser,” Proc. SPIE, vol. 0007, pp. 07 – 14, Jun. 1966.

[20] F. Hanson and S. Radic, “High bandwidth underwater optical commu-nication,” Appl. Opt., vol. 47, no. 2, pp. 277–283, Jan. 2008.

[21] S. Arnon, “Underwater optical wireless communication network,” Opt.Engineering, vol. 49, no. 6, pp. 1–15, Jan. 2010.

Page 32: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

32

[22] M. Doniec, M. Angermann, and D. Rus, “An end-to-end signal strengthmodel for underwater optical communications,” IEEE J. Ocean. Eng.,vol. 38, no. 4, pp. 743–757, Oct. 2013.

[23] J. S. Jaffe, “Underwater optical imaging: The past, the present, andthe prospects,” IEEE J. Ocean. Eng., vol. 40, no. 3, pp. 683–700, Jul.2015.

[24] J. Xu, M. Kong, A. Lin, Y. Song, X. Yu, F. Qu, J. Han, and N. Deng,“OFDM-based broadband underwater wireless optical communicationsystem using a compact blue LED,” Opt. Commun., vol. 369, no.Supplement C, pp. 100 – 105, Jan. 2016.

[25] C. Wang, H.-Y. Yu, Y.-J. Zhu, T. Wang, and Y.-W. Ji, “Multi-LEDparallel transmission for long distance underwater VLC system withone SPAD receiver,” Opt. Commun., vol. 410, pp. 889 – 895, Dec.2017.

[26] G. Cossu, A. Sturniolo, A. Messa, D. Scaradozzi, and E. Ciaramella,“Full-Fledged 10Base-T Ethernet Underwater Optical Wireless Com-munication System,” IEEE J. Sel. Areas Commun., vol. 36, no. 1, pp.194–202, Jan. 2018.

[27] M. Atef, R. Swoboda, and H. Zimmermann, “Real-Time 1.25-Gb/sTransmission Over 50-m SI-POF Using a Green Laser Diode,” IEEEPhoton. Technol. Lett., vol. 24, no. 15, pp. 1331–1333, Aug. 2012.

[28] B. Cochenour, L. Mullen, and J. Muth, “Temporal response of theunderwater optical channel for high-bandwidth wireless laser commu-nications,” IEEE J. Ocean. Eng., vol. 38, no. 4, pp. 730–742, Oct.2013.

[29] D. Tsonev, S. Videv, and H. Haas, “Towards a 100 Gb/s visible lightwireless access network,” Opt. Express, vol. 23, no. 2, pp. 1627–1637,Jan. 2015.

[30] K. Nakamura, I. Mizukoshi, and M. Hanawa, “Optical wireless trans-mission of 405 nm, 1.45 Gbit/s optical IM/DD-OFDM signals througha 4.8 m underwater channel,” Opt. Express, vol. 23, no. 2, pp. 1558–1566, Jan. 2015.

[31] H. M. Oubei, C. Li, K.-H. Park, T. K. Ng, M.-S. Alouini, and B. S. Ooi,“2.3 Gbit/s underwater wireless optical communications using directlymodulated 520 nm laser diode,” Opt. Express, vol. 23, no. 16, pp.20 743–20 748, Aug. 2015.

[32] S. P. Najda, P. Perlin, T. Suski, L. Marona, M. Leszczyski, P. Wis-niewski, R. Czernecki, R. Kucharski, G. Targowski, M. A. Watson,H. White, S. Watson, and A. E. Kelly, “AlGaInN laser diode technologyfor GHz high-speed visible light communication through plastic opticalfiber and water,” Opt. Engineering, vol. 55, no. 6, p. 55, Jan. 2016.

[33] H. M. Oubei, J. R. Duran, B. Janjua, H.-Y. Wang, C.-T. Tsai, Y.-C. Chi,T. K. Ng, H.-C. Kuo, J.-H. He, M.-S. Alouini, G.-R. Lin, and B. S. Ooi,“4.8 Gbit/s 16-QAM-OFDM transmission based on compact 450-nmlaser for underwater wireless optical communication,” Opt. Express,vol. 23, no. 18, pp. 23 302–23 309, Sep. 2015.

[34] J. Xu, Y. Song, X. Yu, A. Lin, M. Kong, J. Han, and N. Deng,“Underwater wireless transmission of high-speed QAM-OFDM signalsusing a compact red-light laser,” Opt. Express, vol. 24, no. 8, pp. 8097–8109, Apr. 2016.

[35] T.-C. Wu, Y.-C. Chi, H.-Y. Wang, C.-T. Tsai, and G.-R. Lin, “Bluelaser diode enables underwater communication at 12.4Gbps,” ScientificReports, vol. 7, no. 8, p. 40480, Jan. 2017.

[36] S. Pandey and P. Agrawal, “A survey on localization techniques forwireless networks,” J. of the Chinese Inst. of Engineers, vol. 29, no. 7,pp. 1125–1148, Mar. 2006.

[37] G. Mao, B. Fidan, and B. D. Anderson, “Wireless sensor networklocalization techniques,” Computer Netw., vol. 51, no. 10, pp. 2529– 2553, Jun. 2007.

[38] A. R. Kulaib, R. M. Shubair, M. A. Al-Qutayri, and J. W. P. Ng, “Anoverview of localization techniques for wireless sensor networks,” inInt. Conf. on Innovations in Info. Techno., Apr. 2011, pp. 167–172.

[39] N. A. Alrajeh, M. Bashir, and B. Shams, “Localization techniques inwireless sensor networks,” Int. J. of Dist. Sensor Netw., vol. 9, no. 6,p. 304628, 2013.

[40] J. Kuriakose, S. Joshi, R. Vikram Raju, and A. Kilaru, A Reviewon Localization in Wireless Sensor Networks. Cham: Springer Int.Publishing, 2014, pp. 599–610.

[41] H. Khan, M. N. Hayat, and Z. U. Rehman, “Wireless sensor networksfree-range base localization schemes: A comprehensive survey,” in Int.Con. on Commun., Comp. and Dig. Sys., (C-CODE), Mar. 2017, pp.144–147.

[42] V. Chandrasekhar, W. K. Seah, Y. S. Choo, and H. V. Ee, “Localizationin underwater sensor networks: Survey and challenges,” in Proc. of the1st ACM Int. Workshop on Underwater Networks, 2006, pp. 33–40.

[43] M. Erol-Kantarci, H. T. Mouftah, and S. Oktug, “A survey of ar-chitectures and localization techniques for underwater acoustic sensor

networks,” IEEE Commun. Surveys Tuts., vol. 13, no. 3, pp. 487–502,Mar. 2011.

[44] H.-P. Tan, R. Diamant, W. K. Seah, and M. Waldmeyer, “A surveyof techniques and challenges in underwater localization,” Ocean Engi-neering, vol. 38, no. 14, pp. 1663 – 1676, 2011.

[45] G. Tuna and V. C. Gungor, “A survey on deployment techniques, lo-calization algorithms, and research challenges for underwater acousticsensor networks,” Int. J. of Commun. Sys., vol. 30, no. 17, pp. 1–21,Jun. 2017.

[46] J. Luo, L. Fan, S. Wu, and X. Yan, “Research on LocalizationAlgorithms Based on Acoustic Communication for Underwater SensorNetworks,” Sensors, vol. 18, no. 1, pp. 1–25, Jan. 2018.

[47] F. Akhoundi, A. Minoofar, and J. A. Salehi, “Underwater positioningsystem based on cellular underwater wireless optical cdma networks,”in Wireless and Opt. Commun. Conf., (WOCC), Apr. 2017, pp. 1–3.

[48] N. Saeed, A. Celik, T. Y. Al-Naffouri, and M.-S. Alouini, “Underwateroptical sensor networks localization with limited connectivity,” in sub-mitted to IEEE Int. Conf. on Acoustics, Speech and Signal Processing(ICASSP), Apr. 2018, pp. 1–5.

[49] N. Saeed, A. Celik, T. Y. Al-Naffouri, and M.-S. Alouini, “Energyharvesting hybrid acoustic-optical underwater wireless sensor networkslocalization,” Sensors, vol. 18, no. 1, pp. 1–16, 2017.

[50] N. Saeed, A. Celik, T. Y. Al-Naffouri, and M.-S. Alouini, “Localizationof energy harvesting empowered underwater optical wireless sensornetworks,” Submitted to IEEE Trans. Wireless Commun., 2018.[Online]. Available: http://hdl.handle.net/10754/626393

[51] H. Kaushal and G. Kaddoum, “Underwater optical wireless communi-cation,” IEEE Access, vol. 4, pp. 1518–1547, Apr. 2016.

[52] L. J. Johnson, R. J. Green, and M. S. Leeson, “Underwater opticalwireless communications: depth dependent variations in attenuation,”Appl. Opt., vol. 52, no. 33, pp. 7867–7873, Nov. 2013.

[53] L. J. Johnson, F. Jasman, R. Green, and M. S. Leeson, “Recent advancesin underwater optical wireless communications,” Underwater Techno.,vol. 32, no. 3, pp. 167–175, Nov. 2014.

[54] M. G. G. Camila, S. R. D. Paulo, L. R. C. Marcello, A. M. Wallace,M. C. Felipe, and N. G. Jonathan, “A survey of underwater wirelesscommunication technologies,” J. of Commun. and Info. Sys., vol. 31,no. 1, pp. 242–255, Jun. 2016.

[55] L. K. Gkoura, G. D. Roumelas, H. E. Nistazakis, H. G. Sandalidis,A. Vavoulas, A. D. Tsigopoulos, and G. S. Tombras, “Underwateroptical wireless communication sys.: A concise review,” in TurbulenceModelling Approaches - Current State, Development Prospects, Appli-cations. InTech, 2017.

[56] M. M. Alomari, M. Wafa, A. Rehab, G. G. Babhair, and M. Hemalatha,“Vision and challenges of underwater optical wireless communication- A survey,” Int. J. of Computer Applications, vol. 167, no. 8, pp. 8–10,Jun. 2017.

[57] G. A. Hollinger, S. Choudhary, P. Qarabaqi, C. Murphy, U. Mitra, G. S.Sukhatme, M. Stojanovic, H. Singh, and F. Hover, “Underwater datacollection using robotic sensor networks,” IEEE J. Sel. Areas Commun.,vol. 30, no. 5, pp. 899–911, Jun. 2012.

[58] E. Felemban, F. K. Shaikh, U. M. Qureshi, A. A. Sheikh, and S. B.Qaisar, “Underwater sensor network applications: A comprehensivesurvey,” Int. J. of Dist. Sens. Netw., vol. 11, no. 11, p. 896832, 2015.

[59] G. Han, C. Zhang, L. Shu, N. Sun, and Q. Li, “A survey on deploymentalgorithms in underwater acoustic sensor networks,” Int. J. of Dist.Sens. Netw., vol. 9, no. 12, p. 314049, 2013.

[60] I. F. Akyildiz, D. Pompili, and T. Melodia, “Underwater acoustic sensornetworks: research challenges,” Ad Hoc Netw., vol. 3, no. 3, pp. 257 –279, Jul. 2005.

[61] D. Pompili, T. Melodia, and I. F. Akyildiz, “Deployment analysis inunderwater acoustic wireless sensor networks,” in Proc. of the 1st ACMInt. Workshop on Underwater Networks. New York, NY, USA: ACM,2006, pp. 48–55.

[62] A. Y. Teymorian, W. Cheng, L. Ma, X. Cheng, X. Lu, and Z. Lu,“3D Underwater Sensor Network Localization,” IEEE Trans. MobileComput., vol. 8, no. 12, pp. 1610–1621, Dec. 2009.

[63] I. Vasilescu, P. Varshavskaya, K. Kotay, and D. Rus, “Autonomousmodular optical underwater robot (AMOUR) design, prototype andfeasibility study,” in Proc. of the IEEE Int. Conf. on Rob. and Auto.,Apr. 2005, pp. 1603–1609.

