Hybrid Satellite - Wireless Sensor Networks Architecture for ...

Hybrid Satellite - Wireless Sensor NetworksArchitecture for Telemedicine Applications in theContext of Emergency Satellite Communications inKu/Ka/Q/V BandsRahim Kacimi1,∗, Ponia Pech2

1University of Toulouse, IRIT-UPS, 118 route de Narbonne, 31062 Toulouse, France2Independent researcher, formerly at TeSA Association Lab., 14-16, Port Saint-Etienne, 31000 Toulouse, France

Abstract

Thanks to the facilities offered by telecommunications, telemedicine today allows physicians and cliniciansto access, monitor and diagnose patients remotely. Telemedicine includes several applications such as remotemonitoring of chronically ill patients, monitoring people in their everyday lives to provide early detection andintervention for various types of diseases, computer-assisted physical rehabilitation in ambulatory settings,and assisted living for the elderly at home, as well as remote monitoring of injured people in a post-disaster situation. These new applications require a reliable, wireless communication link between the devicesimplanted in the patient’s skin and a clinician. In this article, this issue is discussed and a list of performancecriteria for the different communication links used are addressed, especially focusing on the satellite link.Then an adaptive air interface which is designed to meet the performance constraints of bidirectionalsatellite communication links in an emergency situation in Ku/Ka/Q/V bands where when strong channelimpairments occur is described and analysed.

Received on 28 February 2014; accepted on 20 December 2014; published on 28 December 2014Keywords: vital sign monitoring, health-care monitoring, wireless sensor networks, wireless body networks, satellite,telemedicine

Copyright © 2014 Rahim Kacimi and Ponia Pech, licensed to ICST. This is an open access article distributed under the terms of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/), which permits unlimited use, distribution and reproduction in any medium so long as the original work is properly cited.

1. IntroductionNowadays, remote health-care is becoming increasinglyattractive due to several advantages it brings: extendingthe health system coverage to rural and isolated areas,ensuring autonomy to chronic patients by letting themstay at home, allowing the clinicians to remotelydiagnose and monitor their patients, ensuring real-time monitoring for critical illnesses, etc. Moreover,advances in key areas such as Wireless Sensor Networks(WSN) and Body Sensor Networks (BSN) are enablingtechnologies for the application domain of unobtrusivemedical monitoring [42]. This field includes cable-freecontinuous monitoring of vital health signs in intensivecare units, remote monitoring of chronically ill patients,

∗Corresponding author. Email: [email protected]

monitoring people in their everyday lives to provideearly detection and intervention for various types ofdiseases, computer-assisted physical rehabilitation inambulatory settings, and assisted living for the elderlyat home, as well as remote monitoring of injuredpeople in a post-disaster situation. These innovativeapplications, where a pacemaker communicates thepatient’s health state and performance data to a basestation, or a BSN integrating a number of devices,require a reliable wireless communication link betweenthe sensing devices implanted in the patient’s skinand a physician. Otherwise, the wireless link can beused to interrogate the implant at either irregularintervals, or on a regular basis, or to provide nearpermanent communication. A one-way wireless linkmay be used to obtain the patient’s health informationor performance data from the implanted device, while

1

Research ArticleEAI Endorsed Transactionson Mobile Communications and Applications

EAI for InnovationEuropean Alliance

EAI Endorsed Transactions on Mobile Communications and Applications

09 - 12 2014 | Volume 2| Issue 5| e1

doi:10.4108/mca.2.5.e1

http://creativecommons.org/licenses/by/3.0/

mailto:<[email protected]>

R. Kacimi and P. Pech

a two-way link allows external reprogramming of animplanted device [42]. Due to the critical and sensitivenature of the medical information transmitted throughthe wireless network, reliable data transfer for BSNsand network reliability are of paramount importance.In addition to wireless terrestrial networks, satellitelink can (or have to) be used in some cases. Forexample, when the patient is outside of the cellularnetwork coverage, the satellite network can be used totransmit data to the medical call center or to contactthe patient for further information. Network reliabilitydirectly affects the quality of the patient’s monitoring,and in a worst-case scenario can be fateful when alife threatening event has gone undetected. However,due to the constraints on communication bandwidthand power consumption, traditional network reliabilitytechniques such as the retransmission mechanismfor TCP (Transmission Control Protocol) may not bepractical for BSN applications, whereas they are usedin satellite links despite their drawbacks (as in theURSAFE [17] and OURSES [18] projects that will bedescribed later on below). With similar constraintson WSNs, researchers have proposed several methodsfor improving their reliability. One simple approachis to use limited retransmission where packets areretransmitted for a fixed number of times until anacknowledgment is received; however, retransmissionoften induces significant overhead to the network.Another approach is to form a multi-path networkand exploit the multiple routes to avoid disruptedlinks. It is expected that this will be an area thatwill raise significant research interest in the comingyears, particularly in exploring the autonomic sensingparadigm for developing self-protecting, self-healing,self-optimizing, and self-configuring BSNs. Unliketypical wired or wireless network architectures inwhich the network configuration is mostly staticand there are limited constraints on resources, theBSN architecture is highly dynamic, placing morerigorous constraints on power supply, communicationbandwidth, storage and computational resources. Theremainder of this paper is organized as follows:the context of this work and its critical problemsare discussed in section 2. Section 3 states thereliability problem and performance criteria forWBAN (Wireless Body Area Networks) and WSNlinks. Section 4 focuses on satellite link issuesin telemedicine applications. Section 5 thoroughlydescribes an adaptive DVB-S2-based air interfacewhich is designed to meet the performance constraintsof bidirectional satellite communication links in anemergency situation in Ku/Ka/Q/V bands where whenstrong channel impairments occur. In section 6, themain conclusions of our work are summarized, and aset of open issues and future directions is presented.

2. Context and problemTeSA laboratory conducted with its partners twoprojects in the context of remote health-care. Thesetwo projects are UR-Safe (Universal Remote SignalAcquisition For hEalth) and OURSES (Offer of Servicesusing Satellite for Rural Usage). In addition, TeSA alsostudied a low bit rate adaptive air interface for satellitebidirectional links designed for emergency situationsthat is suited to post-disaster telemedicine applications,and will be thoroughly presented in section V.

2.1. UR-Safe project descriptionThe UR-Safe project [17] aimed at creating a mobiletelemedicine care environment for the elderly, thushelping mitigate the problems of health care provisionobserved in the Western societies caused by an agingpopulation and the associated increasing costs. Theadopted technological solution maximizes the conceptsof autonomous living and quality of life for the patient,in alignment with the emerging models of health careprovision, while at the same time addressing safetyand alarm detection issues. The technological solutionconsists in placing medical sensing devices on thepatient’s body, all of them being connected via ashort range Wireless Personal Area Network (WPAN)to a central, portable electronic unit called PersonalBase Station (PBS). These wearable sensors enable torecord electrocardiograms (ECG) and oxygen saturationlevel for instance, while a shock/fall detector sendsan alarm when the patient falls or presses a button.Thanks to speech recognition algorithms, the PBSallows the exchange of simple spoken sentences withthe patient in order to better analyze the patient’s healthcondition. The pieces of information coming from thedifferent sensors and from the shock/fall detector aregathered. Based on these data, preliminary computer-aided diagnosis is performed by the PBS. The data arethen sent to a medical call center.

2.2. OURSES project descriptionThe OURSES project [18] proposed three telemedicineapplications related to services offered to elderlypeople. It focused on the use of satellites as acomplement to terrestrial communication technologiesto ensure the deployment of teleservices in areaswhere telecommunication infrastructure is lacking.These three telemedicine applications are describedin the following. A typical architecture of a remotetelemedicine solution conjugates tree or four differentcommunication technologies according to differentwireless architecture links:

1. The first link: the first link connects the medicalnodes and the coordinator, in order to form aWBAN mainly using a star topology. The WBAN

2EAI for InnovationEuropean Alliance


09 - 12 2014 | Volume 2| Issue 5| e1

Hybrid Satellite - Wireless Sensor Networks Architecture for Telemedicine Applications in the Context of Emergency Satellite Communications in Ku/Ka/Q/V Bands

Figure 1. Global scheme of ECG monitoring in OURSES project[18]

is often composed of a limited number of medicalnodes (body sensors) connected to a coordinator(that may be a normal sensor node or a PDA(Personal Digital Assistant)).

2. The second link: a second link exists betweencoordinators and a base station collector (localPC). The routing of information between the coor-dinators and the base station can be done invarious ways and with different communicationstechnologies: on the one hand, a WSN can beformed with all coordinators. The coordinatorsare also associated with environmental sensorsthat monitor the physical environment surround-ing the patients (temperature, humidity, light,...).The information is then transmitted to the basestation via a WSN mesh topology. On the otherhand, centralized WiFi technology can replace theWSN by setting direct links between coordinatorsand access points to collect information.

3. The third link: a last link is used to transmitdata from the base station (local PC) to aremote physician. In the absence of wide-coverageterrestrial networks, a satellite link is used in thiscase.

