- Email: [email protected]
Effect of Heterogeneous Nodes Location on the Performance of Clustering Algorithms for Wireless Sensor Networks
Available online at www.sciencedirect.com
ScienceDirect Procedia Computer Science 57 (2015) 1042 – 1048
Third International Conference on Recent Trends in Computing (ICRTC 2015)
Effect of Heterogeneous Nodes Location on the Performance of Clustering Algorithms for Wireless Sensor Networks Vipin Pala,∗, Yogitab , Girdhari Singhc , R P Yadavc a Vivekananda
Global University, Jaipur - 302012, India Panjab University, Chandigarh - 110061, India c Malaviya National Institute of Technology, Jaipur - 302017, India b UIET,
Abstract Improved network lifetime without much increase in the cost contributes popularity to heterogeneous wireless sensor networks. Clustering algorithms designed to utilize advantage of heterogeneity of nodes allow these nodes to be cluster head more times than normal nodes to have load balanced network. Cluster head selection is pivotal for the performance of clustering algorithms as cluster quality in terms of communication distance depends upon the location of selected head in the cluster. Work of this paper analyzes the effect of location of heterogeneous nodes on the performance of clustering algorithms. Worst case, average (random) case and best case for location of heterogeneous nodes are considered for analyzing the effect on the performance of clustering algorithms. © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of organizing committee of the 3rd International Conference on Recent Trends in Computing 2015 (ICRTC-2015)
Keywords: Clustering Algorithm; Heterogeneous; Network Lifetime; Wireless Sensor Network
1. Introduction Wireless sensor networks 1,2 consist of various densely deployed sensor nodes. Sensor nodes sense the application area and send the sensed data to base station vie single hop or multi-hop. In-network processing is done either by the node itself or by other relying nodes to reduce the data. Sensor nodes work in-collaboration in the application area to complete the task. So, sensor node is the basic entity of wireless sensor networks. Advancement in MicroElectro-Mechanical-Systems (MEMS) provides low cost yet powerful sensor node that consists of sensing and data processing unit, wireless communication transceiver, battery and memory 3 . Battery power of sensor nodes is limited and harsh/remote application area makes it quite impossible to recharge or replace battery of these nodes. Battery power of sensor nodes is consumed by sensing, processing and communicating the data and also in other operations performed by nodes. So, energy carried by nodes is prime constraints to complete the task successfully and timely therefore there is more emphasis on efficient and economical energy consumption of sensor nodes from circuitry of nodes to protocols of network to application level. ∗
Corresponding author. Tel.: +91-7737701804. E-mail address: [email protected]
1877-0509 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of organizing committee of the 3rd International Conference on Recent Trends in Computing 2015 (ICRTC-2015) doi:10.1016/j.procs.2015.07.376
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
1043
Clustering approach 4,5 is considered energy efficient for wireless sensor networks and also provides scalability to the network. Clustering in wireless sensor networks is the idea to group nodes into clusters and selecting a cluster head node for each cluster to take responsibility of all nodes. Clustering approach is cross-layer approach that has impact on layers of network. Nodes sense the phenomenon and send data to cluster head node. Cluster head node is the point of in-network processing and applies data aggregation/fusion technique 6,3 to reduce that collected data to some meaningful information. Control on radio transmission power, reduced collisions, data aggregation/fusion, and TDMA scheduling in clustering approach provides energy efficiency to wireless sensor networks. Clustering algorithms seem to be attractive for achieving desired performance of wireless sensor network while consuming energy of sensor nodes efficiently. There are various clustering algorithms in literature for homogenous as well as heterogeneous wireless sensor networks. Heterogeneous wireless sensor networks are gaining importance because heterogeneity of nodes improves performance of network without demanding much increase in cost. Clustering algorithms for heterogeneous networks take advantage of these nodes in cluster head selection and give more chances to be selected as cluster head. Intra-cluster communication distance is the measure of quality of clusters which depends upon the position of cluster head in the cluster. As heterogeneous nodes are selected more times as cluster head therefore location of these nodes affects performance of clustering algorithms. Work of this paper analyzes the effect of location of heterogeneous nodes on the performance of clustering algorithms. Rest of paper is organized as: literature review is done in section 2. Section 3 describes network model. Section 4 describes effect of cluster head position on quality of cluster. Analysis of simulation results is done in section 5 and section 6 concludes the work of paper.