[64] I. Vasilescu, C. Detweiler, M. Doniec, D. Gurdan, S. Sosnowski,J. Stumpf, and D. Rus, “AMOUR V: A hovering energy efficientunderwater robot capable of dynamic payloads,” The Int. J. of Rob.Res., vol. 29, no. 5, pp. 547–570, Jan. 2010.

Page 33: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

33

[65] M. Dunbabin, P. Corke, I. Vasilescu, and D. Rus, “Experiments withcooperative control of underwater robots,” The Int. J. of Rob. Res.,vol. 28, no. 6, pp. 815–833, May 2009.

[66] M. Dunbabin, J. Roberts, K. Usher, G. Winstanley, and P. Corke, “AHybrid AUV Design for Shallow Water Reef Navigation,” in Proc. ofthe 2005 IEEE Int. Conf. on Robotics and Automation, Apr. 2005, pp.2105–2110.

[67] B. Truax, Acoustic Communication. Westport, CT, USA: Ablex, 2001.[68] E. M. Sozer, M. Stojanovic, and J. G. Proakis, “Underwater acoustic

networks,” IEEE J. Ocean. Eng., vol. 25, no. 1, pp. 72–83, Jan. 2000.[69] D. Pompili and I. F. Akyildiz, “Overview of networking protocols for

underwater wireless communications,” IEEE Commun. Mag., vol. 47,no. 1, pp. 97–102, Jan. 2009.

[70] J. Partan, J. Kurose, and B. N. Levine, “A survey of practical issuesin underwater networks,” in Proc. of the 1st ACM Int. Workshop onUnderwater Networks, 2006, pp. 17–24.

[71] W. W. Au, P. E. Nachtigall, and J. L. Pawloski, “Acoustic effects ofthe ATOC signal (75 Hz, 195 dB) on dolphins and whales,” The J. ofthe Acous. Soc. of Am., vol. 101, no. 5, pp. 2973–2977, 1997.

[72] C. A. Altgelt, “The worlds largest radio station,” https://www.hep.wisc.edu/∼prepost/ELF.pdf, online; accessed 19 Dec. 2017.

[73] M. R. Frater, M. J. Ryan, and R. M. Dunbar, “Electromagneticcommunications within swarms of autonomous underwater vehicles,”in Proc. of the 1st ACM Int. Workshop on Underwater Networks, 2006,pp. 64–70.

[74] P. Da, “Evaluation of wi-fi underwater networks in freshwater,” Ph.D.dissertation, Ph. D. Thesis, Universidade Do Porto, Porto, Portugal,2014.

[75] V. I. Haltrin, “Two-term Henyey-Greenstein light scattering phasefunction for seawater,” in Proc. of IEEE Int. Symp. on Geoscienceand Remote Sensing, vol. 2, Aug. 1999, pp. 1423–1425.

[76] ——, “One-parameter two-term Henyey-Greenstein phase function forlight scattering in seawater,” Appl. Opt., vol. 41, no. 6, pp. 1022–1028,Feb. 2002.

[77] D. Toublanc, “Henyey-Greenstein and Mie phase functions in MonteCarlo radiative transfer computations,” Appl. Opt., vol. 35, no. 18, pp.3270–3274, Jun. 1996.

[78] Y. Dong, S. Tang, and X. Zhang, “Effect of random sea surface ondownlink underwater wireless optical communications,” IEEE Com-mun. Lett., vol. 17, no. 11, pp. 2164–2167, Nov. 2013.

[79] H. Zhang and Y. Dong, “Link misalignment for underwater wirelessoptical communications,” in Adva. in Wireless and Opt. Commun.,(RTUWO), Nov. 2015, pp. 215–218.

[80] X. Yi, Z. Li, and Z. Liu, “Underwater optical communication perfor-mance for laser beam propagation through weak oceanic turbulence,”Appl. Opt., vol. 54, no. 6, pp. 1273–1278, Feb. 2015.

[81] M. Chaplin, “Water absorption spectrum,” http://www1.lsbu.ac.uk/water/water vibrational spectrum.html, online; accessed 10 Dec. 2017.

[82] J. R. V. Zaneveld, “Light and water: Radiative transfer in naturalwaters,” 1995.

[83] R. W. Spinrad, K. L. Carder, and M. J. Perry, Ocean optics. OxfordUniversity Press, 1994, vol. 25.

[84] C. Mobley, Light and Water: Radiative Transfer in Natural Waters.Academic Press, 1994.

[85] C. Mobley, E. Boss, and C. Roesler, “Ocean optics web book,” 2010.[86] N. G. Jerlov, Marine Optics. Elsevier, 1976, vol. 14.[87] R. C. Smith and K. S. Baker, “Optical properties of the clearest natural

waters (200–800 nm),” Appl. Opt., vol. 20, no. 2, pp. 177–184, Jan.1981.

[88] A. Bricaud, A. Morel, and L. Prieur, “Absorption by dissolved organicmatter of the sea (yellow substance) in the UV and visible domains,”Limnology and Oceanography, vol. 26, no. 1, pp. 43–53, Jan. 1981.

[89] W. Breves and R. Reuter, “Bio-optical properties of gelbstoff in thearabian sea at the onset of the southwest monsoon,” J. of Earth Sys.Sci., vol. 109, no. 4, pp. 415–425, Dec. 2000.

[90] V. I. Haltrin, “Chlorophyll-based model of seawater optical properties,”Appl. Opt., vol. 38, no. 33, pp. 6826–6832, Nov. 1999.

[91] D. Hansell and C. Carlson, Biogeochemistry of Marine DissolvedOrganic Matter. Academic Press, 2002.

[92] J. R. Apel, “Edited by,” in Principles of Ocean Physics, ser. Int.Geophysics, J. R. Apel, Ed. Academic Press, 1988, vol. 38, pp. 1– 634.

[93] M. Kishino, N. Okami, M. Takahashi, and S. Ichimura, “Light uti-lization efficiency and quantum yield of phytoplankton in a thermallystratified sea,” Limnology and Oceanography, vol. 31, no. 3, pp. 557–566, Jan. 1986.

[94] M. J. Costello, A. Cheung, and N. De Hauwere, “Surface area and theseabed area, volume, depth, slope, and topographic variation for theworlds seas, oceans, and countries,” Env. Sci. & Tech., vol. 44, no. 23,pp. 8821–8828, 2010.

[95] F. Yang, J. Cheng, and T. A. Tsiftsis, “Free-space optical communi-cation with nonzero boresight pointing errors,” IEEE Trans. Commun.,vol. 62, no. 2, pp. 713–725, Feb. 2014.

[96] R. Sanchez and N. J. McCormick, “Analytic beam spread function forocean optics applications,” Appl. Opt., vol. 41, no. 30, pp. 6276–6288,Oct. 2002.

[97] M. V. Jamali, J. A. Salehi, and F. Akhoundi, “Performance studiesof underwater wireless optical communication systems with spatialdiversity: MIMO scheme,” IEEE Trans. Commun., vol. 65, no. 3, pp.1176–1192, Mar. 2017.

[98] S. Jaruwatanadilok, “Underwater wireless optical communication chan-nel modeling and performance evaluation using vector radiative transfertheory,” IEEE J. Sel. Areas Commun., vol. 26, no. 9, pp. 1620–1627,Dec. 2008.

[99] C. Gabriel, M. A. Khalighi, S. Bourennane, P. Lon, and V. Rigaud,“Misalignment considerations in point-to-point underwater wirelessoptical links,” in OCEANS, Jun. 2013, pp. 1–5.

[100] J. H. Smart, “Underwater optical communications Systems part 1:variability of water optical parameters,” in IEEE Military Commun.Conf., (MILCOM), Oct. 2005, pp. 1140–1146 Vol. 2.

[101] J. W. Giles and I. N. Bankman, “Underwater optical communicationssys. part 2: basic design considerations,” in IEEE Military Commun.Conf., (MILCOM), Oct. 2005, pp. 1700–1705.

[102] A. L. Alldredge and M. W. Silver, “Characteristics, dynamics andsignificance of marine snow,” Progress in Oceanography, vol. 20, no. 1,pp. 41–82, Apr. 1988.

[103] B. M. Cochenour, L. J. Mullen, and A. E. Laux, “Characterization ofthe beam-spread function for underwater wireless optical communica-tions links,” IEEE J. Ocean. Eng., vol. 33, no. 4, pp. 513–521, Oct.2008.

[104] H. Tan, R. Doerffer, T. Oishi, and A. Tanaka, “A new approach tomeasure the volume scattering function,” Opt. Express, vol. 21, no. 16,pp. 18 697–18 711, Aug. 2013.

[105] H. V. D. Hulst, Multiple Light Scattering. Academic Press, 1980.[106] C. Li, K. H. Park, and M. S. Alouini, “On the use of a direct radiative

transfer equation solver for path loss calculation in underwater opticalwireless channels,” IEEE Wireless Commun. Lett., vol. 4, no. 5, pp.561–564, Oct. 2015.

[107] S. Arnon, J. Barry, G. Karagiannidis, R. Schober, and M. Uysal,Advanced Optical Wireless Communication Sys., 1st ed. New York,NY, USA: Cambridge University Press, 2012.

[108] B. Cochenour, L. Mullen, and A. Laux, “Spatial and temporal dis-persion in high bandwidth underwater laser communication links,” inIEEE Military Commun. Conf., (MILCOM), Nov. 2008, pp. 1–7.

[109] J. R. Potter, M. B. Porter, and J. C. Preisig, “UComms: A Conf.and Workshop on Underwater Commun., Channel Modeling, andValidation,” IEEE J. Ocean. Eng., vol. 38, no. 4, pp. 603–613, Oct.2013.

[110] L. Wang, S. L. Jacques, and L. Zheng, “MCMLMonte Carlo modelingof light transport in multi-layered tissues,” Computer Methods andPrograms in Biomedicine, vol. 47, no. 2, pp. 131 – 146, Feb. 1995.

[111] R. M. Lerner and J. D. Summers, “Monte carlo description of time-and space-resolved multiple forward scatter in natural water,” Appl.Opt., vol. 21, no. 5, pp. 861–869, Mar. 1982.

[112] R. A. Leathers, T. V. Downes, C. O. Davis, and C. D. Mobley, “Montecarlo radiative transfer simulations for ocean optics: a practical guide,”Naval Research Lab Washington DC Applied Optics Branch, Tech.Rep., 2004.

[113] C. Gabriel, M. A. Khalighi, S. Bourennane, P. Leon, and V. Rigaud,“Channel modeling for underwater optical communication,” in IEEEGLOBECOM Workshops (GC Wkshps), Dec. 2011, pp. 833–837.

[114] J. Li, Y. Ma, Q. Zhou, B. Zhou, and H. Wang, “Monte carlo studyon pulse response of underwater optical channel,” Opt. Engineering,vol. 51, pp. 51 – 51 – 6, 2012.

[115] ——, “Channel capacity study of underwater wireless optical commu-nications links based on monte carlo simulation,” J. of Opt., vol. 14,no. 1, pp. 1–7, 2012.

[116] F. Hanson and M. Lasher, “Effects of underwater turbulence on laserbeam propagation and coupling into single-mode optical fiber,” Appl.Opt., vol. 49, no. 16, pp. 3224–3230, Jun. 2010.

[117] W. Liu, Z. Xu, and L. Yang, “SIMO detection schemes for underwateroptical wireless communication under turbulence,” Photon. Res., vol. 3,no. 3, pp. 48–53, Jun. 2015.

Page 34: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

34

[118] W. W. Hou, “A simple underwater imaging model,” Opt. Lett., vol. 34,no. 17, pp. 2688–2690, Sep. 2009.

[119] S. Woods, W. Hou, W. Goode, E. Jarosz, and A. Weidemann, “Mea-surements of turbulence for quantifying the impact of turbulence onunderwater imaging,” in 2011 IEEE/OES 10th Current, Waves andTurbulence Measurements (CWTM), Mar. 2011, pp. 179–183.

