Research Article A Novel Energy-Efficient Clustering Based ...downloads.hindawi.com/journals/ijdsn/2015/198456.pdf · assignment policy for cognitive-radio networks, in which all

Research ArticleA Novel Energy-Efficient Clustering Based CooperativeSpectrum Sensing for Cognitive Radio Sensor Networks

Ashish Rauniyar and Soo Young Shin

Wireless and Emerging Network System Lab, School of Electronics Engineering, Kumoh National Institute of Technology,Gumi 730-701, Republic of Korea

Correspondence should be addressed to Soo Young Shin; [email protected]

Received 9 February 2015; Revised 2 June 2015; Accepted 4 June 2015

Academic Editor: Shaojie Tang

Copyright © 2015 A. Rauniyar and S. Y. Shin. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Cognitive radio has been proposed as a promising way to effectively utilize the scarce spectrum resources. A cognitive radio sensornetwork (CRSN) is a wireless sensor network that is equipped with cognitive radio capability. Clustering is a popular techniquethat can be applied to wireless sensor networks, although it has been proven to be a challenge to implement it in CRSNs. Moreover,few proposals have successfully applied energy-efficient clustering techniques in CRSNs. Therefore, with the aim of increasingenergy efficiency, network lifetime, network stability, and optimal cluster-head selection process, this paper proposes a novel energy-efficient clustering based on cooperative spectrum sensing (ECS) for CRSNs. The proposed ECS scheme utilizes the concept ofpairing among sensor nodes and switches between Awake and Sleep modes for energy efficiency. A comprehensive simulation inMATLABwas carried out to validate the proposedmethod.The simulation results show that, comparedwith conventionalmethods,the proposed method is more energy efficient and that the overall CRSN’s lifetime is prolonged.

1. Introduction

A wireless sensor network is a network of densely deployeddistributed sensors that monitor physical or environmentalconditions such as temperature, sound, and pressure, inorder to cooperatively transfer data through the network toa sink node or base station [1]. Wireless sensor networks arecomposed of multiple limited-power sensor nodes delpoyedrandomly in a region of interest. Clustering is a populartechnique applied in such networks, where nearby nodesform a group, known as “cluster” [2].

Because of the explosive growth in wireless device overthe past few years, spectral congestion and inefficient spec-trum usage have been a pressing concern. In a survey con-ducted by the Federal Communication Commission (FCC)on spectrum utilization revealed that the actual licensedspectrum is largely underutilized, given its vast geographi-cal dimensions [3]. Cognitive radio provides opportunisticaccess to unused licensed bands. With cognitive radio, unli-censed secondary users can use licensed frequencies whenthe primary user is inactive [4, 5]. In practical applications,

the received signal at each cognitive user may suffer from theso-called hidden primary-terminal problem and uncertaintydue to fading and shadowing. Thus, cooperative spectrumsensing has become a popular technique to address suchissues, rendering the spectrum usage more efficient andproviding a high level of protection to the primary user [6–8].

Cognitive radio sensor networks (CRSNs) are a smartcombination of wireless sensor network and a cognitiveradio, these have recently attracted increased attention [9].Clustering algorithms for wireless sensor networks havebeen reasonably successful at improving the performance ofnetworks. Clustering refers to the task of grouping a set ofsensing nodes in such a way that the nodes form a group(called a “cluster”).The nodes in a cluster aremore similar (insome sense or another) to each other than to those in otherclusters. Each cluster consists of one cluster head.The cluster’smembers sense the attributes of the target environment andsend this sensor data to the fusion center or base station. In aCRSN, a cluster head is mainly responsible for all spectrum-management tasks, such as acquiring the sensor data from thenodes in the cluster and forwarding it to the fusion center

Hindawi Publishing CorporationInternational Journal of Distributed Sensor NetworksVolume 2015, Article ID 198456, 8 pageshttp://dx.doi.org/10.1155/2015/198456

2 International Journal of Distributed Sensor Networks

in order to reduce the overall energy consumed. However,clustering in a CRSN cannot be applied directly, as it can in awireless sensor network. Clustering in a CSRN is particularlydelicate. For instance, in order to make clustering possiblein a CSRN, all of the sensor nodes in the cluster must belocated within the transmission range of the others, andthey must share a common communication channel amongthem.