Figure 1 shows a usage scenario on wireless vital signmonitoring using different technologies. This solutionis based on a wireless wearable sensor which transmitsECG to a portable device in real time. The latter is thenconnected to a local PC which is located in the patient’shouse through a Wi-Fi link. ECG signals are then sentto the physician’s office using a satellite link. When anabnormal event during an ECG analysis is detected bya specific automatic signal-processing-based diagnosisapplication running on the PC, an alarm is raised andsent to the physician’s office.

2.3. Problem statement

The purpose of this paper is to join togetherdifferent scientific communities involved in suchchallenging projects as telemedicine and biomedicalvital signs remote monitoring, namely, the medicalend-users (physicians, medical experts), the digitalsignal processing scientists from academia, and thetelecommunication scientists and engineers from bothacademia and industry. Indeed, telemedicine projectsinclude so wide and multidisciplinary technical andscientific expertise fields that a minimum of dialoguebetween the different disciplines should be sought forthe sake of a better mutual understanding and a moreunified and integrated approach. More precisely, onepeculiar problem arose in the experiments carried outin the framework of the aforementioned telemedicineprojects: packet loss was observed at the physician’sPC side on the satellite return channel downlink,which manifested itself in “hole” periods, that is,missing samples in the received ECG signal. Inother words, a few ECG samples were missing dueto undetected causes occurring somewhere in thewireless/satellite transmission chain that are attributedto the fact that the different links involved (fixed accessnetwork, mobile access network and satellite DVB-RCS interfacing) are liable to inducing errors or losson the transmitted data. Such data corruption/losscan occur anytime and anywhere. The timeout ofseveral ARQ (Automatic Repeat Request) procedureswas shown to induce packet loss: 1 − 2% for GPRS,and 8% for the satellite [43]. To cope with thatproblem, TeSA devised a recovering method whichhybridized Papoulis-Gerchberg (PG) algorithm and anAuto-Regressive (AR)-based reconstruction algorithm.The principle of the PG exact reconstruction algorithmlies in an interpolation of the missing samples usingthe band-limited property of the ECG signal, via aniterative process which allows to replace the missingpart of the signal with the result of the Inverse FastFourier Transform (IFFT). The PG algorithm performswell with a reduced number of missing samples, butits drawback lies in the long convergence time ofits iterative process. Therefore another ECG missingsample reconstruction method was proposed, namely,an audio signal reconstruction method based on ARmodeling [43]. This method is used to predict forwardand backward signal samples. The final step in theproposed reconstruction algorithm is to jointly use thetwo PG and AR methods, the AR algorithm beingexecuted at initialization phase to be followed by thePG algorithm. The reconstruction performance wasmeasured using the local signal to reconstruction error(noise) ratio given by:

10log(

σ2xgap

σ2(x−x̂)gap

)in dB



09 - 12 2014 | Volume 2| Issue 5| e1


where σ2xgap represents the variance of the original

signal in the missing part and σ2(x−x̂)gap represents the

variance of the reconstruction error also restricted inthe missing part. Tested on the MIT-BIH ArrhythmiaDatabase, the combined method yielded better perfor-mance than the PG and AR algorithms for missingparts of up to 30 consecutive samples (120 ms). Thepaper aims at identifying and qualitatively surveyingall the different factors related to the wireless andsatellite links that may be a cause of packet error or loss,and thus undermine the reconstruction performance ofTeSA’s algorithm, without any quantitative assessmentat this stage. Packet errors and loss can be traced back tomainly three different sources: (i) general performancecriteria of the transmission system; (ii) wireless issues;(iii) satellite link issues.

2.4. Performance criteria

The performance criteria of wireless network linkswhich have significant implications in the healthcaredomain include the number of collisions, the energyconsumption at the nodes, the network throughput, thenumber of unicast packets delivered, the number ofpackets delivered to each node, the signals received andforwarded to the Medium Access Control (MAC) layer,and the change in energy consumption with variationin transmission range, etc. Some of these criteria aredescribed hereafter:

− Delay: the delay for data packets delivery is ofparamount importance in health-care monitoringand its criticality depends on the way datatraffic is transmitted. Indeed, the data can betransmitted on either a periodical basis or a non-periodical basis. A periodic transmission modecorresponds to the case where the traffic datapackets collected by the body sensors are sentperiodically, while a non-periodic transmissionmode refers for instance to the case where analarm is sent by a body sensor every time it detectsan anomaly.

− Packet Received Rate (PRR): in a health-carenetwork solution, a high PRR is necessary so thatthe physician can make a good diagnostic. Forinstance, it may sometimes be difficult for himto interpret ECG data, which makes a diagnosticcompletely impossible.

− Quality of Service (QoS): loss of data is moresignificant in BSNs, and may require additionalmeasures to ensure QoS and real-time datainterrogation capabilities.

3. Wireless sensor and body area networks for vitalsigns monitoring

The advent of smart wireless sensors that are ableto form a BSN would not be possible without theavailability of appropriate and inexpensive low powershort-range transceivers for low to moderate data rates.These are capable of transmitting real-time data with alatency of typically less than one second within a rangeof up to five meters.

Current standardization efforts affect most of thelayers of a communication stack, starting fromthe Physical (PHY) layer, including the MediumAccess Control (MAC) layer and reaching the higherlayers, such as network or routing layers, and evensometimes the data representation and applicationlayers. Different standardization bodies may work in acooperative fashion, as is the case with ZigBee and IEEE802.15.4.

The problems encountered in BSNs involved respec-tively on the first and the second links listed above aresummarized as follows:

− Energy consumption: It is widely recognizedthat limiting energy is an inescapable issue inthe design of wireless BSNs due to the strictconstraints which it imposes on the networkoperations. In fact, the energy consumption is acrucial factor impacting the network lifetime thathas become the prevailing performance criterionin this area. If the network is to operate as longas possible, these energy constraints require makea trade-off between various activities at both thenode and network levels, so that the less energyconsumed by the nodes, the longer the networklifetime to satisfy the running application [15].

− Scalability: Scalability is an important factorin designing efficient WSN solutions. A goodsolution has to be scalable in the sense ofbeing adaptable to future changes in the networktopology. Thus scalable protocols should performwell as the network grows larger or as theworkload increases.

− Congestion: In healthcare WSN applications (par-ticularly for medical emergencies or closely mon-itoring critically ailing patients), it is obviouslydesirable in the first place to avoid congestion,and should it occur, to reduce data loss due tocongestion.

− Mobility issues: The wireless network solutionmust manage the mobility of equipments andmobility of persons in order to maintain a goodconnectivity.



09 - 12 2014 | Volume 2| Issue 5| e1


3.1. IEEE-802.15.4 StandardAlthough a wireless BSN is not always a lowest dutycycle application (such as continuous ECG streaming),the ZigBee/IEEE 802.15.4 framework appears to bethe most intriguing and suitable protocol suite for it.The IEEE 802.15.4 MAC offers a number of valuableingredients for BSNs: the MAC is optimized for lowpower and short messages and includes peer-to-peernetwork support, guaranteed time slots, etc. IEEE802.15.4 is also likely to be chosen as the radio layerbasis for IEEE P1451.5-based wireless sensors [39].Highly integrated single-chip IEEE 802.15.4-complianttransceivers are already available from a number ofIC manufacturers, yet they are a bit more powerhungry than simple FSK (Frequency Shift Keying)transceivers because of DSSS (Direct Sequence SpreadSpectrum), but they offer better robustness and betterinteroperability compared with FSK. Of course, the datarate is not sufficient to carry video data in ambientapplications, but it could well convey pre-processeddata, e.g. from a camera system that detects whena person is moving or falling. The IEEE 802.15.4aalternate PHY may add another interesting flavourto BSNs in the not-too-distant future. Worldwideinterest in ZigBee-/IEEE 802.15.4-compliant productswill inspire global creativity and keep costs down. In[39], the authors present a usage scenario on wirelessvital sign monitoring using IEEE802.15.4a standard.The 15.4a piconet is composed of one piconet controller(PNC) and a number of vital-sign sensors attached tothe patient. The PNC, which plays the role of a dataaggregator, is located at bedside; thus, the distancesbetween the PNC and the sensors are generally shorterthan 2 meters. Vital signs typically monitored in apatient’s body can be categorized into two types: onetype corresponds to continuous data, that is, datainformation (such as ECG) continuously transmittedfrom a sensor, and the other type corresponds to routinedata, which is generated sporadically from the sensorsincluding body temperature (BT), oxygen saturation(SpO2), and blood pressure (BP). For the continuousdata type, wireless transmission which supports delayQoS maintenance is required because ECG waveform isa streamed data signal.

3.2. IEEE-802.15.4 Link QualityIn this section, we introduce LQI (Link QualityIndicator). The LQI measurement is a characterizationof the strength and/or the quality of a received packet.It can be implemented using the Energy DetectionReceiver (ED Receiver), an estimation of Signal-to-noiseratio or a combination of the two methods. The useof this measure in the network and the applicationlayers has not yet been specified in the IEEE-802.15.4standard.