[120] M. Holohan and J. Dainty, “Low-order adaptive optics: a possible usein underwater imaging?” Opt. & Laser Techno., vol. 29, no. 1, pp. 51– 55, Nov. 1997.

[121] V. V. Nikishov and V. I. Nikishov, “Spectrum of turbulent fluctuationsof the sea-water refraction index,” Int. J. of Fluid Mech. Res., vol. 27,no. 1, pp. 82–98, 2000.

[122] O. Korotkova and N. Farwell, “Effect of oceanic turbulence on po-larization of stochastic beams,” Opt. Commun., vol. 284, no. 7, pp.1740–1746, Aug. 2011.

[123] W. C. Cox Jr, Simulation, modeling, and design of underwater opticalcommunication Sys. North Carolina State University, 2012.

[124] Y. Ata and Y. Baykal, “Field correlation of spherical wave in under-water turbulent medium,” Appl. Opt., vol. 53, no. 33, pp. 7968–7971,Nov. 2014.

[125] S. Tang, X. Zhang, and Y. Dong, “Temporal statistics of irradiance inmoving turbulent ocean,” in OCEANS, Jun. 2013, pp. 1–4.

[126] Y. Ata and Y. Baykal, “Scintillations of optical plane and sphericalwaves in underwater turbulence,” J. Opt. Soc. Am. A, vol. 31, no. 7,pp. 1552–1556, Jul. 2014.

[127] S. Tang, Y. Dong, and X. Zhang, “On link misalignment for underwaterwireless optical communications,” IEEE Commun. Lett., vol. 16, no. 10,pp. 1688–1690, Oct. 2012.

[128] M. A. Khalighi and M. Uysal, “Survey on Free Space Optical Com-munication: A Communication Theory Perspective,” IEEE Commun.Surveys Tuts., vol. 16, no. 4, pp. 2231–2258, Jun. 2014.

[129] F. Akhoundi, J. A. Salehi, and A. Tashakori, “Cellular UnderwaterWireless Optical CDMA Network: Performance Analysis and Imple-mentation Concepts,” IEEE Trans. Commun., vol. 63, no. 3, pp. 882–891, Mar. 2015.

[130] Z. Ghassemlooy, W. Popoola, and S. Rajbhandari, Optical WirelessCommunications: System and Channel Modelling with MATLAB(R),Second Edition. Taylor & Francis, 2018.

[131] D.-S. Shiu and J. M. Kahn, “Differential pulse-position modulationfor power-efficient optical communication,” IEEE Trans. Commun.,vol. 47, no. 8, pp. 1201–1210, Aug. 1999.

[132] Z. Ghassemlooy, A. R. Hayes, N. L. Seed, and E. D. Kaluarachchi,“Digital pulse interval modulation for optical communications,” IEEECommun. Mag., vol. 36, no. 12, pp. 95–99, Dec. 1998.

[133] Z. Ghassemlooy and N. M. Aldibbiat, “Multilevel digital pulse intervalmodulation scheme for optical wireless communications,” in Int. Conf.on Transparent Opt. Netw., vol. 3, Jun. 2006, pp. 149–153.

[134] H. Sari and B. Woodward, “Underwater voice communications using amodulated laser beam,” in OCEANS, vol. 2, Sep. 1998, pp. 1183–1188vol.2.

[135] M. Chen, S. Zhou, and T. Li, “The Implementation of PPM inUnderwater Laser Communication System,” in Int. Conf. on Commun.,Circ. and Sys., vol. 3, Jun. 2006, pp. 1901–1903.

[136] S. Meihong, Y. Xinsheng, and Z. Fengli, “The evaluation of modulationtechniques for underwater wireless optical communications,” in Int.Conf. on Commun. Software and Netw., Feb. 2009, pp. 138–142.

[137] D. Anguita, D. Brizzolara, and G. Parodi, “VHDL modeling of PHYand MAC Layer modules for underwater optical wireless communica-tion,” in 5th European Conf. on Circ. and Sys. for Commun., (ECCSC),Nov. 2010, pp. 185–188.

[138] ——, “Optical wireless communication for underwater wireless sensornetworks: Hardware modules and circuits design and implementation,”in OCEANS, Sep. 2010, pp. 1–8.

[139] R. M. Hagem, D. V. Thiel, S. G. O’Keefe, and T. Fickenscher, “Opticalwireless communication for real time swimmers feedback: A review,”in 2012 Int. Symp. on Commun. and Info. Techno., (ISCIT), Oct. 2012,pp. 1080–1085.

[140] X. He and J. Yan, “Study on performance of M-ary PPM underwateroptical communication Systems using vector radiative transfer theory,”in ISAPE2012, Oct. 2012, pp. 566–570.

[141] P. Swathi and S. Prince, “Designing issues in design of underwaterwireless optical communication system,” in Int. Conf. on Commun.and Signal Process., Apr. 2014, pp. 1440–1445.

[142] R. J. G. Yangyu Fan, “Comparison of pulse position modulation andpulse width modulation for application in optical communications,”Opt. Engineering, vol. 6, pp. 46–53, Jun. 2007.

[143] C. Gabriel, M. A. Khalighi, S. Bourennane, P. Lon, and V. Rigaud,“Investigation of suitable modulation techniques for underwater wire-less optical communication,” in Int. Works. on Opt. Wireless Commun.,(IWOW), Oct. 2012, pp. 1–3.

[144] M. Doniec, I. Vasilescu, M. Chitre, C. Detweiler, M. Hoffmann-Kuhnt,and D. Rus, “Aquaoptical: A lightweight device for high-rate long-range underwater point-to-point communication,” in OCEANS, Oct.2009, pp. 1–6.

[145] M. Doniec and D. Rus, “BiDirectional optical communication withAquaOptical II,” in IEEE Int. Conf. on Commun. Sys., Nov. 2010, pp.390–394.

[146] M. Doniec, C. Detweiler, I. Vasilescu, and D. Rus, “Using opticalcommunication for remote underwater robot operation,” in IEEE/RSJInt. Conf. on Intelligent Robots and Sys., Oct. 2010, pp. 4017–4022.

[147] B. Cochenour, L. Mullen, and A. Laux, “Phase coherent digital commu-nications for wireless optical links in turbid underwater environments,”in OCEANS, Sep. 2007, pp. 1–5.

[148] W. C. Cox, B. L. Hughes, and J. F. Muth, “A polarization shift-keying system for underwater optical communications,” in OCEANS,Oct. 2009, pp. 1–4.

[149] Y. Dong, T. Zhang, and X. Zhang, “Polarized pulse position modulationfor wireless optical communications,” in 47th Annual Conf. on Info.Sci. and Sys., (CISS), Mar. 2013, pp. 1–5.

[150] R. M. Gagliardi and S. Karp, “Opt. commun.” New York, Wiley-Interscience, 1976. 445 p., 1976.

[151] B. E. Saleh, M. C. Teich, and B. E. Saleh, Fundamentals of photonics.Wiley New York, 1991, vol. 22.

[152] F. Xu, M. A. Khalighi, and S. Bourennane, “Impact of different noisesources on the performance of PIN- and APD-based FSO receivers,” inProc. of the 11th Int. Conf. on Telecommun., Jun. 2011, pp. 211–218.

[153] F. M. Davidson and X. Sun, “Gaussian approximation versus nearlyexact performance analysis of optical communication systems withppm signaling and apd receivers,” IEEE Trans. Commun., vol. 36,no. 11, pp. 1185–1192, Nov. 1988.

[154] M. Cole and T. Driscoll, “The Lighting Revolution: If We Were ExpertsBefore, We’re Novices Now,” IEEE Trans. Ind. Appl., vol. 50, no. 2,pp. 1509–1520, Mar. 2014.

[155] L. Grobe, A. Paraskevopoulos, J. Hilt, D. Schulz, F. Lassak, F. Hartlieb,C. Kottke, V. Jungnickel, and K. D. Langer, “High-speed visible lightcommunication systems,” IEEE Commun. Mag., vol. 51, no. 12, pp.60–66, Dec. 2013.

[156] H. Haas, L. Yin, Y. Wang, and C. Chen, “What is LiFi?” J. LightwaveTechnol., vol. 34, no. 6, pp. 1533–1544, Mar. 2016.

[157] H. Marshoud, P. C. Sofotasios, S. Muhaidat, and G. K. Karagiannidis,“Multi-user techniques in visible light communications: A survey,” inInt. Conf. on Adv. Commun. Sys. and Info. Sec., (ACOSIS), Oct. 2016,pp. 1–6.

[158] M. Jaber, M. A. Imran, R. Tafazolli, and A. Tukmanov, “5G Back-haul Challenges and Emerging Research Directions: A Survey,” IEEEAccess, vol. 4, pp. 1743–1766, Apr. 2016.

[159] T. Ho, S. Trisno, I. I. Smolyaninov, S. D. Milner, and C. C. Davis,“Studies of pointing, acquisition, and tracking of agile optical wirelesstransceivers for free-space optical communication networks,” in Proc.SPIE, Feb. 2004, pp. 5237–5237.

[160] A. S. Fletcher, S. A. Hamilton, and J. D. Moores, “Undersea lasercommunication with narrow beams,” IEEE Commun. Mag., vol. 53,no. 11, pp. 49–55, Nov. 2015.

[161] J. B. Snow, J. P. Flatley, D. E. Freeman, M. A. Landry, C. E. Lindstrom,J. R. Longacre, and J. A. Schwartz, “Underwater propagation of high-data-rate laser communications pulses,” in Proc. SPIE, vol. 1750, Dec.1992, pp. 1750 – 1750 – 9.

[162] C. Pontbriand, N. Farr, J. Ware, J. Preisig, and H. Popenoe, “Diffusehigh-bandwidth optical communications,” in OCEANS, Sep. 2008, pp.1–4.

[163] S. Arnon and D. Kedar, “Non-line-of-sight underwater optical wirelesscommunication network,” J. Opt. Soc. Am. A, vol. 26, no. 3, pp. 530–539, Mar. 2009.

[164] J. C. Juarez, A. Dwivedi, A. R. Hammons, S. D. Jones, V. Weerackody,and R. A. Nichols, “Free space optical communications for next-generation military networks,” IEEE Commun. Mag., vol. 44, no. 11,pp. 46–51, Nov. 2006.

[165] D. Liu, X. Zhang, and S. Zhang, “LED Lens for Rectangular Beamwith Small Divergence Angles,” J. Opt. Soc. Korea, vol. 20, no. 6, pp.739–744, Dec 2016.

[166] J. M. Kahn and J. R. Barry, “Wireless infrared communications,” Proc.of the IEEE, vol. 85, no. 2, pp. 265–298, Feb. 1997.

Page 35: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

35

[167] X. Ning, R. Winston, and J. O’Gallagher, “Dielectric totally internallyreflecting concentrators,” Appl. Opt., vol. 26, no. 2, pp. 300–305, Jan.1987.

[168] J. P. Savicki and S. P. Morgan, “Hemispherical concentrators andspectral filters for planar sensors in diffuse radiation fields,” Appl. Opt.,vol. 33, no. 34, pp. 8057–8061, Dec. 1994.

[169] G. Baiden, Y. Bissiri, and A. Masoti, “Paving the way for a futureunderwater omni-directional wireless optical communication systems,”Ocean Engineering, vol. 36, no. 9, pp. 633–640, Mar. 2009.

[170] F. Jasman and R. J. Green, “Monte Carlo simulation for underwateroptical wireless communications,” in 2nd Int. Workshop on Opt.Wireless Commun., (IWOW), Oct. 2013, pp. 113–117.

[171] D. Kedar and S. Arnon, “Non-line-of-sight optical wireless sensornetwork operating in multiscattering channel,” Appl. Opt., vol. 45,no. 33, pp. 8454–8461, Nov. 2006.