Furthermore, clustering schemes for wireless sensornetworks are designed with the aim of accumulating tar-get information with minimized energy consumption [10].However, they cannot deal with the spectrum-aware sens-ing and communication in the context of cognitive radio.Clustering methods for CRSNs are relevant insofar as theydirectly affect the energy consumption and the lifetime ofthe network. Clustering algorithms, such as the low-energyadaptive clustering hierarchy (LEACH) method and thedistributed energy-efficient clustering (DEEC) method, haveachieved a reasonable degree of success at optimizing theperformance of a network. In [11], an energy-efficient LEACHprotocol is proposed, whereby cluster heads are selectedwith predetermined probability and energy drain, and thenother nodes join their nearest cluster heads. With a DEECprotocol, nodes are independently elected as cluster headsbased on the initial energy and residual energy [12]. Thenodes with high initial and residual energy are more likelyto be cluster heads than nodes with low energy, under aDEEC protocol. Reference [13] proposed a stable electionprotocol (SEP) protocol to prolong the time interval beforethe death of the first node in a heterogenous network. Withtheir proposal, advanced and normal nodes are distinguishedwhen electing the cluster head. Another protocol, calledthe hybrid energy-efficient distributed (HEED) protocol, isproposed in [14].This proposal involves choosing a nodewithmore residual energy and more neighboring nodes as thecluster head through coordinated election. However, all ofthese clustering schemes assume a fixed-channel allocationand cannot handle dynamic spectrumaccess.Therefore, theseschemes are not suitable for CRSNs.

A learning-inspired and dynamic channel decision andaccess technique for cognitive-radio-based wireless sensornetworks (CRWSNs) is proposed in [15].With their proposedmethod, the CRWSN agent decides and accesses any avail-able channel, based on previous information regarding theenergy-consumption rate and the energy efficiency achiev-able by the CRWSN. Reference [16] investigated dynamic-spectrum access issues for multichannel CRSNs. Here, theprimary user’s behavior is modeled as a two-state Markovchain, and its transition probabilities are estimated using amaximum-likelihood estimation. They also employed vary-ing packet size adaptation techniques to adapt the transmis-sion over the state-varying channel, and this, in turn, moreefficiently utilizes the battery life of the sensors. A novelquality-of-service-aware spectrum-access algorithm is pro-posed in [17], with power allocation for both noncooperativeand cooperative users to maximize spectrum utilization andminimize power consumption. An efficient medium-access

control (MAC) protocol with selective grouping and coop-erative sensing in cognitive-radio networks called group-based cooperative MAC (GC-MAC) is proposed in [18].The GC-MAC scheme can quickly discover the spectrumopportunities without degrading the accuracy of the sensors.They also proposed an algorithm for selecting the secondaryuser, specifically choosing cooperative secondary users basedon channel dynamics and usage patterns, in order to reducethe sensor overhead in both time-invariant and time-varyingchannels. Reference [19] proposes a cluster-based cooperativespectrum-sensing strategy for obtaining a suitable spectrum-assignment policy for cognitive-radio networks, in which allthe cluster members are cooperative in sensing the samechannels. The cooperative spectrum-sensing problem in acluster was further formulated as a maximum weight one-sided biclique problem, and a greedy heuristic algorithmwas used to find the appropriate spectrum-assignment pol-icy.

A distributed spectrum-aware clustering (DSAC) is pro-posed in [20, 21].This scheme uses constrained clustering [22,23] to cluster CRSN nodes under spectrum-aware constraint.The aim is to merge two nearby nodes or clusters thatshare the same available channel, or where at least onechannel is common between the two nodes or clusters forcommunication. Whether two nodes or clusters are mergeddepends on the local minimum distance obtained fromthe information exchanged between two nearby nodes orclusters. The cluster-formation process is repeated until anoptimal number of clusters is reached.

Extending the idea of spectrum-aware constrained clus-tering as proposed in [20, 21], we propose a novel energy-efficient clustering scheme based on cooperative spectrumsensing (ECS) for CRSNs. In our proposed method, a pair ofnodes that are grouped together can switch between Awakeand Sleepmodes such that the sensing process is more energyefficient. Moreover, the proposed method operates in a self-organizedmanner, extending the network’s lifetime, and withmore stability and an optimal cluster-head selection process.

The principal contributions of this paper are outlined asfollows.

(i) We developed a clustering method suitable for CRSNwith low energy consumption, increasing the lifetimeand stability of a CRSN network.

(ii) Moreover, we develop a novel approach to clusteringby introducing spectrum-aware node-grouping suit-able for CRSNs, such that the pair of nodes in a groupcan alternate between Sleep and Awakemodes duringthe sensing process to increase the energy efficiency ofa CRSN network.

(iii) We develop an algorithm to configure the Awake andSleep modes for coupled CRSN nodes.

The rest of the paper is organized as follows. Section 2describes our proposed method along with its distinct fea-tures. A performance evaluation of the proposed methodis discussed in Section 3. Finally, conclusions are drawn inSection 4.