Figure 2. Sensor network with 3 nodes and a BS

LQI measurements must be performed for eachreceived packet, and the result must be reported to theMAC layer using the original PD-DATA indication as aninteger from 0-255 [39]). Minimum and maximum LQIvalues (0 and 255) should be associated respectivelywith the lowest and the highest quality of the signalsthat can be detected by the receiver. The LQI valuesare uniformly distributed within these two limits. Lateron, we will see that it really depends on the type ofthe receiver that is used, including that of the CC2420radio module where these limits are actually 50 and110. Many platforms, (e.g. micaZ, Telos, Intel Mote2,. . . ) use a CC2420 radio module.

As a matter of fact, the LQI is a hardware indicatorprovided by the CC240 module, and actually is ameasure of the bit error rate. In addition to the LQI, theCC2420 module also provides the RSSI (Received SignalStrength Indicator. The observed limits of the RSSI inold platforms have led several routing protocols as [40]to adopt the LQI as the preferred metric. The resultsgiven in [41] also indicate that the RSSI is stronglycorrelated with the PRR, except when operating atthe receiver sensitivity limits. Meanwhile, the LQI canmake more accurate estimates, requiring averagingmany readings, which reduces flexibility and increasesthe cost of the estimation. The results in [41] suggestthat there is a correlation between hardware indicatorsbehavior and the protocols using them.

Experimental test. Experiments were performed usingTmoteSky devices. The platform is smaller than abusiness card, and includes a microcontroller operatingat 8MHz, with 48K of ROM, and 10K of RAM, a 2.4GHzZigBee wireless transceiver, and a USB interface fordevice programming and logging. Each device operateson 2 AA batteries.

In order to understand the link quality indicator weachieved some tests and shows how this parametervaries depending on the transmission power and thedistance between the nodes. As shown in Fig.2, thefirst test involves four sensor nodes including a basestation to collect the data packets. The nodes form a star



09 - 12 2014 | Volume 2| Issue 5| e1


Figure 3. Link Quality vs. time and TPL

topology and use different transmission power levels(TPL). In fact, the equipment software does not directlymanipulate the power in dBm. Thus, the power level iscoded by an integer between 1 and 31 correspondingrespectively to −25 dBm and 0 dBm.

Fig. 3 plots the LQI variation over time for thenetwork links. First, we note that LQI measurementsdepend on the transmission power. Indeed, the higherthe transmitter power is, the better the measured LQIis. Considering the packets received by the Base Station,on average, the LQI of the link 1 -> BS is slightly higherthan that of the link 2− > BS.

In the second test, we studied the LQI variationaccording to the distance between node 1 and the BS.The distance ranges from 3 to 35 meters and the nodesends one packet every 3 seconds. The BS measures theLQI when it receives these packets. The two nodes areplaced in a corridor with an ambient temperature of20C. We observed the average LQI of each set of 1000packets based on the distance between the sensor nodeand the BS. We also noted a variation of the average IQIin function of the distance. Generally the average LQIdecreases when the distance increases.

Further measurement analysis showed that if weconsider the average LQI values, the PRR follows asmooth curve suggesting a better correlation betweenLQI and PRR. These results are plotted in Fig.4, and thecorrelation coefficient (Pearson) is 0.87. However, somecases, for instance when LQI = 85, could result in anyvalue of PRR ranging from 10% − 100%. Although wedid not fully understand what can cause this, we believethat it may be related to the environment changes andinterferences with other network technologies such as802.11.

Finally, in this experimental study, we show howthe network links are sensitive to parameters such asthe distance between the nodes and their transmissionpower.

Figure 4. PRR vs. LQI.

4. Satellite issues in telemedicine applicationsA satellite link raises several issues in a networkdeployed for telemedicine applications. Some of themmay have a multifold impact on the biomedical signalin terms of: receive signal quality; signal reconstruc-tion algorithm performance; quality of service (QoS)performance. In the following, the focus will be placedupon, (i) the satellite channel itself in relation withpropagation impairments in high frequency bands, andthe nature of errors occurring in a satellite link; (ii) QoSissues with a special emphasis on performance require-ments related to the transmitted IP-based traffic usablein telemedicine applications, (iii) and quite significantissues related to the TCP (Transmission Control Proto-col) over the satellite link. The discussion is limited togeostationary satellites.

4.1. The satellite channel issueThe satellite channel is a sensitive link in two respects:

(i) with regards to atmospheric impairments in highfrequency bands, which can cause severe linkoutages;

(ii) with regards to the error behavior, bit errorstending to cluster in bursts. An analysis and athorough characterization of error patterns arerequired in order to assess the impact of bit errorson higher layers protocols.

These two aspects are detailed in the following.Channel impairments in high frequency bands: Inhigh frequency bands above the Ka band, troposphericeffects in the satellite propagation channel may bestrong, and thus detrimental to communications. Twocategories will be considered [19]:

Atmospheric attenuation due to gas, water vapor,clouds, the melting layer, and rain, the rain componentbeing the prevailing factor for percentages of time lowerthan 1% (cf. figure 5). The rain attenuation componentresults in slow signal fading variations and can yield



09 - 12 2014 | Volume 2| Issue 5| e1


Figure 5. Components of the atmospheric attenuation versus thefrequency for 0.01% of the time

a magnitude of 20 dB for 0.05% of an average yearin Ka band. It represents the most serious limitationto the performance of satellite communication linksin the millimeter wave domain. In addition, secondorder statistics of rain attenuation should also be takeninto account in the design of a satellite communicationsystem, since they directly relate to outage durations.Amplitude scintillation manifesting itself in the form ofrapid signal fluctuations. Scintillation may impact thelong term system availability (time percentages higherthan 1%).

Channel error behavior modeling. The numerous statisti-cal propagation models that exist today for Ku bandand above enable to estimate the degradation on a givenlink in terms of signal-to-noise ratio (SNR) C/N0. Nev-ertheless, for new SatCom systems, C/N0 loss does notdirectly relate to the degradation of the QoS offered tothe end user. Thus, in order to be able to obtain a soundestimation of the QoS information, it becomes necessaryto resort to methods that link the channel with higherlayers mechanisms or more QoS-oriented parameters.In this perspective, an analysis and a modeling of thesatellite channel error behaviour are also required.

Parametric analytical models. A first approach in thatdirection lies upon parametric analytical models. Mostof the models built for the wireless fading channel usediscrete-amplitude and discrete-time Markov chains,among which the two-state Gilbert model is a well-known one [33]. It was demonstrated that the errorpattern can have a considerable impact on theperformance of protocols at higher layers such asTCP or any ARQ-typed retransmission protocols [29],especially when errors occur in bursts. Combinedwith a broadened definition of outage events, differentfrom the conventional definition as the exceeding of athreshold by a first order statistical parameter (such asthe BER [Bit Error Rate] or the PER [Packet Error Rate][35]), or with a peculiar framing strategy (for instance

interleaving) [33], N-state Markov models enable tocapture the intrinsic bit errors correlation and deriveaccurate predictions of the system performance whenN increases.

Error performance methodology for ATM (AsynchronousTransfer Mode) by satellite after the ITU-R Rec. S.1062-1WP-4B. The ITU-R Rec. S.1062-1 WP-4B constitutesa second approach towards the goal expounded above,that is, linking higher layers performance parameters tothe physical layer parameters [3]. The recommendationdefines a specific methodology to be applied whendesigning a satellite system using ATM with thepurpose of satisfying the G.826 recommendation whichaddresses physical layer performance. The S.1062-1 WP-4B recommendation also links ITU I.356 QoSparameters at ATM layer with higher layers QoSparameters. It can be considered that the ITU-RS.1062-1 WP-4B recommendation provides a fruitfulmethodological framework for linking performanceparameters located at different levels. These relationshave been exploited in other contexts and studies [7, 8].A same type of methodology should also be appliedto assess the error performance of satellite links, andtheir deep and maybe subtle impacts on higher layersprotocols such as TCP, which will directly translatesinto a level of biomedical signal restitution quality inour case of interest.

(i) Errors model: In order to properly study errorperformance in a satellite link, a valid modelof error statistics is required. The most commonmodel is that of random errors in which asequence of statistically independent bits isobserved from two possible Bernoulli outcomes:“errored” or “non-errored”. The number of errorsduring the observation period then follows abinomial distribution, but if the observationperiod is quite long and the bit error probability isvery low, it can also be characterized by a Poissondistribution.

(ii) Statistical characterization of error bursts usinglattice diagrams: The random errors model isinappropriate when errors occur in bursts mainlybecause of the memory introduced by signalprocessing techniques in communication systems.It is possible to statistically characterize thepatterns of the errors bursts (times and inter-arrival times of the bursts) at the output ofthe decoder by means of the lattice diagram ofthe coder, by invoking the concept of transferfunction (Viterbi) of the convolutional code,and using an algorithm that systematicallyand exhaustively collects all patterns of bursterrors. The approach can be extended so asto characterize the effects of the scrambler, the



09 - 12 2014 | Volume 2| Issue 5| e1


descrambler, the interleaver and the concatenatedcoding, which are processes that are commonlyused in DVB-RCS satellite communications.