[172] D. Kedar, “Multiaccess interference in a non-line-of-sight ultravioletoptical wireless sensor network,” Appl. opt., vol. 46, no. 23, pp. 5895–5901, Aug. 2007.

[173] L. Mullen, B. Cochenour, W. Rabinovich, R. Mahon, and J. Muth,“Backscatter suppression for underwater modulating retroreflector linksusing polarization discrimination,” Appl. Opt., vol. 48, no. 2, pp. 328–337, Jan. 2009.

[174] B. Cochenour, L. Mullen, W. Rabinovich, and R. Mahon, “Underwateroptical communications with a modulating retro-reflector,” in Proc.SPIE, Apr. 2009, pp. 1–11.

[175] A. Celik, N. Saeed, T. Y. Al-Naffouri, and M.-S. Alouini, “Modelingand performance analysis of multihop underwater optical wireless sen-sor networks,” in IEEE Wireless Commun. and Netw. Conf., (WCNC),Apr. 2018, pp. 1–6.

[176] P. Eraerds, M. Legre, A. Rochas, H. Zbinden, and N. Gisin, “SiPMfor fast photon-counting and multiphoton detection,” Optics Express,vol. 15, no. 22, pp. 14 539–14 549, Oct. 2007.

[177] ITUT, “Forward error correction for high bit rate dwdm submarinesystems,” International Telecommunication Union (ITU-T), Tech. Rep.,2004.

[178] G. Baiden and Y. Bissiri, “High bandwidth optical networking forunderwater untethered telerobotic operation,” in OCEANS, Sep. 2007,pp. 1–9.

[179] M. Bilgi and M. Yuksel, “Multi-element Free-Space-Optical sphericalstructures with intermittent connectivity patterns,” in Proc. IEEE Int.Conf. on Computer Commun. (INFOCOM), Apr. 2008, pp. 1–4.

[180] N. Farr, A. Chave, L. Freitag, J. Preisig, S. White, D. Yoerger, andP. Titterton, “Optical modem technology for seafloor observatories,” inOCEANS, Sep. 2005, pp. 928–934.

[181] D. E. Ingber, “The architecture of life,” Scientific American, vol. 278,no. 1, pp. 48–57, 1998.

[182] M. V. Jamali, F. Akhoundi, and J. A. Salehi, “Performance character-ization of relay-assisted wireless optical CDMA networks in turbulentunderwater channel,” IEEE Trans. Wireless Commun., vol. 15, pp.4104–4116, Jun. 2016.

[183] F. Akhoundi, M. V. Jamali, N. B. Hassan, H. Beyranvand, A. Minoofar,and J. A. Salehi, “Cellular Underwater Wireless Optical CDMA Net-work: Potentials and Challenges,” IEEE Access, vol. 4, pp. 4254–4268,Jul. 2016.

[184] U. Nadeem, N. U. Hassan, M. A. Pasha, and C. Yuen, “Indoorpositioning system designs using visible LED lights: performancecomparison of TDM and FDM protocols,” Electron. Lett., vol. 51, no. 1,pp. 72–74, Jan. 2015.

[185] Y. Hou, S. Xiao, H. Zheng, and W. Hu, “Multiple access scheme basedon block encoding time division multiplexing in an indoor positioningsystem using visible light,” IEEE J. Opt. Commun. Netw., vol. 7, no. 5,pp. 489–495, May 2015.

[186] Y. Wang, N. Chi, Y. Wang, L. Tao, and J. Shi, “High speed LED basedvisible light communication networks for beyond 10Gb/s wirelessaccess,” in Sixth Int. Conf. on Wireless Commun. and Signal Process.,(WCSP), Oct. 2014, pp. 1–6.

[187] J. Armstrong, R. J. Green, and M. D. Higgins, “Comparison of ThreeReceiver Designs for Optical Wireless Communications using WhiteLEDs,” IEEE Commun. Lett., vol. 16, no. 5, pp. 748–751, May 2012.

[188] H. Elgala, R. Mesleh, and H. Haas, “Indoor optical wireless commu-nication: potential and state-of-the-art,” IEEE Commun. Mag., vol. 49,no. 9, pp. 56–62, Sep. 2011.

[189] M. F. Sanya, L. Djogbe, A. Vianou, and C. Aupetit-Berthelemot, “DC-biased optical OFDM for IM/DD passive optical network systems,”IEEE J. Opt. Commun. Netw., vol. 7, no. 4, pp. 205–214, Apr. 2015.

[190] J. Tan, Z. Wang, Q. Wang, and L. Dai, “Near-Optimal Low-ComplexitySequence Detection for Clipped DCO-OFDM,” IEEE Photon. Technol.Lett., vol. 28, no. 3, pp. 233–236, Feb. 2016.

[191] H. Zhang, Y. Yuan, and W. Xu, “PAPR Reduction for DCO-OFDMVisible Light Communications via Semidefinite Relaxation,” IEEEPhoton. Technol. Lett., vol. 26, no. 17, pp. 1718–1721, Sep. 2014.

[192] Z. Yu, R. J. Baxley, and G. T. Zhou, “Iterative Clipping for PAPRReduction in Visible Light OFDM Communications,” in IEEE MilitaryCommun. Conf., (MILCOM), Oct. 2014, pp. 1681–1686.

[193] X. Li, J. Vucic, V. Jungnickel, and J. Armstrong, “On the Capacityof Intensity-Modulated Direct-Detection Systems and the InformationRate of ACO-OFDM for Indoor Optical Wireless Applications,” IEEETrans. Commun., vol. 60, no. 3, pp. 799–809, Mar. 2012.

[194] J. Dang and Z. Zhang, “Comparison of optical OFDM-IDMA andoptical ofdma for uplink visible light communications,” in Int. Conf. onWireless Commun. and Signal Process., (WCSP), Oct. 2012, pp. 1–6.

[195] E. Dinc, O. Ergul, and O. B. Akan, “Soft handover in OFDMA basedvisible light communication networks,” in IEEE 82nd Vehicular Techno.Conf., (VTC), Sep. 2015, pp. 1–5.

[196] B. Lin, X. Tang, H. Yang, Z. Ghassemlooy, S. Zhang, Y. Li, andC. Lin, “Experimental demonstration of IFDMA for uplink visible lightcommunication,” IEEE Photon. Techno. Lett., vol. 28, no. 20, pp. 2218–2220, Oct. 2016.

[197] M. Jazayerifar and J. A. Salehi, “Atmospheric optical CDMA commu-nication systems via optical orthogonal codes,” IEEE Trans. Commun.,vol. 54, no. 9, pp. 1614–1623, Sep. 2006.

[198] B. M. Ghaffari, M. D. Matinfar, and J. A. Salehi, “Wireless opticalCDMA LAN: digital design concepts,” IEEE Trans. Commun., vol. 56,no. 12, pp. 2145–2155, Dec. 2008.

[199] K. Fouli and M. Maier, “OCDMA and optical coding: Principles,applications, and challenges,” IEEE Commun. Mag., vol. 45, no. 8,pp. 27–34, Aug. 2007.

[200] A. Stok and E. H. Sargent, “The role of optical CDMA in accessnetworks,” IEEE Commun. Mag., vol. 40, no. 9, pp. 83–87, Sep. 2002.

[201] H. Marshoud, V. M. Kapinas, G. K. Karagiannidis, and S. Muhaidat,“Non-orthogonal multiple access for visible light communications,”IEEE Photon. Techno. Lett., vol. 28, no. 1, pp. 51–54, Jan. 2016.

[202] H. Marshoud, P. C. Sofotasios, S. Muhaidat, G. K. Karagiannidis,and B. S. Sharif, “On the performance of visible light communicationsystems with non-orthogonal multiple access,” IEEE Trans. Commun.,vol. 16, no. 10, pp. 6350–6364, Oct. 2017.

[203] D. Tsonev, S. Videv, and H. Haas, “Light fidelity (li-fi): towards all-optical networking,” Proc. SPIE, vol. 9007, no. 10, pp. 1–10, Feb.2014.

[204] Z. Chen and H. Haas, “Space division multiple access in visible lightcommunications,” in IEEE Int. Conf. on Commun., (ICC), Jun. 2015,pp. 5115–5119.

[205] T.-H. Ho, S. Trisno, I. I. Smolyaninov, S. D. Milner, and C. C. Davis,“Studies of pointing, acquisition, and tracking of agile optical wirelesstransceivers for free-space optical communication networks,” Proc.SPIE, vol. 5237, no. 12, pp. 1 – 13, 2004.

[206] V. Nikulin, R. Khandekar, and J. Sofka, “Performance of a lasercommunication system with acousto-optic tracking: an experimentalstudy,” Proc. SPIE, vol. 6105, no. 10, pp. 1–10, Mar. 2006.

[207] P. LoPresti, H. Refai, J. Sluss, and I. Varela-Cuadrado, “Adaptivedivergence and power for improving connectivity in free-space opticalmobile networks,” Appl. Opt., vol. 45, no. 25, pp. 6591–6597, Sep.2006.

[208] M. Uysal and M. Uysal, Cooperative Communications for ImprovedWireless Network Transmission: Framework for Virtual Antenna ArrayApplications. Hershey, PA: Information Science Reference - Imprintof: IGI Publishing, 2009.

[209] N. D. Chatzidiamantis, D. S. Michalopoulos, E. E. Kriezis, G. K.Karagiannidis, and R. Schober, “Relay selection protocols for relay-assisted free-space optical systems,” J. Opt. Commun. Netw., vol. 5,no. 1, pp. 92–103, Jan. 2013.

[210] S. Kazemlou, S. Hranilovic, and S. Kumar, “All-optical multihopfree-space optical communication systems,” J. of Lightwave Techno.,vol. 29, no. 18, pp. 2663–2669, Sep. 2011.

[211] M. Karimi and M. Nasiri-Kenari, “BER analysis of cooperative systemsin free-space optical networks,” J. of Lightwave Techno., vol. 27, no. 24,pp. 5639–5647, Dec. 2009.

[212] J. M. Jornet, M. Stojanovic, and M. Zorzi, “Focused beam routingprotocol for underwater acoustic networks,” in Proc. of the Third ACMInt. Workshop on Underwater Netw., 2008, pp. 75–82.

Page 36: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

36

[213] M. Ayaz, I. Baig, A. Abdullah, and I. Faye, “A survey on routingtechniques in underwater wireless sensor networks,” J. of Netw. andComputer App., vol. 34, no. 6, pp. 1908 – 1927, 2011.

[214] N. Li, J.-F. Martinez, J. M. M. Chaus, and M. Eckert, “A Surveyon Underwater Acoustic Sensor Network Routing Protocols,” Sensors,vol. 16, no. 3, pp. 1–28, Mar. 2016.

[215] M. Ahmed, M. Salleh, and M. I. Channa, “Routing protocols forunderwater wireless sensor networks based on data forwarding: areview,” Telecommun. Systems, vol. 65, no. 1, pp. 139–153, May 2017.

[216] Q. Lu, F. Liu, Y. Zhang, and S. Jiang, “Routing protocols forunderwater acoustic sensor networks: A survey from an applicationperspective,” in Advances in Underwater Acoustics. Rijeka: InTech,2017.

[217] S. Sahana, K. Singh, R. Kumar, and S. Das, “A review of underwaterwireless sensor network routing protocols and challenges,” in Next-Generation Networks, Nov. 2018, pp. 505–512.

[218] S. Hirai, Y. Tanigawa, and H. Tode, “Integration method betweenlocalization and routing in underwater sensor network,” in IEEE 15thInt. Conf. on Computational Sci. and Eng., Dec 2012, pp. 689–693.

[219] Y. Tanigawa, S. Hirai, and H. Tode, “Lightweight data transfer unifiedwith active localization and robust routing in underwater networks,” inIEEE Int. Conf. on Commun., (ICC), Jun. 2015, pp. 2534–2539.