International Journal of Distributed Sensor Networks 3

2. Energy-Efficient Clustering-BasedCooperative Spectrum Sensing (ECS)Scheme

In this section, we present our energy-efficient clustering-based spectrum-sensing scheme (hereafter, the “ECSscheme”). Below, a “node” refers to a CRSN node. In theproposed ECS scheme, and the following assumptions andobjectives are used.

(i) Each node is equipped with cognitive-radio capabil-ities, such as spectrum sensing, dynamic spectrumaccess, and transmission-parameter configurability.Moreover, each node is aware of the radio environ-ment.

(ii) The deployment of nodes is random, dense, andredundant.

(iii) Nodes that belong to the same cluster should haveat least one common available channel that is notcurrently occupied by a neighboring primary user.

(iv) Each node is randomly deployed in a region, and itslocation is determined with a global positioning sys-tem (GPS) that transmits its location, along with theapplication type and the node-identification number,to the fusion center or base station.

(v) The MAC protocol is based on a time-divisionmultiple-access (TDMA) method.

(vi) Efficient application-oriented source sensing. Thereis a cluster head for every cluster in the CRSN. Thesource information sensed by different cooperatingnodes in a cluster should first be aggregated in thecluster head and then relayed to the fusion center.

2.1. Spectrum-Aware Pairwise Coupling. In the proposed ECSscheme, spectrum-aware pairwise coupling is somewhatsimilar to distributed spectrum-aware clustering (DSAC)scheme proposed in [20]. With DSAC, the local minimumdistance between a pair of nodes is exchanged throughneighborhood-information interchange, merging the closestlocal pair. Moreover, with the DSAC scheme, both nodes in apair are active for sensing the target, and they communicatewith the cluster head, adding to the energy consumed. Ourmain goal is to minimize the energy used by the CSRN toenhance the stability and the lifetime of the network.

As shown in Figure 1, Nodes A and B can be pairedtogether, provided that they have at least one channel incommon. The numbers marked beside Nodes A and Brepresent the available channels. As seen in the middle ofFigure 1, Nodes A and B can be coupled together, becausethey have Channel 1 in common. However, on the right sideof Figure 1, Nodes A and B do not have a channel in common.Hence, they cannot be coupled together.

For our ECS scheme, we assume that the fusion centeror base station has received the location information, appli-cation type, and node-identification numbers for all of thenodes deployed in the region of interest. This informationis used by the fusion center to compute the mutual distancebetween nodes. CRSN nodes of the same application type

A

B

A B

A B

A

B

Yes

No

Can link

Cannot link

[1, 2] [1, 2]

[1, 3] [3, 4]

Figure 1: Spectrum-aware pairwise coupling.

0 10 20 30 40 50 60 70 80 90 1000

10

20

30

40

50

60

70

80

90

100

FC

CH CRSNnode

CoupledCRSN node

Leftout CRSNnode duringcoupling

Coupled CRSNnode

Figure 2: Example of spectrum-aware pairwise coupling of differentCRSN nodes.

that have at least one channel in common and are at aminimum distance from each other within their intraclustertransmission range are coupled together, as directed by thefusion center. This process is known as spectrum-awarepairwise coupling. The fusion center broadcasts the couplinginformation to all CRSN nodes in the network. Each nodeis thus aware of its pairing. Moreover, during the pairwise-coupling process, some nodes remain unpaired, because theydo not fall within the intracluster transmission range of anyother node. Figure 2 shows an example of spectrum-awarepairwise coupling with different CRSN nodes. The fusioncenter does the work of spectrum-aware pairwise couplingfor CRSN nodes. The cluster-head selection process andthe process by which nodes are clustered are explained inSection 2.2, below.

The coupled nodes in the proposed ECS scheme canswitch between Sleep and Awake modes during a singlecommunication interval for energy efficiency. Initially, a nodein a coupled node which has already been paired by fusioncenter switches into “Awake” mode, if its distance from thefusion center is less than the coupled node. The active nodein Awake mode senses the channel status and relays thisinformation to the cluster heads. The other node in the pairswitches to Sleep mode and neither senses the channel statusnor communicates with the cluster head. During the nextcommunication interval, the Awake node switches to Sleepmode and the other node becomes active (i.e., it switches


to Awake mode), sensing the channel and communicatingwith the cluster head. The reason for alternating betweenmodes in our proposedECS scheme is tominimize the energyconsumption. Nodes in Sleep mode do not require energy tosense the channel status and communicate with the clusterheads. Rather, they delegate these tasks to the other node.Thecluster-head selection process and the Awake-Sleep modesof each node-pair are, respectively, detailed in Sections 2.2and 2.3, below. The unpaired nodes remain in Awake modecontinuously, until their energy resources are depleted.