(iii) Generic models of burst errors: The character-ization methods presented previously are spe-cific to a particular coding scheme. Consequentlytheir interest is limited for globally evaluatingthe performance parameters of digital satellitelinks. Therefore more generic models are basedon a small number of statistical quantities suchas typically the average length L of the burstsand the average number of errors per burst. Bursterrors are generally assumed to follow a Neyman-A contagion distribution.

4.2. QoS issuesFor satellite communications, one of the most criticalrequirements is to provide the desired QoS level tothe different services. The discussion will focus onIP applications. The QoS issue should be dealt withaddressing the diverse network layers and the QoSarchitecture, and assessing the application and networkbehaviour. In addition to classic QoS metrics [16],subjective metrics should also be introduced in orderto evaluate the IP applications from a user pointof view, such as PSNR (Peak Signal to Noise Ratio)and MOS (Mean Opinion Score) [28]. The data to betransmitted is characterized mainly in terms of bit rate,overhead, error rates (BER [Bit Error Rate], or PER[Packet Error Rate]), delay, jitter, average and maximumpacket sizes. In the following, two types of servicesare considered and their salient QoS features will bepresented along with a number of constraints fromhigher layer protocols [20–22]: real-time applications;data-like loss-sensitive, but not or very little delay-sensitive applications.

4.3. QoS requirements for basic applicationsReal-time applications. A first typical real-time com-monly required service is VoIP, which is a delay-sensitive but very little loss-sensitive application. AVoIP call is expected to be intelligible. A strictlyminimum bit rate of 5.3 kbps (assuming ITU H.323G.723.1 ACELP [Algebraic Code Excited Linear Predic-tion] codec) is required. It has been shown however thatoptimal bandwidth occupation for VoIP over satelliteis around 12 kbps. VoIP bit rate also varies dependingon the codec, on whether RTP (Real-Time Protocol)is compressed, and on the redundancy introduced bythe headers of the protocol suite (Ethernet, IP, UDP[User Datagram Protocol], RTP). The bit rate can thusbe considered to range between 5.3 and 13 kbps. Aminimum MOS requirement of 3.5 ensures a good voicequality. Moreover, ITU-T G.114 Recommendation [12]

specifies a maximum latency value of 150 ms for one-way VoIP communications. Lastly, in terms of packetcorruption and loss, some experiments have shown thatthe satellite link is quite robust to packet corruption inclear sky or moderately degraded channel environmentup to a BER of 10-5 (that is, Frame Erasure Rate orFER of 2%) [25]. A second type of real-time servicesconcern video applications. These applications rangefrom real time communications to surveillance, Internetvideo streaming, as well as collaborative scene visual-ization, broadcasting and virtual meetings. AlthoughQoS constraints are strongly dependent on the appli-cation considered, the following QoS specifications canbe used as a baseline for video services [23]: the variableaverage bit rate allocated to a video application shallnot be lower than 256 kbps in the two ways. Criticalapplications such as telemedicine require a fairly goodvideo quality. The maximum transmission delay shouldbe lower than 400 ms. The video codec which is used(for instance H.323) should be able to provide a goodpicture quality for telemedicine applications. A videoconnection should be established in less than 30 sec-onds for high priority applications.

Data-like loss-sensitive, but not or very little delay-sensitiveapplications. In this category are for instance SMS(Short Message Service) / MMS (Multimedia MessagingSystem), email applications, file exchange and Internetbrowsing. The transmitted mean bit rate must be atleast 32 kbps for Web browsing and file exchange,and 200 kbps for email applications [31]. BER valuesof up to 10-6 can be supported [31]. Moreover, thetime interval between the sending of an SMS and itsreception by the receiver must be between 6 and 8seconds in average, given that actually 98% of sentSMSs are successfully delivered by a mobile user toa fixed network within a 5 sec time period, accordingto some telecom operators [4, 5]. Since the integrityof SMS messages is 100%, it is obvious that SMSs arewell fitted to telemedicine communications, especiallyin emergency situations [20–22], where there is a needto transmit an alarm.

Other QoS issues. Another crucial QoS optimizationmethod consists in properly handling and managingdata traffic, especially when different services areaggregated. Differentiated services QoS architecture hasreceived much attention these last years as traffic flowsare of mixed categories (TCP flows and UDP flowsfor instance). By assigning each IP packet a specifictraffic class, a more optimized management bufferis made possible resulting in improved bandwidthresource utilization. In particular, if excess TCP andexcess UDP were both treated equally, TCP flowswould reduce their rates on packet drops while UDPflows would not change, and instead monopolize theentire excess bandwidth [2]. All this leads to proper



09 - 12 2014 | Volume 2| Issue 5| e1


buffer management associated with efficient droppingstrategies.

4.4. Issues related to TCP over the satelliteTCP is the most widely used Internet transmission pro-tocol located at the OSI transport layer. TCP allows anend-to-end flow control mechanism between a senderand a receiver on the Internet with acknowledgments(ACKs) being sent back by the receiver to the sender,and to vary the transmission packet rate based on therate at which acknowledgements are received back fromthe receiver. This mechanism enables to verify the cor-rect delivery of data between a client and a server. TCPsupports error or data loss detection, and implements aretransmission technique until the packets sent are cor-rectly and completely received [26]. It also implementsa network congestion control. All these TCP featuresmake it a reliable and efficient transport protocol overthe Internet stack, independently from the applicationsabove it and the Internet below it. Moreover, the Inter-net is quite a particular network because it consistsof different network topologies, bandwidth, delays andpacket sizes. The satellite link possesses inherent char-acteristics that may have negative impacts on the TCP.Two of them are mentioned here:

− Transmission errors: satellite channels are muchmore prone to bit errors than typical terrestrialnetworks. A characterization of the burst errorsin satellite links has been given in section 4.TCP assumes that all packet drops are caused bynetwork congestion to avoid congestion collapse[27].

− Latency: latency is due to propagation delay,transmission delay, and queuing delay [9]. Ofcourse, the round trip time (RTT) propagationdelay of about 275 ms in geostationary linksis the prevailing term. The dominant additionto the end-to-end one way latency will beroughly 300 ms of fixed propagation delay [9].This delay mainly impacts some of the TCPcongestion control algorithms. Originally, theTCP protocol suite was designed for a terrestrialenvironment with short transmission delays thatseldom exceed 250 ms [1]. When applied to ageostationary satellite link, TCP performs poorlydue to the long latency introduced between aground Earth station and the satellite. Latencycalls for protocol-specific acceleration.

Hence, the use of TCP over the satellite raises importantissues to be pondered. Two of them deserve specialattention and are tackled in the following [6, 9].

Capacity, latency and congestion. TCP is responsible forflow and congestion control, ensuring that data is

transmitted at a rate consistent with the capacitiesof both the receiver and the intermediate links inthe network path. Since there may be multiple TCPconnections active in a link, TCP is also responsiblefor ensuring that a link capacity is responsiblyshared among the connections using it. As a result,most throughput issues are associated to TCP. Fourcongestion control mechanisms exist in TCP: slow start,congestion avoidance, fast retransmit before the RTO(Retransmission Time-Out) expires, fast recovery toavoid slow start [26]. The basic principle of the TCPprotocol congestion mechanism can be summarizedas follows [9]: a congestion window is initialized toa value of one segment upon connection startup. Itdetermines the TCP sending rate. During the slow startphase, the congestion window doubles every roundtrip time (RTT), until congestion is experienced dueto a data packet loss. The congestion avoidance phaseis entered upon detection of congestion. Then TCPretransmits the missing segment, and the window isemptied down to its half content. If retransmittedpackets happen to be lost again, the TCP sender isforced to retransmit the missing packets, but this isdone with an imposed timeout where slow start isresumed and the window is reduced to one segment[26]. Consequently, the throughput becomes very low.Congestion control mechanisms in TCP thus degradethe performance of individual TCP connections oversatellite links because the algorithms slowly probe thenetwork for additional capacity, which in turn wastesbandwidth. Indeed, the satellite latency which caneasily exceed 2000 ms is seen as evidence of a congestednetwork or packet loss and thus TCP will not increasethe rate at which it sends packets, even though there isno actual congestion or packet loss across the satellitelink [1].