[220] P. Wang, D. h. Fu, C. q. Zhao, J. c. Xing, Q. l. Yang, and X. f. Du, “Areliable and efficient routing protocol for underwater acoustic sensornetworks,” in IEEE Int. Conf. on Cyber Techno. in Auto., Cont. andIntelligent Sys., May 2013, pp. 185–190.

[221] D. Hwang and D. Kim, “DFR: Directional flooding-based routingprotocol for underwater sensor networks,” in OCEANS, Sep. 2008, pp.1–7.

[222] S. H. Ahmed, S. Lee, J. Park, D. Kim, and D. B. Rawat, “iDFR:Intelligent directional flooding-based routing protocols for underwatersensor networks,” in 14th IEEE Annual Consumer Commun. Netw.Conf., (CCNC), Jan. 2017, pp. 560–565.

[223] L. Emokpae and M. Younis, “Signal reflection-enabled geographicalrouting for underwater sensor networks,” in IEEE Int. Conf. on Com-mun., (ICC), Jun. 2012, pp. 147–151.

[224] N. Chirdchoo, W. S. Soh, and K. C. Chua, “Sector-based routing withdestination location prediction for underwater mobile networks,” in Int.Conf. on Adv. Info. Netw. and App. Workshops, May 2009, pp. 1148–1153.

[225] S. Zhang and D. Li, “A beam width and direction concerned routingfor underwater acoustic sensor networks,” in IEEE 9th Int. Conf. onMobile Ad-hoc and Sensor Netw., Dec. 2013, pp. 17–24.

[226] M. Khalid, Z. Ullah, N. Ahmad, H. Khan, H. S. Cruickshank, andO. U. Khan, “A comparative simulation based analysis of locationbased routing protocols in underwater wireless sensor networks,” in2nd Workshop on Recent Trends in Telecommun. Res., (RTTR), Feb.2017, pp. 1–5.

[227] Y. Wei and D. S. Kim, “Reliable and energy-efficient routing protocolfor underwater acoustic sensor networks,” in Int. Conf. on Info. andCommun. Techno. Conv., (ICTC), Oct. 2014, pp. 738–743.

[228] E. Kim, J. Kang, P. K. Chong, S. Yoo, and D. Kim, “Energy efficientlocal area source routing protocol of underwater sensor networks in thedeep ocean,” in Int. Symp. on Commun. and Info. Techno., Oct. 2007,pp. 948–953.

[229] Y. D. Chen, Y. W. Chen, C. Y. Lien, and K. P. Shih, “A channel-aware depth-adaptive routing protocol for underwater acoustic sensornetworks,” in OCEANS, Apr. 2014, pp. 1–6.

[230] X. Chen and G. Lin, “Energy efficient routing protocol for underwateracoustic networks using shortest path algorithm,” in 4th Int. Conf. onInfo. Sci. and Cont. Eng., (ICISCE), Jul. 2017, pp. 295–300.

[231] Y. D. Chen, C. Y. Lien, C. H. Wang, and K. P. Shih, “DARP: A depthadaptive routing protocol for large-scale underwater acoustic sensornetworks,” in OCEANS, May 2012, pp. 1–6.

[232] R. W. L. Coutinho, L. F. M. Vieira, and A. A. F. Loureiro, “DCR:Depth-Controlled Routing protocol for underwater sensor networks,”in IEEE Symp. on Computers and Commun., (ISCC), Jul. 2013, pp.453–458.

[233] L. C. Kuo and T. Melodia, “Cross-layer routing on MIMO-OFDMunderwater acoustic links,” in Annual IEEE Commun. Soc. Conf. onSensor, Mesh and Ad Hoc Commun. and Netw., (SECON), Jun. 2012,pp. 227–235.

[234] X. Wu, G. Chen, and J. Chen, “Energy-Efficient and Topology-AwareRouting for Underwater Sensor Networks,” in Proc. of 19th Int. Conf.on Computer Commun. and Netw., Aug. 2010, pp. 1–6.

[235] Z. Guo, G. Colombi, B. Wang, J. H. Cui, D. Maggiorini, and G. P.Rossi, “Adaptive Routing in Underwater Delay/Disruption Tolerant

Sensor Networks,” in Fifth Annual Conf. on Wireless on Demand Netw.Sys. and Serv., Jan. 2008, pp. 31–39.

[236] D. Pompili, T. Melodia, and I. F. Akyildiz, “Routing algorithms fordelay-insensitive and delay-sensitive applications in underwater sensornetworks,” in Proc. of the 12th Annual Int. Conf. on Mobile Computingand Netw., 2006, pp. 298–309.

[237] M. Goetz and I. Nissen, “GUWMANET Multicast routing in Under-water Acoustic Networks,” in Military Commun. and Info. Sys. Conf.,(MCC), Oct. 2012, pp. 1–8.

[238] Z. Zhou, Z. Peng, J. H. Cui, and Z. Shi, “Efficient multipath com-munication for time-critical applications in underwater acoustic sensornetworks,” IEEE/ACM Trans. Netw., vol. 19, no. 1, pp. 28–41, Feb.2011.

[239] J. Xu, K. Li, and G. Min, “Reliable and energy-efficient multipathcommunications in underwater sensor networks,” IEEE Trans. ParallelDistrib. Syst., vol. 23, no. 7, pp. 1326–1335, Jul. 2012.

[240] S. Basagni, C. Petrioli, R. Petroccia, and D. Spaccini, “Channel-awarerouting for underwater wireless networks,” in OCEANS, May 2012, pp.1–9.

[241] K. R. Anupama, A. Sasidharan, and S. Vadlamani, “A location-basedclustering algorithm for data gathering in 3D underwater wirelesssensor networks,” in Int. Symp. on Telecommun., Aug. 2008, pp. 343–348.

[242] M. C. Domingo and R. Prior, “Design and analysis of a gps-free routingprotocol for underwater wireless sensor networks in deep water,” in Int.Conf. on Sensor Techno. and App., (SENSORCOMM), Oct. 2007, pp.215–220.

[243] C. Wang and G. Liu, “LUM-HEED: A Location Unaware, Multi-hoprouting protocol for Underwater Acoustic Sensor Networks,” in Proc.of Int. Conf. on Computer Sci. and Netw. Techno., vol. 4, Dec. 2011,pp. 2336–2340.

[244] D. N. Sandeep and V. Kumar, “Review on Clustering, Coverage andConnectivity in Underwater Wireless Sensor Networks: A Communica-tion Techniques Perspective,” IEEE Access, vol. 5, pp. 11 176–11 199,Jun. 2017.

[245] T. Hu and Y. Fei, “An adaptive routing protocol based on connectivityprediction for underwater disruption tolerant networks,” in IEEE GlobalCommun. Conf., (GLOBECOM), Dec. 2013, pp. 65–71.

[246] ——, “QELAR: A Machine-Learning-Based Adaptive Routing Pro-tocol for Energy-Efficient and Lifetime-Extended Underwater SensorNetworks,” IEEE Trans. Mobile Comput., vol. 9, no. 6, pp. 796–809,Jun. 2010.

[247] ——, “MURAO: A multi-level routing protocol for acoustic-opticalhybrid underwater wireless sensor networks,” in 9th Annual IEEECommun. Soc. Conf. on Sensor, Mesh and Ad Hoc Commun. and Netw.,(SECON), Jun. 2012, pp. 218–226.

[248] J. Dall and M. Christensen, “Random geometric graphs,” Phys. Rev. E,vol. 66, pp. 16–21, Jul. 2002.

[249] P. Erdos and A. Renyi, “On the evolution of random graphs,” inPublication of the mathematical institute of the Hungarian academyof sciences, Jul. 1960, pp. 17–61.

[250] C. Bettstetter, “On the connectivity of ad hoc networks,” The ComputerJ., vol. 47, no. 4, pp. 432–447, Jul. 2004.

[251] Z. Liao, D. Li, and J. Chen, “A network access mechanism for multihopunderwater acoustic local area networks,” IEEE Sensors J., vol. 16,no. 10, pp. 3914–3926, May 2016.

[252] H. AlQuwaiee, I. S. Ansari, and M. S. Alouini, “On the performanceof free-space optical communication systems over double generalizedgamma channel,” IEEE J. Sel. Areas Commun., vol. 33, no. 9, pp.1829–1840, Sep. 2015.

[253] P. Gupta and P. R. Kumar, “Critical power for asymptotic connectivity,”in Proc. of the 37th IEEE Conf. on Decision and Control, vol. 1, Dec.1998, pp. 1106–1110.

[254] C. Y. Wu and C. W. Yu, “Computing the probability of specificrandom sector graphs with applications in wireless networks withdirectional antennas,” in Tenth Int. Conf. on Intelligent Info. Hidingand Multimedia Signal Process., Aug. 2014, pp. 678–681.

[255] A. Vavoulas, H. G. Sandalidis, and D. Varoutas, “Underwater opticalwireless networks: A k-connectivity analysis,” IEEE J. Ocean. Eng.,vol. 39, no. 4, pp. 801–809, Oct. 2014.

[256] N. Saeed, A. Celik, T. Y. Al-Naffouri, and M.-S. Alouini, “Connectivityanalysis of underwater optical wireless sensor networks: A graphtheoretic approach,” in submitted to IEEE Int. Conf. on Commun.,(ICC), May 2018.

[257] P. Gupta and P. R. Kumar, Critical Power for Asymptotic Connectivityin Wireless Networks. Boston, MA: Birkhauser Boston, 1999, pp.547–566.

Page 37: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

37

[258] ——, “The capacity of wireless networks,” IEEE Trans. Inf. Theory,vol. 46, no. 2, pp. 388–404, Mar. 2000.

[259] A. Ozgur, O. Leveque, and D. Tse, “Hierarchical cooperation achieveslinear capacity scaling in ad hoc networks,” in Proc. IEEE Int. Conf.on Computer Commun. (INFOCOM), May 2007, pp. 382–390.

[260] F. A. Onat, I. Stojmenovic, and H. Yanikomeroglu, “Generating randomgraphs for the simulation of wireless ad hoc, actuator, sensor, andinternet networks,” Perv. and Mobile Comp., vol. 4, no. 5, pp. 597–615,May 2008.

[261] H. M. Ammari and S. K. Das, “Critical density for coverage andconnectivity in three-dimensional wireless sensor networks using con-tinuum percolation,” IEEE Trans. Parallel Distrib. Syst., vol. 20, no. 6,pp. 872–885, Jun. 2009.

[262] H. M. Ammari, Challenges and Opportunities of Connected k-CoveredWireless Sensor Networks. Berlin Heidelberg: Springer-Verlag, 2009.

[263] J. P. Coon, O. Georgiou, and C. P. Dettmann, “Connectivity scalinglaws in wireless networks,” IEEE Wireless Commun. Lett., vol. 4, no. 6,pp. 629–632, Dec. 2015.

[264] Y. Liu, J. Gao, J. Yu, and C. Yin, “Local connectivity for heterogeneousoverlaid wireless networks,” Ad Hoc Netw., vol. 58, no. Supplement C,pp. 205 – 212, Jun. 2017.

[265] V. Ravelomanana, “Extremal properties of three-dimensional sensornetworks with applications,” IEEE Trans. Mobile Comput., vol. 3, no. 3,pp. 246–257, Jul. 2004.

[266] K. Akkaya and A. Newell, “Self-deployment of sensors for maximizedcoverage in underwater acoustic sensor networks,” Computer Commun.,vol. 32, no. 7-10, pp. 1233–1244, May 2009.

[267] F. Senel, K. Akkaya, M. Erol-Kantarci, and T. Yilmaz, “Self-deployment of mobile underwater acoustic sensor networks for maxi-mized coverage and guaranteed connectivity,” Ad Hoc Netw., vol. 34,no. Supplement C, pp. 170 – 183, Oct. 2015.

[268] Q. Wang, H. N. Dai, C. F. Cheang, and H. Wang, “Link connectivityand coverage of underwater cognitive acoustic networks under spec-trum constraint,” Sensors, vol. 17, no. 12, pp. 1–15, Dec. 2017.