2.2. Clustering and Cluster-Head Selection. LEACH is a hier-archical protocol according to which most nodes transmitinformation to the cluster heads, and the cluster headsaggregate and compress this data before forwarding it to thebase station. Each node uses a stochastic algorithm at eachround to determine whether it will become a cluster headduring this round. Nodes that have already been cluster headscannot become cluster heads again for 𝑃 rounds, where 𝑃

is the desired percentage of cluster heads. Thereafter, eachnode has a 1/𝑃 probability of becoming a cluster head ineach round. At the end of each round, each node that is nota cluster head selects the closest cluster head and joins thatcluster.The cluster head then creates a schedule for each nodein its cluster to transmit its data. During the process of datacollection and transmission, the energy consumed by datatransmission is greater than that of data fusion [24]. If thecurrent energy of a cluster head is less than that of othernodes, the cluster head will die quickly because of a heavyenergy burden.

In our ECS scheme, the cluster heads selected after thefirst round are determined based on the remaining energyof each node. Furthermore, only nodes in Awake modeengage in the cluster-head selection process. This reducesthe communication costs. In the first round, when all nodeshave an initial energy level 𝐸0, nodes that are active will electthemselves as cluster heads on the basis of the probability ofselecting a cluster head using a distributed algorithm. EachAwake node randomly picks a number between 0 and 1 andcompares this number to a threshold value 𝑇ℎ(𝑛), which isdetermined as follows:𝑇ℎ (𝑛)

=

{

{

{

𝑃

1 − 𝑃 ∗ ((first round) mod (1/𝑃))∀𝑛 ∈ 𝐴

0 otherwise,

(1)

where 𝐴 is the set of active nodes in the first round.If the random number picked by a CRSN node in Awake

mode is less than the threshold𝑇ℎ(𝑛), this nodewill elect itselfas a cluster head, that is, as the primary cluster head (PCH).When a node has been selected as the PCH, it broadcastsan advertisement message to the entire network. Only nodesin Awake mode receive these messages from different PCHs,and they select their PCH on the basis of the received signalstrength identification (RSSI) to form a cluster. When anAwake-mode CRSN node decides to associate itself with acluster and a PCH, it sends an association request to thePCH using a carrier-sense multiple-access with collision

avoidance (CSMA/CA)MAC protocol to avoid collision withthe association requests of other active nodes. In additionto the association request to their respective PCHs, Awakenodes also send information regarding their energy anddistance. The PCH in a cluster calculates the remainingenergy of each Awake node in the cluster and selects asecondary cluster head (SCH) that will act as the cluster headfor next round. It should be noted that the SCH is selectedon the basis of the remaining energy of the nodes. Whennodes have the same level of energy, the node closest to thePCH is selected as the SCH.Thus, the PCH, SCH, and clusterare formed for our ECS scheme. The procedural flow for thecluster formation and the cluster-head selection is shown inFigure 3.

2.3. Awake-Sleep Mode Configuration for Coupled CRSNNodes. After spectrum-aware pairwise coupling, clustering,and cluster-head selection, each node configures its respec-tiveAwake and Sleepmodes for the next round. Both nodes ina spectrum-aware coupled pair take turns switching betweenAwakemode and Sleepmode after each round.Nodes that areselected as the SCH by the PCH (i.e., to be the cluster head forthe next round) must be in Awake mode during the currentround. This situation gives rise to a conflict with the node towhich it is paired in terms of alternating between Awake andSleep modes. To resolve this issue, we develop an algorithmfor configuring Awake and Sleep modes for coupled nodes inour ECS scheme.

In Algorithm 1, the node first checks whether spectrum-aware coupling has been done. If the node is coupled, thenone of the nodes in the pair will check whether it is in awakemode andwhether it is flagged as the SCH for the next round.If the SCH next-round flag is “ON,” then the same node willbe in Awake mode during the next round and the other nodewill remain inactive. If the node is in Awake mode and itsSCH next-round flag is “OFF,” then the current node will gointo Sleep mode and other node will take its turn in Awakemode node for the next round.This procedure is repeated forall other Sleep-mode nodes.