TCP acceleration and the security issue. Several tech-niques to accelerate TCP exist. Performance EnhancingProxies (PEPs) are one of them and basically involvean alteration of the TCP header data before and afterthe satellite link in order to hide the high latency ofthe satellite link from the TCP session [1]. Examplesof PEP techniques are TCP spoofing and TCP multi-plexing (also known as cascading TCP or split TCP).TCP spoofing consists in shortening the delay pathand thus bypassing slow start, by adding a spoofingdevice/software (e.g. a rooter near the satellite link)which is in charge of returning ACKs to the sender,and in the meantime suppressing the ACKs from thereceiver [1, 26]. The spoofing device also retransmitsany segments lost and contains storage buffers. TheTCP multiplexing technique accelerates data transferrates across the satellite link by converting a singleTCP sessions into several parallel TCP sessions. At the



09 - 12 2014 | Volume 2| Issue 5| e1


receive side, all TCP sessions are recombined into a sin-gle session [1]. Problems arise when TCP accelerationis achieved simultaneously with security by means ofa VPN (Virtual Private Network) in the tunnel mode.Assuming that the data packets enter the VPN tunnelbefore TCP is accelerated, one is left with TCP packetsentirely encrypted and the header of which cannotbe altered anymore, or otherwise the authenticationsafeguards would be violated [1]. TCP multiplexingtechnique alone is compatible with VPN, but the pro-cessing must not take place inside the satellite modem.In addition, a number of performance degradation fac-tors appear with the technique. All these performanceand security issues associated with TCP accelerationhave led to the development of a number of propri-etary solutions to optimize the bandwidth resourceand utilization for TCP over satellite. Among these,the End II End’s patent-pending Broadband NetworkOptimization (BNO) [1] and UDcast solutions are worthto be mentioned [3]. With the revised, mobile versionof the DVB-RCS standard, called DVB-RCS+M [6] aswell as with the recent DVB-RCS2 standard, whichboth use the DVB-S2 waveform and ACM feature forthe return channel, other optimization issues over asatellite link would also need to be discussed. In par-ticular, IP encapsulation efficiency depending on theencapsulation technique employed, whether it be MPE,GSE or ULE, should be investigated for the return linkas for the forward link [11].

5. DVB-S2-based air interfaceThis section describes a DVB-S2-based adaptive airinterface designed by TeSA in the framework of anR&T contract with the French Space Agency (CNES)to meet the performance constraints of bidirectionalsatellite communication links to be established in apost-disaster emergency situation in Ku/Ka and Q/Vbands where strong channel impairments occur. Theproposed solution allows to establish very quicklya minimum low bit rate satellite return link, usingthe available resources of a primary geostationarysatellite system, and very small, low-cost and power-limited dedicated ground terminals. The bi-directionaltransmission link is modeled within the Juzzle opensource environment software with an emphasis on thereturn channel, deploying an adaptive strategy basedon the DVB-S2 adaptive coding and modulation (ACM)scheme. This section expounds the enhanced DVB-S2-like air interface proposed to support telemedicineapplications in an emergency situation in a severelyimpaired channel environment in high frequencies; italso outlines the combined Excel/Juzzle/Matlab high-level transmission link software simulation platformthat was developed in order to assess the performanceof the proposed transmission scheme. The simulator

architecture follows a cross-layer approach, integratingpropagation, DVB-S2-based physical layer, and trafficcomponents. Some simulation results are also provided,emphasizing on the adopted adaptive strategy.

5.1. System architecture and scenarioThe proposed system architecture and scenario wereas follows [21]: the emergency mission signals aresuperimposed with those of the primary systemcharacterized by a star topology, a classic multibeam,multicarrier, broadband bent-pipe satellite operatingeither in Ku/Ka or Q/V-band, and with a 120-MHz transponder. Thus a dedicated channel for theemergency mission is not required. The gateway hasat its disposal a bandwidth of 480 MHz (forwardchannel), while the user links share a total bandwidthof 240 MHz on the return channel, in four 120-MHz sub-bands and two polarizations, with a reusefrequency factor of 1 over 4. Furthermore, an all-IP(Internet Protocol) architecture is assumed. For theemergency mission, a set of 4 different types of userterminals was considered: two mobile user terminalsof very low transmission power (between 0.5 and 2W), the first one (UTA) having a patch antenna, andthe second one (UTB) having a less directive antennaand thus degraded link budget performance and twodeployable or transportable user terminals: UTC is arapidly deployable âĂĲmini-gatewayâĂİ mounted ona van, transmitting at up to 50 W, while UTD can betransported by a human user and has a transmissionpower of 5 W.

5.2. Enhanced DVB-S2 air interfaceIt was proposed to adopt the ETSI DVB-S2 ACMMODCODs [25] for the return channel as it providesexcellent performance close to the theoretical Shannonlimit due to an advanced Forward Error Correction(FEC) scheme (concatenated BCH and LDPC codes),and allows an attractive waveform flexibility inpresence of channel fading, with its inherent ACMcapability. Incidentally, this adoption of DVB-S2ACM schemes for the return channel was recentlystandardized within the DVB-RCS+M working group[6], but in the latter standard, very short (4 kbits)DVB-S2 PLFRAMEs were envisaged instead of normaland short lengths (64,800 bits for the normal frame,and 16,200 bits for the short frame). By contrast, here,standard-length DVB-S2 PLFRAMEs were assumed. Inquasi-error free (QEF) environment (PER of 10−7), inan Additive White Gaussian Noise (AWGN) channel,DVB-S2 operates at ideal Es/N0 ranging from 16down to -2.35 dB. The performance of the receiverin terms of signal acquisition/ reacquisition time,decoding thresholds, etc., all are well known [26].The proposed novelty was the use of DVB-S2 along



09 - 12 2014 | Volume 2| Issue 5| e1


with spread spectrum and other adaptive mechanismsfor return channel interactive services while DVB-S2 was specifically designed for the forward linkand broadcasting services. This thus raises someperformance challenges to be coped with. Since theconsidered applications all were assumed to be IP-based ones, the DVB-S2 waveform was coupled withan efficient encapsulation mechanism, namely theGSE (Generic Stream Encapsulation) protocol at theSegmentation and Reassembly (SAR) layer, aimed atsegmenting network layer IP datagrams (PDUs orPayload Data Units) into link layer DVB-S2 basicdata units called BBFRAMEs (base band frames).The GSE mechanism was designed with the purposeof fully taking advantage of the innovative featuresof the DVB-S2, primarily in terms of reliability,flexibility, and enhanced capacity. MPE (Multi-ProtocolEncapsulation) and ULE (Unidirectional LightweightEncapsulation) have been the standard encapsulationtechniques that were classically used in DVB-typedsatellite systems, and as such they received abundantattention in the literature. Nonetheless, GSE constitutesa much more efficient encapsulation scheme fittedto the DVB-S2 standard in that it allows to fullyexploiting the adaptive ACM capability, implementingQoS scheduling decisions and flexible placement andenhanced fragmentation of PDUs in the flow [27]. Inparticular, DVB-S2 GS (Generic Stream) data flows maybe packetized or continuous streams. The first onesare suited to carrying PDUs of constant size, whereasthe latters were designed to seamlessly adapt to inputstream of any format, including continuous bit streamsand variable-sized PDUs such as IP datagrams. GSEcan avoid using MPEG2 packets as with MPE andULE, which would be sub-optimal in the frameworkof DVB-S2. In effect, the GS flow is more suited tointeractive services due to a liberty from inadequateMPEG2-TS constraints of constant bit rate and end-to-end delay. In addition, due to the large sizes ofa BBFRAME payload (up to 40 times as long as anMPEG2 packet), datagram fragmentation occurs lessoften. Measures in the Internet network backbone showthat the mean size of an IP datagram is about 500 bytes,which roughly amounts to 7000/500 = 14 IP datagramscarried in the longest available BBFRAME, against 2 or3 fragmentations on the MPEG2-TS layer, and up to 10in the case of ATM [27]. GS streams are tailored into21 possible BBFRAME frames, thus offering a varietyof efficiency versus error protection compromises, andpredefined sizes ranging from 384 to 1,779 bytes(short BBFRAMEs), or 2,001 to 7,274 bytes (for longBBFRAMEs). Consequently, all these characteristics ofGSE result in IP datagrams being delivered morerapidly, efficiently and optimally in a cross-layerperspective, with reduced redundancy and complexity[27]. A last technique deployed for the purpose of

enhancing the DVB-S2 transmission performance wasthe Spread Spectrum (SS) in its Direct Sequence (DS)variant. Besides its resistance to interference, jammingand multipath impairments, a quite powerful propertyof SS exploited in the framework of this study is itsprocessing gain GP , defined as the ratio of the spreadbandwidth over the original bandwidth in dB. Thisprocessing gain can thus be added to the S side of theSNR calculation [28]. This fruitful property is due tothe power-bandwidth trade-off that exists in any radiocommunication system: using a spread spectrum signalenables the system to operate at negative signal to noiseratios, thus allowing to deploy smaller terminals withreduced transmission power with respect to the non-spread case. This consequently means improved batterylife in the case of portable terminals, as well as an easierand quicker deployment of the terminals. Thereforethe adequacy of SS is straightforward for emergencycommunications, in heavy rain environment [29]. TheDirect Sequence Spread Spectrum (DS-SS) block canonly be inserted before the base-band filter and beforethe modulator. In this technique, the pseudo randomnoise (PRN) is applied directly to the data entering thecarrier modulator. The modulator therefore sees a muchhigher bit rate, which corresponds to the chip rate ofthe PRN sequence. The purpose of modulating an RFcarrier with such a code sequence is to produce a direct-sequence-modulated spread spectrum with ((sinx)/x)2

frequency spectrum, centered at the carrier frequency.There is no changing the point in the system where theDS-SS must be placed otherwise it would be a quitedifferent SS technique. As a result, that means that astandard DVB-S2 transmitter cannot be used as a blackbox, but that the transmission chain must undergo somedesign adaptations, so as to conform to the provisionenvisaged in clause 5.1 of the mobile version of theDVB-RCS standard [6].