[269] J. Diaz, J. Petit, and M. Serna, “A random graph model for opticalnetworks of sensors,” IEEE Trans. Mobile Comput., vol. 2, no. 3, pp.186–196, Jul. 2003.

[270] D. Kundur, U. N. Okorafor, and W. Luh, “Holistic: Heterogeneouslightweight sensornets for trusted visual computing,” in Int. Conf. onIntelligent Info. Hiding and Multimedia, Dec. 2006, pp. 267–270.

[271] U. N. Okorafor and D. Kundur, “On the connectivity of hierarchicaldirectional optical sensor networks,” in IEEE Wireless Commun. andNetw. Conf., (WCNC), Mar. 2007, pp. 3524–3528.

[272] R. Xie, F. Tong, D. Kang, and Y. C. Kim, “A reconfigurable routingprotocol for free space optical sensor network,” in The Int. Conf. onInfo. Netw., (ICOIN), Jan. 2011, pp. 165–170.

[273] R. Xie, L.-M. Peng, W. Tang, F. Tong, D. ki Kang, W.-H. Yang, and Y.-C. Kim, “A simulation study of neighborhood discovery algorithm infree space optical sensor networks,” in Second Int. Conf. on Ubiquitousand Future Networks (ICUFN), Jun. 2010, pp. 87–91.

[274] R. Ferrero, M. V. Bueno-Delgado, and F. Gandino, “In- and out-degreedistributions of nodes and coverage in random sector graphs,” IEEETrans. Wireless Commun., vol. 13, no. 4, pp. 2074–2085, Apr. 2014.

[275] M. J. B. Appel and R. P. Russo, “The minimum vertex degree of agraph on uniform points in [0,1],” Adv. in App. Probability, vol. 29,no. 3, pp. 582–594, 1997.

[276] AOSN, “Autonomous ocean sampling network (AOSN),” https://www3.mbari.org/aosn/, online; accessed 26 Dec. 2017.

[277] ERI, “Bermuda bio-optics project,” http://www.oceancolor.ucsb.edu/bbop/, online; accessed 26 Dec. 2017.

[278] X. Yu, T. Dickey, J. Bellingham, D. Manov, and K. Streitlien, “Theapplication of autonomous underwater vehicles for interdisciplinarymeasurements in Massachusetts and Cape Cod bays,” Continental ShelfRes., vol. 22, no. 15, pp. 2225 – 2245, 2002.

[279] T. D. Dickey, “Emerging ocean observations for interdisciplinary dataassimilation systems,” J. of Marine Sys., vol. 40-41, no. SupplementC, pp. 5 – 48, Jan. 2003.

[280] W. Philpot, C. O. Davis, W. P. Bisset, C. D. Mobley, D. D. R.Kohler, Z. Lee, J. Bowles, R. G. Steward, Y. Agrawal, J. Trowbridge,R. W. Gould, and R. A. Arnone, “Coastal ocean optics and dynamics,”Oceanography, vol. 17, no. 2, pp. 5 – 48, Jan. 2004.

[281] J. Orcutt, “The ocean research interactive observatory networks(ORION) program,” EoS, Trans. American geophysical union, vol. 84,no. 10, pp. 411 – 415, Oct. 2003.

[282] Y. Nishida, K. Nagahashi, T. Sato, A. Bodenmann, B. Thornton,A. Asada, and T. Ura, “Autonomous survey system of underwatermanganese crusts using acoustic and visual instruments,” in The Proc.

of JSME annual Conf. on Robotics and Mechatronics (Robomec), vol.2016. Japan Society of Mechanical Engineers, Jun. 2017.

[283] S. Srinivas, P. Ranjitha, R. Ramya, and G. K. Narendra, “Investigationof oceanic environment using large-scale UWSN and UANETs,” in 8thInt. Conf. on Wireless Commun., Netw. and Mobile Comp., Sep. 2012,pp. 1–5.

[284] G. Acar and A. E. Adams, “ACMENet: an underwater acoustic sensornetwork protocol for real-time environmental monitoring in coastalareas,” IEE Proc. - Radar, Sonar and Navigation, vol. 153, no. 4, pp.365–380, Aug. 2006.

[285] J. Heidemann, W. Ye, J. Wills, A. Syed, and Y. Li, “Research challengesand applications for underwater sensor networking,” in IEEE WirelessCommun. and Netw. Conf., (WCNC), vol. 1, no. 1, Apr. 2006, pp. 228–235.

[286] J. Lloret, S. Sendra, M. Garcia, and G. Lloret, “Group-based un-derwater wireless sensor network for marine fish farms,” in IEEEGLOBECOM Workshops (GC Wkshps), Dec. 2011, pp. 115–119.

[287] J. Lloret, M. Garcia, S. Sendra, and G. Lloret, “An underwater wire-less group-based sensor network for marine fish farms sustainabilitymonitoring,” Telecommun. Sys., vol. 60, no. 1, pp. 67–84, Sep. 2015.

[288] A. Yalcuk and S. Postalcioglu, “Evaluation of pool water quality oftrout farms by fuzzy logic: monitoring of pool water quality for troutfarms,” Int. J. of Env. Sci. and Tech., vol. 12, no. 5, pp. 1503–1514,May 2015.

[289] M. Lopez, S. Martinez, J. Gomez, A. Herms, L. Tort, J. Bausells, andA. Errachid, “Wireless monitoring of the pH, NH4+ and temperaturein a fish farm,” Procedia Chemistry, vol. 1, no. 1, pp. 445 – 448, Sep.2009.

[290] N. Nasser, A. Zaman, L. Karim, and N. Khan, “CPWS: An efficientrouting protocol for RGB sensor-based fish pond monitoring system,”in IEEE 8th Int. Conf. on Wireless and Mobile Comput., Netw. andCommun., (WiMob), Oct. 2012, pp. 7–11.

[291] G. Cario, A. Casavola, P. G. M. Lupia, C. Petrioli, and D. Spaccini,“Long lasting underwater wireless sensors network for water qualitymonitoring in fish farms,” in OCEANS, Jun. 2017, pp. 1–6.

[292] F. Kratzert and H. Mader, “Advances of FishNet towards a fullyautomatic monitoring system for fish migration,” in EGU GeneralAssembly Conf., Apr. 2017, p. 7932.

[293] C. Alippi, R. Camplani, C. Galperti, and M. Roveri, “A robust, adaptive,solar-powered wsn framework for aquatic environmental monitoring,”IEEE Sensors J., vol. 11, no. 1, pp. 45–55, Jan. 2011.

[294] J. Trevathan, R. Johnstone, T. Chiffings, I. Atkinson, N. Bergmann,W. Read, S. Theiss, T. Myers, and T. Stevens, “SEMAT the nextgeneration of inexpensive marine environmental monitoring and mea-surement systems,” Sensors, vol. 12, no. 7, pp. 9711–9748, May 2012.

[295] C. A. Perez, M. Jimenez, F. Soto, R. Torres, J. A. Lopez, and A. Iborra,“A system for monitoring marine environments based on wirelesssensor networks,” in OCEANS, Jun. 2011, pp. 1–6.

[296] S. Zhang, J. Yu, A. Zhang, L. Yang, and Y. Shu, “Marine vehicle sensornetwork architecture and protocol designs for ocean observation,”Sensors, vol. 12, no. 1, pp. 373–390, Dec. 2012.

[297] H. Yang, H. Wu, and Y. He, “Architecture of wireless sensor networkfor monitoring aquatic environment of marine shellfish,” in 7th AsianControl Conf., Aug. 2009, pp. 1147–1151.

[298] G. Xu, W. Shen, and X. Wang, “Applications of wireless sensor net-works in marine environment monitoring: A survey,” Sensors, vol. 14,no. 9, pp. 16 932–16 954, Sep. 2014.

[299] S. Bainbridge, D. Eggeling, and G. Page, “Lessons from the field twoyears of deploying operational wireless sensor networks on the greatbarrier reef,” Sensors, vol. 11, no. 7, pp. 6842–6855, Jun. 2011.

[300] M. Palaniswami, A. S. Rao, and S. Bainbridge, “Real-time monitoringof the great barrier reef using internet of things with big data analytics,”ITU J.: ICT Discoveries,, vol. 1, no. 13, pp. 1–10, Oct. 2017.

[301] A. Faustine1, A. N. Mvuma1, H. J. Mongi1, M. C. Gabriel1, A. J.Tengel, and S. B. Kucel2, “Wireless sensor networks for water qualitymonitoring and control within lake victoria basin: Prototype develop-ment,” Wireless Sensor Netw., vol. 6, no. 1, pp. 281–290, Jun. 2014.

[302] M. H. Waseem, M. Alamzeb, B. Mustafa, F. Malik, M. Shakir, andM. A. Khan, “Design of a low-cost underwater wireless sensor networkfor water quality monitoring,” IETE J. of Res., vol. 59, no. 5, pp. 523–534, 2013.

[303] G. Tuna, O. Arkoc, and K. Gulez, “Continuous monitoring of waterquality using portable and low-cost approaches,” Int. J. of Dist. Sens.Netw., vol. 9, no. 6, pp. 1–11, Jan. 2013.

[304] K. A. U. Menon, D. P, and M. V. Ramesh, “Wireless sensor network forriver water quality monitoring in india,” in Third Int. Conf. on Comput.Commun. Netw. Techno., (ICCCNT), Jul. 2012, pp. 1–7.

Page 38: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

38

[305] A. Khan and L. Jenkins, “Undersea wireless sensor network for oceanpollution prevention,” in 3rd Int. Conf. on Commun. Sys. Soft. andMidd., (COMSWARE), Jan. 2008, pp. 2–8.

[306] K. S. Adu-Manu, C. Tapparello, W. Heinzelman, F. A. Katsriku, and J.-D. Abdulai, “Water quality monitoring using wireless sensor networks:Current trends and future research directions,” ACM Trans. Sen. Netw.,vol. 13, no. 1, pp. 1–41, Jan. 2017.

[307] M. Waldmeyer, H. P. Tan, and W. K. G. Seah, “Multi-stage AUV-aided localization for underwater wireless sensor networks,” in IEEEWorkshops of Int. Conf. on Adv. Infor. Netw. and Applications, Mar.2011, pp. 908–913.

[308] P. Carroll, K. Mahmood, S. Zhou, H. Zhou, X. Xu, and J. H. Cui, “On-demand asynchronous localization for underwater sensor networks,”IEEE Trans. Signal Process., vol. 62, no. 13, pp. 3337–3348, Jul. 2014.

[309] S. Kemna, M. J. Hamilton, D. T. Hughes, and K. D. LePage, “Adaptiveautonomous underwater vehicles for littoral surveillance,” IntelligentService Robotics, vol. 4, no. 4, p. 245, Jul. 2011.

[310] A. Caiti, V. Calabro, A. Munafo, G. Dini, and A. Lo Duca, “Mobileunderwater sensor networks for protection and security: Field experi-ence at the uan11 experiment,” J. of Field Robotics, vol. 30, no. 2, pp.237–253, 2013.

[311] E. Cayirci, H. Tezcan, Y. Dogan, and V. Coskun, “Wireless sensornetworks for underwater survelliance systems,” Ad Hoc Netw., vol. 4,no. 4, pp. 431 – 446, 2006.

[312] K. Munasinghe, M. Aseeri, S. Almorqi, M. F. Hossain, M. B. Wali,and A. Jamalipour, “EM-based high speed wireless sensor networks forunderwater surveillance and target tracking,” J. of Sensors, vol. 2017,p. 6731204, 2016.

[313] G. Ferri, A. Munafo, A. Tesei, P. Braca, F. Meyer, K. Pelekanakis,R. Petroccia, J. Alves, C. Strode, and K. LePage, “Cooperative roboticnetworks for underwater surveillance: an overview,” IET Radar, SonarNavigation, vol. 11, no. 12, pp. 1740–1761, Dec. 2017.