2.4. Data Transmission and Reporting in a CRSN. Afterspectrum-aware pairwise coupling, clustering, and PCH andSCH selection, all active nodes send their sensor data to theirrespective cluster heads during their individual TDMA slots.It should be noted that nodes in Sleep mode do not takepart in data transmission under our proposed ECS schemeThe cluster heads aggregate the data from different nodes andsend it to the fusion center for the decision-making process.The various decision-making and data-fusion techniques falloutside the scope of this paper. In terms of data transmission,then, we are interested in successfully gathering data for thecluster heads and transmitting it to the fusion center. Dataaggregation is a key technique for compressing the amount ofdata and decreasing the energy consumption of the network.

If there are 𝑁 total nodes and 𝐾 is the optimal numberof cluster heads, then the average number of nodes in eachcluster will be

(𝑁

𝐾− 1) . (2)


Start

Round 1, AwakeCRSN nodes of

coupled pair elect themselves as CH

Randomly pick a number between 0

and 1 by awake CRSN node

Is random number selected

CRSNAwakenodes

becomePCH

Yes

PCHbroadcastsmessage to the wholenetwork

CRSN nodes sendassociation request to PCH based on

RSSI using CSMA/CAMAC protocol along

with remaining energy and its distance Info

PCHsacknowledge the

association request from CRSN nodes and allow them to

join and form a cluster

PCHcalculatesremaining

energy and distance of

other CRSN nodes

PCHsselect

SCH for next

round

Stop

No≤ threshold?

Figure 3: Procedural flow of cluster formation and cluster-head selection.

(1) 𝑖𝑓 (𝐶𝑅𝑆𝑁𝑁𝑜𝑑𝑒𝑠 == 𝑆𝑝𝑒𝑐𝑡𝑟𝑢𝑚 − 𝑎𝑤𝑎𝑟𝑒𝑃𝑎𝑖𝑟𝑤𝑖𝑠𝑒𝐶𝑜𝑢𝑝𝑙𝑖𝑛𝑔) 𝑡ℎ𝑒𝑛

(2) 𝑖𝑓 (𝐶𝑅𝑆𝑁𝑁𝑜𝑑𝑒 == 𝐴𝑤𝑎𝑘𝑒 &&𝑆𝐶𝐻 𝑁𝐸𝑋𝑇 𝑅𝑂𝑈𝑁𝐷 𝐹𝐿𝐴𝐺 == 1) 𝑡ℎ𝑒𝑛

Current CRSN Node Mode = Awake /∗ Keep Current CRSN Node inAwake Mode for Next Round ∗/

(3) 𝑒𝑙𝑠𝑒 𝑖𝑓 (𝐶𝑅𝑆𝑁𝑁𝑜𝑑𝑒 == 𝐴𝑤𝑎𝑘𝑒 &&𝑆𝐶𝐻 𝑁𝐸𝑋𝑇 𝑅𝑂𝑈𝑁𝐷 𝐹𝐿𝐴𝐺 == 0) 𝑡ℎ𝑒𝑛

Current CRSN Node Mode = Sleep /∗ Keep Current CRSN Node inSleep Mode for Next Round and Activate Other Coupled Sleep-mode Node toAwake Mode ∗/

(4) 𝑒𝑙𝑠𝑒CRSN Node Mode = Awake

(5) 𝑒𝑛𝑑 𝑖𝑓

(6) 𝑒𝑛𝑑 𝑖𝑓

(7) Repeat the Same Procedure for Other Sleep-mode Nodes Taking Turns in AwakeMode During the Next Round

Algorithm 1: Awake-Sleep mode setup for coupled CRSN nodes.

The energy consumed during data transmission andreporting under our ECS scheme can be divided as follows:

(i) data transmission by the node to the cluster head,

(ii) data received by the cluster head from the node,

(iii) aggregate data by the cluster head,

(iv) aggregated data transmission to the fusion center.

During the data transmission of 𝐷𝑐-bit message from anode to the cluster head, each node dissipates 𝐸𝑇 energy torun the transmission circuitry and 𝐸amp for the transmissionamplifier to reach an acceptable signal-to-noise ratio (SNR)for transmitting the same 𝐷𝑐-bit message. If the distancebetween the node and the cluster head is 𝑑2toCH, the nodeexpends 𝐸Node energy. This is derived as follows:

𝐸Node = (𝑁

𝐾− 1) (𝐸𝑇 ∗ 𝐷𝑐 ∗ 𝐸amp ∗ 𝐷𝑐 ∗ 𝑑

2toCH) . (3)


If 𝐸𝑅 is the energy dissipated by the receiver circuitry ofthe cluster head for receiving the data from other nodes, thenthe energy expenditure 𝐸Rec from the cluster head is given asfollows:

𝐸Rec = (𝐸𝑅 ∗ 𝐷𝑐) (𝑁

𝐾− 1) . (4)