5.3. Integrated simulation platformGeneral description. A software model of the enhancedDVB-S2 transmission link was developed in mixed stan-dard C and Java languages within an open source soft-ware environment called Juzzle [24]. The whole simu-lation platform was an integrated Excel/Juzzle/Matlabsoftware package which also implements the C DIS-LIN [30] data plotting library. The Juzzle simulationplatform The Juzzle simulation platform is composedof two separate and standalone Juzzle components: achannel propagation time series generator componentand a processing core component, and of an extern post-processing Matlab routine.

Off-line propagation module. The propagation channeltime series generation module is a pre-processing Juzzlecomponent allowing to generate an attenuation timeseries configurable with a number of user parameters.



09 - 12 2014 | Volume 2| Issue 5| e1


Figure 6. Juzzle link model

The synthesizer relies on the theoretical principlesof the enhanced Maseng-Bakken model which wasused by Lacoste [34, 35], and recently improved byCarrié. Actually, this new stochastic model enableseither to stochastically interpolate initial samplesto synthesize fast fluctuations of rain attenuationor to generate “on-demand” rain attenuation eventsstarting from the three parameters: the duration ofthe event, the maximum attenuation value and thetime position of the peak attenuation. The modelimprovement lies in the characterization of theconditional probability p(A(t)|A(t −Dt1), A(t +Dt2))which enables very fast interpolation or stochasticgeneration of rain events “on-demand”. Three of theinput parameters (m, s, Aof f set) can be assessed for alllink configurations using ITU-R recommended models.For the last parameter, b, a rough estimate equal to 10-4 s-1 can be used independently of the sampling timeand whatever the link considered in the North-westernEurope for elevation angles between 25◦ and 38◦ andfrequencies between 12 GHz and 50 GHz [36].

The Juzzle core component. The Juzzle core componentis composed of 5 different nodes (Gateway, Satellite,GatewaySiteDiversitySlave, User terminal 1, and Userterminal 2) as shown in Figure 3 hereafter:

This core processing module was developed in Javadue to the portability, reusability, and object-orientedfeature of the programming language. The moduleuses the Java event engine allowing to model thevarious delays within the system more easily. Themodule models the emergency satellite system at highlevel as shown in Figure 6, with the following nodes:two user terminals (“User terminals 1 and 2”), themaster gateway, the slave gateway (to be switched to incase of strong impairments above a defined thresholdwhen site diversity (SD) is activated) and the satellite,especially in terms of radio packet transmission, andits related link budget computation functions. Asa result, the simulator is a powerful and complexintegrated simulation platform handling more than 700parameters, and being composed of 44 Java classes.

Traffic modules. Four types of traffic are modeled: VoIP,Web browsing, and SMS for the return link, andInternet aggregate traffic as an example of continuous

forward link traffic. The underlying theoretical modelsused are:

1. for VoIP, classic ON/ OFF state machine whichoperates as an alternating process of talk andsilence periods which are distributed according tonegative exponential laws with means a-1 and b-1respectively;

2. for Web browsing traffic, the model implementedin the simulator is that of Choi and Limb [32];

3. SMS traffic is simply modeled by a generatorof SMS messages whose mode of operationis as follows [4, 38]: A random number Nfrom 1 to 10 (produced by a uniform randomnumber generator) of fixed 140-byte (amountingto 160 alphanumeric characters) concatenatedSMS messages is generated every period TSMSwhere TSMS is generated according to a negativeexponential distribution.

4. the IP aggregate traffic source is modeled bya Fractional Gaussian Noise (FGN) using theFast Fourier Transform (FFT) and Paxson’s FFTalgorithm. Paxson’s FFT method [33] consists insynthesizing a sample path having a same powerspectral density (PSD) as an FGN process. Thissample path can then be used in simulations astraces of real self-similar traffic. The algorithmrelies on an implementation of Paxon’s self-similar generator written in ANSI C, and providedwith in the form of a routine fft_fgn.c developedby Christian Schuler [33].

Adaptive strategy module. The adaptive strategy mech-anism is at the heart of the simulator, and relies on acombination of several techniques aimed at ensuring ahigh availability of the transmission links in spite ofsevere channel impairments in the selected frequencybands (eg. about 20 dB in Ka band and more than80 dB in Q band 0.01 % of the time). The techniquesemployed are: ACM, spread spectrum with varying DS-SS spread factor (L), gateway site diversity (SD) (inorder to improve the downlink budget for the returnchannel when it is impaired by rain), and ARQ-like(Automatic Repeat Request) time diversity (TD). ARQis only applied to SMSs, the user terminal automati-cally attempting at retransmitting the same SMS mes-sage at different times, until the channel conditionsimprove. Currently, retransmission is done on a puredeterministic basis, that is, periodically after a fixedtime interval. Nevertheless, more elaborate strategiescould be devised exploiting fade duration, fade slope,and inter-fade statistics so as to implement an efficientmethod able to predict the channel attenuation in themedium term (several minutes). For UTC and UTD-typed terminals, uplink power control (ULPC) is also

12

EAI for InnovationEuropean Alliance


09 - 12 2014 | Volume 2| Issue 5| e1


used. It is moreover assumed that all of the user ter-minals of the emergency mission which are deployedover the disaster geographical area are of the sametype and undergo the same ACM mode. This makessense since the area is assumed to be quite limited inextent (less than 100 km2 which is the size of a greatcity such as Paris in France), which allows to considerthat the link budget and transmission parameters areroughly constant over the whole area. Nevertheless, itmust be kept in mind that this uniform ACM schemeover the whole area is also a consequence of a sim-plistic assumption concerning the channel propagationmodelling, in which no spatial variability is taken intoaccount. Only a temporal variability is modeled usingthe Carrié’s enhancement of Lacoste’s CNES/ONERArain fading time series stochastic generator based onMaseng-Bakken model [31, 34–37].

The adaptive strategy combines ACM with spread-factor-varying SS, the spread factor L being adapted(increased) before ACM activation depending on theattenuation level with respect to predefined thresholds,and with three different service classes being defined:

1. Premium service: VoIP, and Web browsing/SMSwith a bit rate of about 128 kbps;

2. Gold service: VoIP/SMS with a bit rate of 4 to 16kbps;

3. Basic service: SMS only with a bit rate of less than4 kbps.

The DS-SS spread factor L can range from 1to up to several thousand provided the satellitebandwidth is not exceeded. Within each service (orprimary bit rate mode), the channel fluctuationsare dynamically compensated by full DVB-S2 ACMMODCOD switching.

5.4. Some resultsA number of results will be presented in this sectionnot in an exhaustive manner, but rather selectivelymerely in order to illustrate the ability of the integratedsimulation tool to perform relevant analyses of theperformance of the low bit rate satellite communicationlink proposed in the context of an emergency situationand disaster management. The following configurationswill be studied:

1 UTA and UTB terminals in Ku band.

2 UTD terminals in the most attenuated frequencyband, that is, V band.

Moreover, our analysis will mainly focus on the generalperformance of the adaptive strategy. The normalPLFRAME (64,800 bits).

Table 1. Return link performance report for UTA and UTBterminals in Ku bandeb

Parameters Terminal type ...UTA UTB

Maximum uplink attenuation (dB) 6.88 6.88Maximum downlink attenuation (dB) 4.11 4.11Time series duration (s) 10,800 10,800Used MODCODs 15 15Carrier bandwidth (KHz) 51.406 51.406

constant constantMaximum allowed bit rate (kbps) 128 128

constant constantDS-SS factor 1 constant 1 constantMaximum budget margin (dB) 18.03 13.93Minimum budget margin (dB) 12.22 8.123Mean budget margin (dB) 15.96 11.861C/(N0+I0)max (dB) 28.66 22.87C/(N0+I0)min (dB) 26.61 24.56C/(N0+I0)mean (dB) 18.77 22.51Mean 1-way IP packet delay (s) 1.03071 1.030711-way IP packet delay standard 0.14323 0.14323deviation (s)Total number of ATM cells sent 601,736 601,736by the terminal under studyTotal number of ATM cells 601,736 601,736correctly received by the gatewayFraction of ATM cells correctly 100 100received by the gateway (%)Transmitted throughput (kbps) 23.627 23.627Correctly received throughput (kbps) 23.627 23.627Mean spectral efficiency (bit/s/Hz) 2.49 2.49Minimum spectral efficiency (bit/s/Hz) 2.49 2.49Maximum spectral efficiency (bit/s/Hz) 2.49 2.49Fraction of VoIP traffic (%) 40.4 40.4Number of VoIP ATM cells sent 243,080 243,080Number of VoIP ATM cells correctly 243,080 243,080received by the gatewayVoIP throughput sent (kbps) 9.544 9.544VoIP throughput correctly 9.544 9.544received (kbps)Number of Web ATM cells sent 358,656 358,656Fraction of Web traffic (%) 59.6 % 59.6 %Number of Web ATM cells correctly 358,656 358,656received by the gatewayVoIP throughput sent (kbps) 14.082 14.082VoIP throughput correctly 14.082 14.082received by the gateway (kbps)Number of SMS ATM cells sent 0 0Fraction of SMS traffic (%) 0 0SMS throughput sent (kbps) 0 0SMS throughput correctly 0 0received by the gateway (kbps)

UTA and UTB terminals in Ku band. In Ku band, UTA andUTB terminals can be used and are entirely sufficientfor the defined purpose. Table summarizes the resultsobtained for the two terminal types using two 10,800-s channel propagation attenuation time series. Theuplink attenuation time series has a peak attenuationof 6.88 dB while the downlink one has a maximumattenuation of 4.11 dB.