[314] S. Kumar, A. R. Perry, C. R. Moeller, D. C. Skvoretz, M. J. Ebbert,R. K. Ostrom, S. L. Bennett, and P. V. Czipott, “Real-time trackingmagnetic gradiometer for underwater mine detection,” in OCEANS,vol. 2, no. 2, Nov. 2004, pp. 874–878.

[315] D. P. Williams, “On optimal AUV track-spacing for underwater minedetection,” in IEEE Int. Conf. on Robotics and Automation, May 2010,pp. 4755–4762.

[316] C. Rao, K. Mukherjee, S. Gupta, A. Ray, and S. Phoha, “Underwatermine detection using symbolic pattern analysis of sidescan sonarimages,” in American Control Conf., Jun. 2009, pp. 5416–5421.

[317] I. Leonard, A. Alfalou, and C. Brosseau, “Sensitive test for seamine identification based on polarization-aided image processing,” Opt.Express, vol. 21, no. 24, pp. 29 283–29 297, Dec. 2013.

[318] P. M. Rajeshwari, G. Kavitha, C. M. Sujatha, and D. Rajapan, “Swarmintelligence based segmentation for buried object scanning SONARimages,” in IEEE Underwater Techno. (UT), Feb. 2015, pp. 1–4.

[319] S. Yu, J. Shao, Z. Han, L. Gao, Y. Lin, Y. Tang, and C. Wu, “Mine likeobject detection and recognition based on intrackability and improvedBOW,” in IEEE Int. Conf. on Cyber Techno. in Autom., Cont., andIntelligent Sys., (CYBER), Jun. 2016, pp. 222–227.

[320] S. Khaledi, H. Mann, J. Perkovich, and S. Zayed, “Design of anunderwater mine detection system,” in Sys. and Info. Eng. Des. Symp.,(SIEDS), Apr. 2014, pp. 78–83.

[321] C. Barngrover, R. Kastner, and S. Belongie, “Semisynthetic versus real-world sonar training data for the classification of mine-like objects,”IEEE J. Ocean. Eng., vol. 40, no. 1, pp. 48–56, Jan. 2015.

[322] K. Denos, M. Ravaut, A. Fagette, and H. S. Lim, “Deep learningapplied to underwater mine warfare,” in OCEANS, Jun. 2017, pp. 1–7.

[323] P. E. Hagen, N. Storkersen, K. Vestgard, and P. Kartvedt, “The HUGIN1000 autonomous underwater vehicle for military applications,” inOCEANS, vol. 2, Sep. 2003, pp. 1141–1145.

[324] J. Keller, “Wide-area maritime surveillance with intelligent floats isgoal of DARPA ocean of things project,” http://www.militaryaerospace.com/articles/2017/12/surveillance-ocean-floats.html, online; accessed28 Dec. 2017.

[325] P. Kumar, P. Kumar, P. Priyadarshini, and Srija, “Underwater acousticsensor network for early warning generation,” in OCEANS, Oct. 2012,pp. 1–6.

[326] K. Casey, A. Lim, and G. Dozier, “A sensor network architecture fortsunami detection and response,” Int. J. Dist. Sen. Netw., vol. 4, no. 1,pp. 28–43, Jan. 2008.

[327] S. Tyan, S.-H. Oh, M. C. Domingo, and A. Caiti, “AUV-RM: Underwa-ter sensor network scheme for AUV based river monitoring,” in ICCA,ASTL, 2013, pp. 53–55.

[328] R. Marin-Perez, J. Garcia-Pintado, and A. S. Gomez, “A real-timemeasurement system for long-life flood monitoring and warning ap-plications,” Sensors, vol. 12, no. 4, pp. 4213–4236, Dec. 2012.

[329] E. N. Udo and E. B. Isong, “Flood monitoring and detection systemusing wireless sensor network,” Asian J. of Computer and Info. Sys.,vol. 1, no. 4, pp. 108–113, Dec. 2013.

[330] H. Afzaal and N. A. Zafar, “Cloud computing based flood detectionand management system using WSANs,” in Int. Conf. on EmergingTechnologies (ICET), Oct. 2016, pp. 1–6.

[331] A. Koulakezian, R. Ohannessian, H. Denkilkian, M. Chalfoun, M. K.Joujou, A. Chehab, and I. H. Elhajj, “Wireless sensor node for real-timethickness measurement and localization of oil spills,” in IEEE/ASMEInt. Conf. on Advanced Intelligent Mechatronics, Jul. 2008, pp. 631–636.

[332] H. Denkilkian, A. Koulakezian, R. Ohannessian, M. S. Chalfoun,M. K. W. Joujou, A. Chehab, and I. H. Elhajj, “Wireless sensor forcontinuous real-time oil spill thickness and location measurement,”IEEE Trans. Instrum. Meas., vol. 58, no. 12, pp. 4001–4011, Dec.2009.

[333] A. Massaro, A. Lay-Ekuakille, D. Caratelli, I. Palamara, and F. C.Morabito, “Optical performance evaluation of oil spill detection meth-ods: Thickness and extent,” IEEE Trans. Instrum. Meas., vol. 61,no. 12, pp. 3332–3339, Dec. 2012.

[334] S. Oh and M. Lee, “Oil spill detection sensor using aritificial illumi-nation with blue LEDs,” in IEEE Sensors J., Oct. 2012, pp. 1–4.

[335] M. N. Pangilinan, R. Anacan, and R. Garcia, “Design and developmentof an oil spill detection and transmission system using artificial illu-mination using LEDs,” in 7th Int. Conf. on Intelligent Sys., Modellingand Sim., (ISMS), Jan. 2016, pp. 407–412.

[336] M. Erol-Kantarci, H. T. Mouftah, and S. Oktug, “Localization tech-niques for underwater acoustic sensor networks,” IEEE Commun. Mag.,vol. 48, no. 12, pp. 152–158, Dec. 2010.

[337] R. Diamant and L. Lampe, “Underwater localization with time-synchronization and propagation speed uncertainties,” IEEE Trans.Mobile Comput., vol. 12, no. 7, pp. 1257–1269, Jul. 2013.

[338] G. Cario, A. Casavola, V. Djapic, P. Gjanci, M. Lupia, C. Petrioli, andD. Spaccini, “Clock synchronization and ranging estimation for controland cooperation of multiple UUVs,” in OCEANS, Apr. 2016, pp. 1–9.

[339] J. Crowell, “Method and apparatus for synchronizing clocks underwaterusing light and sound,” May 2017, US Patent 9,645,272.

[340] H. Ramezani, F. Fazel, M. Stojanovic, and G. Leus, “Collision tolerantand collision free packet scheduling for underwater acoustic localiza-tion,” IEEE Trans. Wireless Commun., vol. 14, no. 5, pp. 2584–2595,May 2015.

[341] A. Rushworth, “A review of: underwater acoustic system analysis.” Int.J. of Rem. Sens., vol. 6, no. 2, pp. 365–366, 1985.

[342] P. Etter, “Recent advances in underwater acoustic modeling and sim-ulation,” J. of Sound and Vib., vol. 240, no. 2, pp. 351 – 383, Feb.2001.

[343] M. Stojanovic and J. Preisig, “Underwater acoustic communicationchannels: Propagation models and statistical characterization,” IEEECommun. Mag., vol. 47, no. 1, pp. 84–89, Jan. 2009.

[344] M. Kanaan and K. Pahlavan, “CN-TOAG: a new algorithm for indoorgeolocation,” in IEEE 15th Int. Symp. on Personal, Indoor and MobileRadio Commun., vol. 3, Sep. 2004, pp. 1906–1910.

[345] L. Emokpae and M. Younis, “Surface based anchor-free localizationalgorithm for underwater sensor networks,” in IEEE Int. Conf. onCommun., (ICC), Jun. 2011, pp. 1–5.

[346] P. A. van Walree, “Propagation and scattering effects in underwateracoustic communication channels,” IEEE J. Ocean. Eng., vol. 38, no. 4,pp. 614–631, Oct. 2013.

[347] S. Poursheikhali and H. Zamiri-Jafarian, “TDOA based target localiza-tion in inhomogenous underwater wireless sensor network,” in 5th Int.Conf. on Computer and Knowledge Engineering (ICCKE), Oct. 2015,pp. 1–6.

[348] X. Cheng, H. Shu, Q. Liang, and D. H. C. Du, “Silent positioningin underwater acoustic sensor networks,” IEEE Trans. Veh. Technol.,vol. 57, no. 3, pp. 1756–1766, May 2008.

[349] G. Isbitiren and O. B. Akan, “Three-dimensional underwater targettracking with acoustic sensor networks,” IEEE Trans. Veh. Technol.,vol. 60, no. 8, pp. 3897–3906, Oct. 2011.

[350] J. Choi and H. T. Choi, “Preliminary results on three dimensionallocalization of underwater acoustic sources,” in 14th Int. Conf. onUbiquitous Robots and Ambient Intelligence (URAI), Jun. 2017, pp.347–348.

Page 39: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

39

[351] J. Scheuing and B. Yang, “Disambiguation of TDOA estimation formultiple sources in reverberant environments,” IEEE Trans. Acoust.,Speech, Signal Process., vol. 16, no. 8, pp. 1479–1489, Nov. 2008.

[352] P. E. M. de Magalhes, C. Ioana, J. Mars, and X. Cristol, “Hausdorffdistance applied on real data experiment for underwater localization,”in OCEANS, Jun. 2017, pp. 1–5.

[353] B. Kouzoundjian, F. Beaubois, S. Reboul, J. B. Choquel, and J. C.Noyer, “A TDOA underwater localization approach for shallow waterenvironment,” in OCEANS, Jun. 2017, pp. 1–4.

[354] J. Jung, K. Kim, W. Lee, and H. Kim, “Acoustic localization withoutsynchronization,” in IEEE 31st Int. Conf. on Adv. Info. Netw. andApplications (AINA), Mar. 2017, pp. 423–430.

[355] Y. Kim, Y. Noh, and K. Kim, “RAR: Real-time acoustic ranging inunderwater sensor networks,” IEEE Commun. Lett., vol. 21, no. 11,pp. 2328–2331, Nov. 2017.

[356] M. Hosseini, H. Chizari, and A. S. Ismail, “New hybrid RSS-basedlocalization mechanism for underwater wireless sensor networks,” Int.J. of Comp. Commun. and Netw., vol. 1, no. 11, pp. 1–10, Nov. 2017.

[357] T. Xu, Y. Hu, B. Zhang, and G. Leus, “Rss-based sensor localization inunderwater acoustic sensor networks,” in IEEE Int. Conf. on Acoustics,Speech and Signal Processing (ICASSP), Mar. 2016, pp. 3906–3910.

[358] B. Zhang, H. Wang, T. Xu, L. Zheng, and Q. Yang, “Received signalstrength-based underwater acoustic localization considering stratifica-tion effect,” in OCEANS, Apr. 2016, pp. 1–8.

[359] W. H. Thorp, “Analytic description of the low frequency attenuationcoefficient,” The J. of the Acous. Soc. of Am., vol. 42, no. 1, pp. 270–270, 1967.

[360] GODAE and Ocenview, “Argo drifters,” http://www.argo.ucsd.edu/,online; accessed 19 Dec. 2017.

[361] R. M. Corless, G. H. Gonnet, D. E. G. Hare, D. J. Jeffrey, and D. E.Knuth, “On the lambertw function,” Adv. in Comput. Math., vol. 5,no. 1, pp. 329–359, Dec. 1996.

[362] W. J. Gould and J. Turton, “Argosounding the oceans,” Weather,vol. 61, no. 1, pp. 17–21, 2006.

[363] J. Griffiths, “Beijing plans underwater observation systemin south china sea,” http://edition.cnn.com/2017/05/29/asia/south-china-sea-underwater-observation-system/index.html, online;accessed 19 Dec. 2017.

[364] J. Rice and D. Green, “Underwater acoustic communications andnetworks for the US navy’s Seaweb program,” in Second Int. Conf.on Sensor Techno. and App., (sensorcomm), Aug. 2008, pp. 715–722.