If 𝐸AD is the energy expenditure that a cluster head needsto aggregate the data𝐷𝑐 from each associated node, then thetotal energy dissipated by the cluster head in aggregating thedata received from the associated nodes is

𝐸Agg𝐸 = (𝐸AD ∗ 𝐷𝑐) (𝑁

𝐾) . (5)

If 𝐷𝐴 is the aggregated data and 𝑑2toFC is the distance

from the cluster head to the fusion center, then the totaltransmission energy 𝐸Total dissipated by the cluster head intransmitting the aggregated data to the fusion center can becalculated as follows:

𝐸Total = 𝐸𝑇 ∗ 𝐷𝐴 ∗ 𝐸amp ∗ 𝐷𝐴 ∗ 𝑑2toFC. (6)

From the above equations, the total energy 𝐸TCH dissi-pated by the cluster head each round comes in the form ofenergy dissipated in receiving the data from associated nodes,aggregating the data, and transmitting the aggregated data tothe fusion center.Therefore, the total energy dissipated by thecluster head each round can be calculated as follows:

𝐸TCH = 𝐸Rec +𝐸Agg𝐸 +𝐸Total. (7)

3. Performance Evaluation of the ProposedECS Scheme

In this section, we analyze the performance of the proposedECS scheme in terms of its stability period, network lifetime,instability period, the number of cluster heads selected, andthe number of packets sent to the fusion center. The metricsused to evaluate the ECS scheme are defined as follows:

(i) stability period: duration of the CRSN’s performancefrom its initialization until the first node dies out,

(ii) network lifetime: duration of the CRSN’s perfor-mance from its initialization until the last node isalive,

(iii) instability period: duration of the CRSN’s perfor-mance from when the first node dies until the lastnode dies,

(iv) the number of cluster heads: the number of clusterheads generated each round,

(v) packet to fusion center (FC): the rate of successfuldata delivery to fusion center from cluster heads.

We used MATLAB to simulate the performance of theproposed scheme. In all simulations, we randomly deployed100 CRSN nodes in a 100∗ 100meter area with initial energyof 𝐸0. Without the loss of generality, the fusion center was

Table 1: Simulation parameters.

Parameter ValueNetwork size 100m ∗ 100mInitial energy 0.5 J𝑃𝑑 0.1Data aggregation energy cost 50 pj/bit jNumber of nodes 100Packet size 4000 bit𝐸𝑇 50 nJ/bit𝐸𝑅 50 nJ/bit𝐸amp 0.0013 pJ/bit/m4

0 1000 2000 3000 4000 50000

10

20

30

40

50

60

70

80

90

100

Number of rounds

Num

ber o

f dea

d CR

SN n

odes

Proposed ECSDEEC scheme

Figure 4: The number of dead CRSN nodes.

placed at (50, 50) in this area. The simulation parameters areprovided in Table 1 [25].

We compared our ECS scheme with the DEEC protocolproposed in [12]. Figure 4 provides a graph of the numberof dead CSRN nodes against the number of rounds. Figure 4shows that the ECS scheme has a longer stability period thanthe DEEC protocol. Under the ECS scheme, the first nodedied near Round 1,600, whereas the first node died nearRound 1,300 under the DEEC protocol. Furthermore, thenodes started dying more rapidly under the DEEC protocolafter Round 1,500, and this did not occur with our ECSscheme.This indicates that the ECS scheme provides a longernetwork lifetime than the DEEC. This is because the ECSscheme deactivates CRSN nodes in Sleep mode.

Figure 5 shows the number of alive CRSN nodes againstnumber of rounds. Initially all nodes are alive. In Figure 5, wecan see that, out of 100 alive nodes, the first node in the ECSscheme died around 1,600 and that subsequent nodes died ata constant rate. Under the DEEC protocol, on the other hand,there was a sudden increase in dead nodes after round 1,500.This shows that, with our proposed ECS scheme, unstableregions appear later than they do with the DEEC protocol.


0 1000 2000 3000 4000 50000

10

20

30

40

50

60

70

80

90

100

Number of rounds

Num

ber o

f aliv

e CRS

N n

odes


Figure 5: The number of alive CRSN nodes.

0 1000 2000 3000 4000 50000

5

10

15

20

25

30

35

40

45

Number of rounds

Num

ber o

f CH

gen

erat

ed p

er ro

und


Figure 6: The number of cluster heads selected each round.