Figure 7 and Figure 8 show the behaviour of theadaptive mechanism for UTA and UTB terminalsrespectively. It can be observed that for the two



09 - 12 2014 | Volume 2| Issue 5| e1


Figure 7. Adaptive strategy operation for UTA terminals in Kuband

types of terminal: The 6.88-dB attenuation is totallycompensated using a fixed MODCOD, namely 8 PSK8/9 characterized by a spectral efficiency of 2.49bit/s/Hz. The transmitted signal is not spread (DS-SSfactor of 1). Site diversity is not required. The premiumservice is guaranteed with a maximum allowed bit rateof 128 kbps, which gives birth to a constant carrierbandwidth of 51.406 KHz.

UTD terminals in V band. In V band, UTA and UTBterminals cannot be used anymore. Larger terminals arerequired. Since UTC is known to be overdimensioned,only UTD needs to be examined here. Table summarizesthe performance results obtained in a simulation withUTD terminals using two 14,160-s channel propagationattenuation time series. The uplink attenuation timeseries has a peak attenuation of 84.74 dB while thedownlink one has a maximum attenuation of 37.51 dB.It was also assumed that only a subset of the 28 DVB-S2

Figure 8. Adaptive strategy operation for UTB terminals in Kuband

MODCODs was used, more precisely, MODCODs 1 to23. In other words, 32 APSK modulation was excluded.

It is worth noting that SD is never triggered,since the downlink attenuation never reaches theSD activation threshold of 64 dB. Only the basicservice (less than 4 kbps) can be guaranteed, due tothe strong channel impairments encountered and thecomparatively moderate transmission power (less than5 W) of UTD terminals. Very strong uplink attenuationvalues between 4,000 and 5,500 s lead to using themost robust MODCOD available, that is, 16 APSK 9/10(MODCOD 23). However, even that is not sufficient toentirely compensate for the impairment, which givesbirth to link outage periods. Hard activation of DS-SScan be noticed with spread factors yielding a value ofmore than 700.

6. ConclusionIn this paper a reflection on some issues relatedto the use of hybrid wireless/satellite links in the



09 - 12 2014 | Volume 2| Issue 5| e1


Table 2. Return link performance report for UTD terminals in Vband

Parameters ValueMaximum uplink attenuation (dB) 84.74Maximum downlink attenuation (dB) 37.51Time series duration (s) 14,160Most used MODCODs QPSK1/4 (92.1 %),

QPSK 1/3 (536 %),QPSK 3/5 (1.1 %),QPSK 1/2 (0.46 %)

Maximum carrier bandwidth (KHz) 6,360Minimum carrier bandwidth (KHz) 396.429Mean carrier bandwidth (KHz) 2,257.678Maximum allowed bit rate (kbps) 4Minimum allowed bit rate (kbps) 4Mean allowed bit rate (kbps) 4Maximum budget margin (dB) 51.09Minimum budget margin (dB) -18.75Mean budget margin (dB) 33.25C/(N0+I0)max (dB) 24.98C/(N0+I0)min (dB) -50C/(N0+I0)mean (dB) 6.91Mean 1-way IP packet delay (s) 1.2581-way IP packet delay standard 0.976deviation (s)Total number of ATM cells sent 1,261,175by the terminal under studyTotal number of ATM cells 1,193,051correctly received by the gatewayFraction of ATM cells 94.6correctly received by the gateway (%)Transmitted throughput (kbps) 37.764Correctly received throughput (kbps) 35.725Mean spectral efficiency (bit/s/Hz) 0.53Minimum spectral efficiency (bit/s/Hz) 0.5Maximum spectral efficiency (bit/s/Hz) 3.58

field of telemedicine was conducted. At the startingpoint of our reflection, was our interest in theanalysis of the performance of a combined PG-ARsignal reconstruction algorithm we proposed to applyin order to remedy the problem of missing ECGsamples at the user-end side. We started from thecrude observation that along the transmission chainthere were data errors and/or loss that could occuranywhere and anytime, and that was our motivationto investigate more thoroughly the multiple causesof such errors and loss from a pure network andtelecommunication point of view. First, the context waspresented with telemedicine projects (U-R-SAFE andOURSES) which enabled us to highlight some issuesstill open with respect to performance criteria havingsignificant implications in health-care applications,in terms of, for example, packet error rate, energyconsumption at the nodes, bandwidth occupation,etc. Furthermore, due to the multifold impact ofthe telecommunications and networking issues onbiomedical signals, a special emphasis was laid ondesign constraints in a wireless network architecture,then on the satellite channel itself which can bestrongly impaired in high frequency bands, and causes

Figure 9. Adaptive strategy operation for UTD terminals in Vband

errors of a particular kind. QoS issues such as thatof performance requirements as to the transmitted IP-based traffic usable in telemedicine applications, andquite significant issues related to the TCP (TransmissionControl Protocol) protocol over the satellite link werealso surveyed. Indeed, strong propagation channelimpairments requiring efficient adaptive strategies,transmission errors occurring in bursts, and highlatency due to the geostationary Round Time Trip (RTT)are the three main drawbacks of the satellite path inhigh frequency bands. These factors combine togethermaking data errors or packet loss very likely, which weknow to be quite critical in telemedicine applications,in that human lives closely depend on high qualitybiomedical signal reception, and correct diagnoses. This



09 - 12 2014 | Volume 2| Issue 5| e1


Figure 10. (a) Histogram of DS-SS factor values for UTDterminals in V band (left). (b) Histogram of MODCODs for UTDterminals in V band (right).

paper only explored a few well known networkingand telecommunication issues from a qualitative pointof view for telemedicine applications. The complexconnections between the two worlds of telemedicineon one side, and networking and telecommunicationson the other side still remain to be investigated inmore details from a quantitative perspective. This willbe carried out in a forthcoming paper. Last but notleast, an enhanced DVB-S2-like air interface suited totelemedicine applications in an emergency situationhas been described. It involves DS-SS technique andother adaptive mechanisms such power control, andsite diversity for a multibeam, bent-pipe satellitesystem. It was designed to rapidly establish a lowbit rate link for emergency communications in Ku/Kaand Q/V bands. Link budget analyses have shownthat even though SS may be deactivated, very lowtransmit power user terminals UTA and UTB yieldreasonable performance for the envisaged purposeup to the Q band, but for rather moderate channelattenuation values. This highlights the relevance andefficiency of a solution combining ACM, SS, bitreduction, and SD techniques in a new adaptive strategyin order to improve the transmission performance.Such an adaptive strategy has been successfullyimplemented within the more general framework of

an integrated Excel/Juzzle/Matlab software simulationplatform designed to model the system in a highlevel cross-layer approach mixing propagation, physicallayer, and higher layers components. Some selectiveresults of the simulations carried out using this DVB-S2-like satellite link software simulator have beenpartially presented highlighting its ability to performrelevant analyses of the performance of the system bothat the physical layer and the network level.

References[1] Anderson T.J, TCP/IP over Satellite: Optimization vs.

Acceleration, End II End Communications, Inc. WhitePaper. April 2005.

[2] Durresi A, Kota S,Goyal M,Jain R and Bharani V,Achieving QoS for TCP traffic in Satellite Networks withDifferentiated Services, Journal of Space Communications,Volume 17, Issue 1-3, April 2001, pp.125-136.

[3] Brandao J.C, Pinto E.L, and Maia M.A.G, A Review ofError Performance Models for Satellite ATM Networks, IEEECommunications Magazine, Broadband Satellite NetworkPerformance, vol. 37 no. 7, pp.80-85, July 1999.

[4] ETSI, Digital cellular telecommunications system (Phase2+); Universal mobile telecommunications system (UMTS);Technical realization of short message service (SMS) (3GPPTS 23.040 version 5.8.1 Release 5), October 2004.