[365] L. Collin, S. Azou, K. C. Yao, and G. Burel, “On spatial uncertaintyin a surface long baseline positioning system,” in Fifth European Conf.on Underwater Acoustics, Lyon, France, Jul. 2000.

[366] B. Foley and D. Mindell, “Precision survey and archaeological method-ology in deep water,” The J. of the Hellenic Inst.of Marine Arch., vol. 6,pp. 49–56, 2002.

[367] R. D. Ballard, “The MEDEA/JASON remotely operated vehicle sys-tem,” Deep Sea Research Part I: Oceanographic Research Papers,vol. 40, no. 8, pp. 1673 – 1687, Apr. 1993.

[368] J. R. Kayser, M. A. Cardoza, W. F. Wade, J. H. Merts, and D. R.Casey, “Weapon Scoring Results from a GPS Acoustic Weapons Testand Training System,” in Proc. of the 2005 National Technical Meetingof The Institute of Navigation, San Diego, CA, Jan. 2005.

[369] L. Chen, S. Wang, K. McDonaldMaier, and H. Hu, “Towards au-tonomous localization and mapping of AUVs: a survey,” Int. J. ofIntelligent Unmanned Sys., vol. 1, no. 2, pp. 97–120, 2013.

[370] Z. Zhou, J.-H. Cui, and S. Zhou, “Efficient localization for large-scaleunderwater sensor networks,” Ad Hoc Netw., vol. 8, no. 3, pp. 267–279,Aug. 2010.

[371] E. Logics, “Underwater acoustic LBL positioning sys.” https://www.evologics.de/en/products/LBL/index.html, online; accessed 01 Jan.2018.

[372] KONGSBERG, “HiPAP - acoustic underwater positioning and naviga-tion sys.” https://www.km.kongsberg.com/ks/web/nokbg0240.nsf, on-line; accessed 01 Jan. 2018.

[373] Sonardyne, “Mini-ranger 2 underwater positioning(usbl) system,” https://www.sonardyne.com/product/underwater-positioning-mini-ranger-2/, online; accessed 01 Jan.2018.

[374] iXblue, “USBL positioning sys.” https://www.ixblue.com/products/usbl-positioning-Sys., online; accessed 01 Jan. 2018.

[375] K. Tech., “Videoray ROV non-acoustic positioning systemsmart tether,” https://www.videoray.com/homepage/new/options/positioning-Sys./smart-tether.html, online; accessed 01 Jan. 2018.

[376] LinkQuest, “Tracklink acoustic tracking and communication sys.” http://www.link-quest.com/html/intro2.htm, online; accessed 01 Jan. 2018.

[377] T. Marine, “Teledyne benthos underwater acoustic sys.” http://www.teledynemarine.com/positioning-Sys., online; accessed 01 Jan. 2018.

[378] Sonardyne, “Bluecomm underwater optical com-munication,” https://www.sonardyne.com/product/bluecomm-underwater-optical-communication-system/, online;accessed 01 Jan. 2018.

[379] STM, “Anglerfish underwater optical communicationsystem,” http://www.navyrecognition.com/index.php/news/naval-exhibitions/dimdex-2016-show-daily-news/3781-stm-unveils-the-anglerfish-uwoc/-underwater-optical-communication-system-at-dimdex-2016.html,online; accessed 01 Jan. 2018.

[380] O. Korotkova, N. Farwell, and E. Shchepakina, “Light scintillation inoceanic turbulence,” Waves in Random and Complex Media, vol. 22,no. 2, pp. 260–266, Feb. 2012.

[381] A. Mora, D. Ganger, G. Wells, J. Zhang, X. Hu, C. Zhou, A. Richa,and C. Youngbull, “Ad-hoc multi-hop underwater optical network fordeep ocean monitoring,” in OCEANS, Sep. 2013, pp. 1–5.

[382] M. V. Jamali, A. Chizari, and J. A. Salehi, “Performance analysis ofmulti-hop underwater wireless optical communication systems,” IEEEPhoton. Techno. Lett., vol. 29, no. 5, pp. 462–465, Mar. 2017.

[383] A. Tabeshnezhad and M. A. Pourmina, “Outage analysis of relay-assisted underwater wireless optical communication systems,” Opt.Commun., vol. 405, no. Supplement C, pp. 297 – 305, Dec. 2017.

[384] W. Wang, P. Wang, T. Cao, H. Tian, Y. Zhang, and L. Guo, “Per-formance investigation of underwater wireless optical communicationsystem using M-ary OAMSK modulation over oceanic turbulence,”IEEE Photon. J., vol. 9, no. 5, pp. 1–15, Oct. 2017.

[385] Y. Song, W. Lu, B. Sun, Y. Hong, F. Qu, J. Han, W. Zhang, and J. Xu,“Experimental demonstration of MIMO-OFDM underwater wirelessoptical communication,” Opt. Commun., vol. 403, no. Supplement C,pp. 205 – 210, Nov. 2017.

[386] S. Han, Y. Noh, R. Liang, R. Chen, Y.-J. Cheng, and M. Gerla,“Evaluation of underwater optical-acoustic hybrid network,” ChinaCommun., vol. 11, no. 5, pp. 49–59, May 2014.

[387] L. J. Johnson, R. J. Green, and M. S. Leeson, “Hybrid underwateroptical/acoustic link design,” in 16th Int. Conf. on Trans. Opt. Netw.,(ICTON), Jul. 2014, pp. 1–4.

[388] C. Moriconi, G. Cupertino, S. Betti, and M. Tabacchiera, “Hybridacoustic/optic communications in underwater swarms,” in OCEANS,May 2015, pp. 1–9.

[389] R. Diamant, F. Campagnaro, M. de Filippo de Grazia, P. Casari,A. Testolin, V. S. Calzado, and M. Zorzi, “On the Relationship Betweenthe Underwater Acoustic and Optical Channels,” IEEE Trans. WirelessCommun., vol. 16, no. 12, pp. 8037–8051, Dec. 2017.

[390] S. Sudevalayam and P. Kulkarni, “Energy harvesting sensor nodes:Survey and implications,” IEEE Commun. Surveys Tuts., vol. 13, no. 3,pp. 443–461, Jul. 2011.

[391] H. Li, C. Tian, J. Lu, M. J. Myjak, J. J. Martinez, R. S. Brown, andZ. D. Deng, “An energy harvesting underwater acoustic transmitter foraquatic animals,” Nature Scientific Reports, vol. 6, p. 33804, Sep. 2016.

[392] H. Wang, J.-D. Park, and Z. J. Ren, “Practical energy harvesting formicrobial fuel cells: A review,” Env. Sci. & Techno., vol. 49, no. 6, pp.3267–3277, 2015.

[393] C.-C. Kao, Y.-S. Lin, G.-D. Wu, and C.-J. Huang, “A comprehensivestudy on the internet of underwater things: Applications, challenges,and channel models,” Sensors, vol. 17, no. 1477, pp. 1–20, May 2017.

[394] Z. Zhou, B. Yao, R. Xing, L. Shu, and S. Bu, “E-CARP: An energyefficient routing protocol for UWSNs in the internet of underwaterthings,” IEEE Sensors J., vol. 16, no. 11, pp. 4072–4082, Jun. 2016.

[395] N. Javaid, S. Cheema, M. Akbar, N. Alrajeh, M. S. Alabed, andN. Guizani, “Balanced energy consumption based adaptive routing foriot enabling underwater wsns,” IEEE Access, vol. 5, no. 1, pp. 40–51,May 2017.

[396] M. Xu and L. Liu, “Sender-receiver role-based energy-aware schedul-ing for internet of underwater things,” IEEE Trans. Emerg. TopicsComput., no. 99, pp. 1–1, Nov. 2016.

[397] M. H. Berlian, T. E. R. Sahputra, B. J. W. Ardi, L. W. Dzatmika,A. R. A. Besari, R. W. Sudibyo, and S. Sukaridhoto, “Design andimplementation of smart environment monitoring and analytics in real-time system framework based on internet of underwater things and bigdata,” in Int. Electronics Symp. (IES), Sep. 2016, pp. 403–408.

Page 40: Underwater Optical Wireless Communications, …communication ranges. Therefore, networking solutions are crucial for mitigating range related deficiencies in order to employ optical

40

Nasir Saeed (S’14-M’16) received his Bachelorsof Telecommunication degree from N.W.F.P Uni-versity of Engineering and Technology, Peshawar,Pakistan, in 2009 and received Masters degree insatellite navigation from Polito di Torino, Italy, in2012. He received his Ph.D. degree in electronicsand communication engineering from Hanyang Uni-versity, Seoul, South Korea in 2015. He was anassistant professor at the Department of ElectricalEngineering, Gandhara Institute of Science and IT,Peshawar, Pakistan from August 2015 to September

2016. Dr. Saeed worked as an assistant professor at IQRA National University,Peshawar, Pakistan from October 2017 to July 2017. He is currently apostdoctoral research fellow at Communication Theory Lab, King AbdullahUniversity of Science and Technology (KAUST). His current areas of interestinclude cognitive radio networks, underwater optical wireless networks, andlocalization.

Abdulkadir Celik (S’14-M’16) received the B.S.degree in electrical-electronics engineering from Sel-cuk University, Turkey [2009], the M.S. degree inelectrical engineering [2013], the M.S. degree incomputer engineering [2015], and the Ph.D. degreein co-majors of electrical engineering and computerengineering [2016], all from Iowa State University,Ames, IA. Dr. Celik is currently a postdoctoralresearch fellow at Communication Theory Labora-tory of King Abdullah University of Science andTechnology (KAUST). His current research interests

include but not limited to cognitive radio networks, green communications,non-orthogonal multiple access, D2D communications, heterogeneous net-works, and optical wireless communications and networking for data centersand underwater sensor networks.

Tareq Y. Al-Naffouri (M10) Tareq Al-Naffouri re-ceived the B.S. degrees in mathematics and electricalengineering (with first honors) from King Fahd Uni-versity of Petroleum and Minerals, Dhahran, SaudiArabia, the M.S. degree in electrical engineeringfrom the Georgia Institute of Technology, Atlanta, in1998, and the Ph.D. degree in electrical engineeringfrom Stanford University, Stanford, CA, in 2004.He was a visiting scholar at California Institute ofTechnology, Pasadena, CA, from January to August2005 and during summer 2006. He was a Fulbright

Scholar at the University of Southern California from February to September2008. He has held internship positions at NEC Research Labs, Tokyo, Japan, in1998, Adaptive Systems Lab, University of California at Los Angeles in 1999,National Semiconductor, Santa Clara, CA, in 2001 and 2002, and BeceemCommunications Santa Clara, CA, in 2004. He is currently an Associateprofessor at the Electrical Engineering Department, King Abdullah Universityof Science and Technology (KAUST). His research interests lie in the areasof sparse, adaptive, and statistical signal processing and their applicationsand in network information theory. He has over 150 publications in journaland conference proceedings, 9 standard contributions, 10 issued patents, and6 pending. Dr. Al-Naffouri is the recipient of the IEEE Education SocietyChapter Achievement Award in 2008 and Al-Marai Award for innovativeresearch in communication in 2009. Dr. Al-Naffouri has also been servingas an Associate Editor of Transactions on Signal Processing since August2013.

Mohamed-Slim Alouini (S’94-M’98-SM’03-F’09)was born in Tunis, Tunisia. He received the Ph.D.degree in Electrical Engineering from the CaliforniaInstitute of Technology (Caltech), Pasadena, CA,USA, in 1998. He served as a faculty memberin the University of Minnesota, Minneapolis, MN,USA, then in the Texas A&M University at Qatar,Education City, Doha, Qatar before joining KingAbdullah University of Science and Technology(KAUST), Thuwal, Makkah Province, Saudi Arabiaas a Professor of Electrical Engineering in 2009.

His current research interests include the modeling, design, and performanceanalysis of wireless communication systems.