Figure 6 shows the number of cluster heads selected eachround. The DEEC protocol resulted in more uncertainty inselecting cluster heads. A random number of cluster headsare selected every round under the DEEC protocol. However,with our ECS scheme, cluster-head selection is energy effi-cient, resulting in less uncertainty in cluster-head selection.Moreover, with the DEEC protocol, there were more clusterheads selected, leading to increased communication costs.With the proposed ECS scheme, only the Awake nodes takepart in cluster-head selection. Thus, the optimal numberof cluster heads is selected for every round in a controlledfashion to prolong the networks lifetime.

Figure 7 shows the data transmission to the fusion centeragainst the number of rounds. Owing to the Sleep and Awakemodes with our ECS scheme, there was less data transmitted

Number of rounds


0 1000 2000 3000 4000 50000

1

2

3

4

5

6

Num

ber o

f pac

kets

to F

C

×104

Figure 7: Data transmitted to the fusion center.

to the fusion center compared to the DEEC protocol. Underthe DEEC protocol, all nodes take part in data transmission.Hence, with the DEEC protocol, there is significantly moredata transmitted. Under the ECS scheme, only Awake nodestake part in data transmission to the cluster heads, and thecluster heads aggregate the data fromassociatedAwake-modeCRSN nodes exclusively, transmitting only this data to thefusion center.Thus, there is considerably less data transmittedunder the ECS scheme. This tradeoff can be justified by theincreased network lifetime and stability of the proposed ECSscheme.

4. Conclusion

In this paper, we proposed a novel energy-efficient clusteringbased cooperative spectrum sensing (what we call the “ECS”scheme) for cognitive radio sensor networks. Our maingoal was to enhance the energy efficiency of a cognitiveradio sensor network, in order to increase the networkslifetime and stability. Further, we developed techniques forenergy-efficient cluster-head selection and clustering. Withour proposed ECS scheme, we used spectrum-aware pairwisecoupling for the sensor nodes. The coupled nodes alternatebetween Awake and Sleep modes in order to minimizethe energy consumed by the network and to increase thelifetime of the network. Through extensive simulations, wedemonstrated a significant improvement in the stability andlifetime of a cognitive radio sensor network.

In future work, we will consider the involvement ofprimary users, andwewill provide network-coverage analysisand discuss data and decision-fusion and convergence ratesfor our proposed ECS scheme.

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper.


Acknowledgment

This research was supported by the MSIP (Ministry ofScience, ICT and Future Planning), Korea, under the GlobalIT Talent Support Program (NIPA-2014-H0904-14-1005).

References

[1] J. Yick, B. Mukherjee, and D. Ghosal, “Wireless sensor networksurvey,”ComputerNetworks, vol. 52, no. 12, pp. 2292–2330, 2008.

[2] A. A. Abbasi andM. Younis, “A survey on clustering algorithmsfor wireless sensor networks,” Computer Communications, vol.30, no. 14-15, pp. 2826–2841, 2007.

[3] US Federal Communications Commission, “Spectrum policytask force report,” ET Docket No. 02-135, 2002.

[4] J. Mitola and G. Q. Maguire Jr., “Cognitive radio: making soft-ware radios more personal,” IEEE Personal Communications,vol. 6, no. 4, pp. 13–18, 1999.

[5] S. Haykin, “Cognitive radio: brain-empowered wireless com-munications,” IEEE Journal on Selected Areas in Communica-tions, vol. 23, no. 2, pp. 201–220, 2005.

[6] G. Ganesan and Y. Li, “Cooperative spectrum sensing incognitive radio, part II: multiuser networks,” IEEE Transactionson Wireless Communications, vol. 6, no. 6, pp. 2214–2222, 2007.

[7] R. Fan and H. Jiang, “Optimal multi-channel cooperativesensing in cognitive radio networks,” IEEE Transactions onWireless Communications, vol. 9, no. 3, pp. 1128–1138, 2010.

[8] S. M. Mishra, A. Sahai, and R. W. Brodersen, “Cooperativesensing among cognitive radios,” in Proceedings of the IEEEInternational Conference on Communications (ICC ’06), pp.1658–1663, Istanbul, Turkey, July 2006.

[9] O. B. Akan, O. B. Karli, and O. Ergul, “Cognitive radio sensornetworks,” IEEE Network, vol. 23, no. 4, pp. 34–40, 2009.

[10] H. Shin, S. Moh, and I. Chung, “Clustering with one-time setupfor reduced energy consumption and prolonged lifetime inwireless sensor networks,” International Journal of DistributedSensor Networks, vol. 2013, Article ID 301869, 10 pages, 2013.

[11] W. R. Heinzelman, A. Chandrakasan, and H. Balakrish-nan, “Energy-efficient communication protocol for wirelessmicrosensor networks,” in Proceedings of the 33rd AnnualHawaii International Conference on System Sciences, p. 10, IEEE,January 2000.