[5] ETSI, TR 102 444 V1.1.1: Analysis of the short messageservice (SMS) and cell broadcast service (CBS) for emergencymessaging applications; emergency messaging; SMS and CBS,February 2006.

[6] ETSI, EN 301 790 V1.5.1: Digital Video Broadcasting(DVB); Interaction channel for satellite distribution systems,May 2009.

[7] Hashimoto A, Outage Probability Analysis in RelocatableWireless Access Systems under Line-of-Sight Non-RayleighFading, IEICE vol.E80-B no.5, pp.746-754, May 1997.

[8] Hashimoto A, Error Performance and ATM Cell TransferCharacteristics in Relocatable Wireless Access Systems,Proceedings of the IEICE, vol.E81-B, no.6, pp.1213-1223,June 1998.

[9] Henderson T. R. and Katz R. H., Transport Protocolsfor Internet-Compatible Satellite Networks, IEEE Journal onSelected Areas of Communications, 1999.

[10] Inigo P, Durin B, Girault N and Verelst G, OURSES:Medical assistance over Satellite, IWSSC 2008.

[11] Jegham N, Girault N, Le Guern C, Roussel G, Lohier S

and Beylot A-L, VoIP over a DVB-S2 ACM link, IWSSC2008.

[12] ITU-T, G.114 Recommendation: Transmission systemsand media, general recommendations on the transmissionquality for the entire Internet telephone connection: One-waytransmission time, number 114 in G. ITU-T, 2000.

[13] Ganesan D, Govindan R, Shenker S, Estrin D, Highly-resilient, energy efficient multipath routing in wireless sensornetworks. Mobile Computing and Communication Review,1(2):28-36, 2002.

[14] Girault N, Jegham N, Lerouge N, Le Guern C, andBeylot A-L, OURSES: Efficiency of IP Encapsulation overDVB-S2 Links, IWSSC 2008.



09 - 12 2014 | Volume 2| Issue 5| e1


[15] Kacimi R, Dhaou R, and Beylot A-L, Load balancingtechniques for lifetime maximizing in wireless sensornetworks. Ad Hoc Networks 11(8): 2172-2186 2013.

[16] Kota S and Marchese M, Quality of service forsatellite IP networks: a survey, International Journal ofSatellite Communications and Networking 2003; 21:303-349 (DOI: 10.1002/sat.765).

[17] Mailhes C, Castanié F, Henrion S, Lareng L, andAlonso A. The URSAFE telemedicine project: improvinghealth care of the elderly. European Federation for MedicalInformatics (MIE’03), Saint Malo, France, 2003.

[18] Mailhes C, Prieto-Guerrero A, Comet B, De

Bernard H, and Campo E. Telemedicine Applicationsin OURSES Project, International Workshop on Satelliteand Space Communications (IWSSC’08), Toulouse,France, 2008.

[19] Pech P. Incidence de la prise en compte des effets detechniques de mécanismes de lutte contre les affaiblissements(FMT) en bande Ka sur la gestion des ressources dansun système d’accès multimedia par satellite géostationnaire,Ph.D dissertation, Supaero, Toulouse, France, december19, 2003.

[20] Pech P, Huang P, Bousquet M, Robert M, andDuverdier A. Simulation of an Adaptive Strategy Designedfor Low Bit Rate Emergency Satellite Communications Linksin Ku/Ka/Q/V Bands, International Workshop on Satelliteand Space Communications 2009 (IWSSC 2009), 10-11September 2009, Siena-Tuscany, Italy.

[21] Pech P, Huang P and Bousquet M. Low Bit RateSatellite Link for Emergency Communications, InternationalWorkshop on Satellite and Space Communications 2008(IWSSC 2008), Toulouse, France, 1-3 October 2008.

[22] Pech P, Robert M, Duverdier A, and Bousquet M,Design and Simulation of a DVB-S2-like Adaptive Airinterface Designed for Low Bit Rate Emergency Commu-nications Satellite Link in Ku/Ka/Q/V Bands, availablefrom: http://sciyo.com/articles/show/title/design-and-simulation-of-a-dvb-s2-like-adaptive-air-interface-designed-for-low-bit-rate-emergency-co. Chapter 9 ofSatellite Communications, edited by Nazzareno Diodato,Sciyo, Rijeka, Croatia, September 2010.

[23] Pech P, Liaison bas débit par satellite pour les situationsd’urgence. Study report for work package 2, release 2.1,TeSA/CNES, 8 June 2009.

[24] Official Juzzle Web site: http://www.juzzle.org/.[25] ETSI. EN 302 307 V1.2.1: Digital video broadcasting

(DVB); Second generation framing structure, channel codingand modulation systems for broadcasting, interactive services,news gathering and other broadband satellite applications,April 2009.

[26] ETSI. ETSI TR 102 376 V1.1.1: Digital video broadcasting(DVB); User guidelines for the second generation system forbroadcasting, interactive services, news gathering and otherbroadband satellite applications (DVB-S2), February 2002.

[27] Cantillo J. Codage multi-couches pour systèmes decommunication par satellites, Ph.D thesis report, TélécomParis, Toulouse, France, 19 May 2008.

[28] Ayala M.R, Gonzales E.A.A, and Franco J.J.R. Perfor-mance in QPSK and BPSK synchronous sequence DS-CDMAsatellite systems, 1st International Conference on Electricaland Electronics Engineering (ICEEE), Iaghouat, Algeria,

24-27 June 2004.[29] Yoon J-H, Lee J-K, Chang D-I, and Oh D-J. Spread

Spectrum Technique for Digital Broadcasting System inRain Environment, 10th International Conference onAdvanced Communication Technology (ICACT), PhoenixPark, South Korea, 17-20 February 2008.

[30] Michels H, DISLIN 9.4, A Data PlottingLibrary, 15 October 2008. available from:http://www.mps.mpg.de/dislin/document.html.

[31] Lacoste F. Modelling of the dynamics of the Earth-spacepropagation channel at Ka and EHF bands, Ph.D thesisreport, SUPAERO, Toulouse, France, September 2005.

[32] Klemm A, Lindemann C, and Lohmann M. Traffic Model-ing and Characterization for UMTS Networks. Proceedingsof the Globecom, Internet Performance Symposium, SanAntonio TX, United States, November 2001.

[33] Paxson V. Fast, Approximate Synthesis of FractionalGaussian Noise for Generating Self-Similar Network Traffic,Computer Communication Review, October 1997; 27(5),pp.5-18.

[34] Lacoste F, Bousquet M, Castanet L, Cornet F,Lemorton J. Improvement of the ONERA-CNES rainattenuation time series synthesiser and validation of thedynamic characteristics of the generated fade events, SpaceCommunication Journal 2005; 20(1-2).

[35] Castanet L (editor) et al. Influence of the variability of thepropagation channel on mobile, fixed multimedia and opticalsatellite communications, SatNEx JA-2310 book, ISBN 978-3-8322-6904-3, Shaker, 2008.

[36] Aroumont A, Alamanac A.B, Castanet L, Bousquet M,Cioni S, and Corazza G.E. Performance of channel qualityestimation algorithms for fade mitigation techniques withKa/Q/V band satellite systems, 9th International Workshopon Signal Processing for Space Communication, Noord-wijk, Netherlands, September 2006.

[37] Lemorton J, Castanet L, Lacoste F, Riva C, Matric-

ciani E, Fiebig U.C, Van de Kamp M, and Martellucci A,Development and validation of time-series synthesizers ofrain attenuation for Ka-band and Q/V-band satellite commu-nication systems, International Journal of Satellite Com-munications and Networking, 2007; 25:575-601.

[38] ETSI. Digital cellular telecommunications system (Phase2+); Universal Mobile Telecommunications System (UMTS);Alphabets and language-specific information (3GPP TS23.038 v6.1.0), September 2004.

[39] IEEE Standard For Information Technology Part 15.4:Wireless LAN medium access control (MAC) and physicallayer (PHY) specification for low rate wireless personal areanetworks (LR-WPANs), 2006.

[40] Tolle G and Culler D. Design of an application-cooperative management system for wireless sensor networks.In Proceedings of the 2nd European Workshop onWireless Sensor Networks (EWSN’05), pages 121_132,Istanbul, Turkey, February 2005.

[41] Srinivasan K and Levis P. RSSI is under appreciated.In Proceedings of the 3rd Workshop on EmbeddedNetworked Sensors (EmNets’06), Cambridge, MA, USA,2006.

[42] Yang G Z, Body Sensor Networks, Springer-VerlagLondon, 2006.



09 - 12 2014 | Volume 2| Issue 5| e1


[43] Prieto-Guerrero A, Mailhes C and Castanié F, LostSample Recovering of ECG Signals in e-Health Applications,Proceedings of the 29th Annual International, Conference

of the IEEE EMBS, Cité Internationale, Lyon, France,August 23-26, 2007.



09 - 12 2014 | Volume 2| Issue 5| e1

Hybrid Satellite - Wireless Sensor Networks Architecture for ...

Documents