[12] L. Qing, Q. Zhu, andM.Wang, “Design of a distributed energy-efficient clustering algorithm for heterogeneous wireless sensornetworks,”Computer Communications, vol. 29, no. 12, pp. 2230–2237, 2006.

[13] G. Smaragdakis, I. Matta, and A. Bestavros, “Sep: a stableelection protocol for clustered heterogeneous wireless sensornetworks,” Tech. Rep., Boston University Computer ScienceDepartment, 2004.

[14] O. Younis and S. Fahmy, “HEED: a hybrid, energy-efficient,distributed clustering approach for ad hoc sensor networks,”IEEE Transactions on Mobile Computing, vol. 3, no. 4, pp. 366–379, 2004.

[15] J. Abolarinwa, N. M. Abdul Latiff, and S. K. Syed, “Energy-efficient, learning-inspired channel decision and access tech-nique for cognitive radio-basedwireless sensor networks,” Inter-national Journal of Multimedia and Ubiquitous Engineering, vol.10, no. 2, pp. 11–24, 2015.

[16] X. Li, D. Wang, J. McNair, and J. Chen, “Dynamic spectrumaccess with packet size adaptation and residual energy balanc-ing for energy-constrained cognitive radio sensor networks,”

Journal of Network and Computer Applications, vol. 41, no. 1, pp.157–166, 2014.

[17] X. Tian, Z. Tian, K. Pham, E. Blasch, and G. Chen, “Qos-aware dynamic spectrum access for cognitive radio networks,”in Sensors and Systems for Space Applications VI, Proceedings ofSPIE, p. 87390P, International Society for Optics and Photonics,May 2013.

[18] Y. Liu, S. Xie, R. Yu, Y. Zhang, and C. Yuen, “An efficientMAC protocol with selective grouping and cooperative sensingin cognitive radio networks,” IEEE Transactions on VehicularTechnology, vol. 62, no. 8, pp. 3928–3941, 2013.

[19] W. Zhang, Y. Yang, C. K. Yeo, and L. Deng, “Cluster-basedcooperative spectrum sensing assignment strategy in cognitiveradio networks,” in Proceedings of the IEEE 80th VehicularTechnology Conference (VTC Fall), pp. 1–5, IEEE, Vancouver,Canada, September 2014.

[20] H. Zhang, Z. Zhang, H. Dai, R. Yin, and X. Chen, “Distributedspectrum-aware clustering in cognitive radio sensor networks,”in Proceedings of the IEEE Global Telecommunications Con-ference (GLOBECOM ’11), pp. 1–6, IEEE, Houston, Tex, USA,December 2011.

[21] H. Z. Zhang, Z. Y. Zhang, and C. Yuen, “Energy-efficientspectrum-aware clustering for cognitive radio sensor networks,”Chinese Science Bulletin, vol. 57, no. 28-29, pp. 3731–3739, 2012.

[22] K. Wagstaff, C. Cardie, S. Rogers, and S. Schrodl, “Constrainedk-means clustering with background knowledge,” in Proceed-ings of the 18th International Conference on Machine Learning(ICML ’01), vol. 1, pp. 577–584, 2001.

[23] D. Klein, S. D. Kamvar, and C. D. Manning, “From instance-level constraints to space-level constraints: making the most ofprior knowledge in data clustering,” in Proceedings of the 19thInternational Conference on Machine Learning (ICML ’02), pp.307–314, Sydney, Australia, July 2002.

[24] R. M. Eletreby, H. M. ElSayed, andM.M. Khairy, “CogLEACH:a spectrum aware clustering protocol for cognitive radio sensornetworks,” in Proceedings of the 9th International Conferenceon Cognitive Radio Oriented Wireless Networks and Communi-cations (CROWNCOM ’14), pp. 179–184, IEEE, Oulu, Finland,June 2014.

[25] M. Ye, C. Li, G. Chen, and J. Wu, “EECS: an energy efficientclustering scheme in wireless sensor networks,” in Proceedingsof the 24th IEEE International Performance, Computing, andCommunications Conference (IPCCC ’05), pp. 535–540, IEEE,April 2005.

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation http://www.hindawi.com

Journal ofEngineeringVolume 2014

Submit your manuscripts athttp://www.hindawi.com

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

SensorsJournal of


Modelling & Simulation in EngineeringHindawi Publishing Corporation http://www.hindawi.com Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks


Research Article A Novel Energy-Efficient Clustering Based ...downloads.hindawi.com/journals/ijdsn/2015/198456.pdf · assignment policy for cognitive-radio networks, in which all

Documents