2. Review of Literature Low Energy Adaptive Cluster Hierarchy (LEACH) 7 is fully distributed clustering algorithm. In set-up phase, cluster head selection, cluster formation and TDMA scheduling of nodes are performed. In steady phase, nodes send data to cluster head and cluster head aggregates the data. Aggregated data is send to base station. Re-clustering is done over regular time periods to rotate role of cluster head among all nodes that makes network load balance. LEACH does not consider heterogeneity of nodes for cluster head selection, i.e. all nodes have equal probability of cluster head. 8 addresses problem of fixed round-time in LEACH and proposed a network adaptive round-time for LEACH. Round-time is adaptive to number of nodes alive in network. Adaptive Decentralized Re-clustering Protocol (ADRP) 9 is base station assisted protocol that selects a set of cluster heads for current round and another set of cluster heads for few next rounds. In the initial phase, nodes send their energy and location information to base station. Base station divides the network into different clusters and selects one cluster head for each cluster, depending upon energy and location. After cluster formation and cluster head selection, base station selects sets of cluster head nodes for few next rounds according to energy consumption approximation. Base station then broadcasts the cluster information, current cluster heads and next cluster heads. Nodes send data periodically to the cluster head and cluster heads send data after aggregation to base station. At the end of current round, new cluster heads are selected from the set of next cluster heads if list is not empty; otherwise initial phase is executed. Nodes save energy by avoiding re-clustering for few rounds as cluster heads are known in advance. LEACH and its variant are not able to handle heterogeneous nodes present in network. Nodes with extra resources are also treated as normal nodes. Stable Election Protocol (SEP) 10 is a clustering algorithm capable of handling heterogeneous nodes to prolong the stable region of network lifetime. SEP divides nodes into advance nodes (having high energy) and normal nodes. Probability of selection of a node as cluster head is different for advanced nodes and normal nodes. Epoch, number of rounds in which all nodes are selected as cluster head once, of the nodes is increased to let advanced nodes to become cluster head for more rounds. SEP-E 11 extends the level of heterogeneity of nodes up to level three. There are three types of nodes Advance nodes, Intermediate nodes and Normal nodes. SEP and its variant does not consider residual energy of nodes or residual network of network. Distributed Energy Efficient Clustering (DEEC) 12 selects cluster heads according to residual energy of node and average energy of network. Two types of nodes - Advance nodes and normal nodes, are considered for heterogeneous network. Election probabilities are different for advance nodes and normal nodes which also takes account residual energy of nodes. High residual energy nodes have more chances for role of cluster head over other nodes in a round. Energy Efficient
1044
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
Heterogeneous Clusters (EEHC) scheme 13 extends the work of DEEC and SEP-E. Three level of nodes are considered Super node, Advance node and Normal node. Energyaware Routing Protocol (ERP) 14 exploits heterogeneous energy of nodes for cluster head selection. ERP minimizes the energy consumption of nodes by area coverage problem. A node with high residual energy to the average of all the neighbor nodes in cluster range has high probability of being cluster head. In the beginning of each round, nodes exchange information message with nodes in their cluster range. With updation in neighbor table, each node calculates average energy of cluster range nodes. For completing cluster head selection, broadcasting delay time is calculated according to residual energy of node and average energy of cluster range nodes. Nodes with high residual energy are elected as cluster heads eventually. Plain nodes join nearest cluster head node. By solving the area coverage problem according to the requirements of application, few nodes are selected for active state while other remaining nodes transit to sleep state to save energy. Heterogeneous clustering algorithms extend the network lifetime but do not analyze the effect of position of these heterogeneous nodes on the performance. 3. Network Model Following network assumptions are considered: • All sensor nodes and base station are stationary once deployed in the field. • There is single base station located outside the field. • The nodes are considered to die only when their energy is exhausted. Most of the energy of nodes is dissipated due to communication between two nodes and it depends on the distance between them. Both sending and receiving of data consumes energy 15 . Energy dissipation model is shown in Fig. 1 and explained next.
Fig. 1. Radio Energy Model
For sending m bit data over a distance d, the total energy consumed by a node is as follows: ET X (m, d) = ET X−elec (m) + ET X−amp (m, d) ET X (m, d) =
mEelec +m f s d2 mEelec +mmp d4
(1)
d < dcrossover d ≥ dcrossover
(2)
where dcrossover is crossover distance, while the energy consumption for receiving that message is as follows: ERX (m) = mEelec For considered network model, values considered for Eelec is 50nJ/bit, The crossover distance dcrossover is considered 87m.
(3) fs
is 10pJ/bit/m2 , and amp is 0.0013pJ/bit/m4 .
4. Problem Definition Clustering algorithms are energy efficient approach for wireless sensor networks. Cluster head selection plays vital role on the performance of clustering algorithms. This section shows the effect of selection of cluster heads.
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
1045
4.1. Effect of Position of Cluster Head Clustering algorithms groups the nodes in clusters and selects one cluster head for each cluster to have the hierarchy in the network. Nodes send the information about the environment to respective cluster head and cluster heads send the collected data to the base station after performing data aggregation/ data fusion. So, communication involved in clustering algorithm can be classified as: Inter-cluster communication communication between cluster heads and base station, and intra-cluster communication communication between member nodes and cluster head 16 . Consequently, distances involved in the communication can also be classified as: Inter-cluster communication distance and intracluster communication distance. Most of clustering algorithms operate in rounds. Data transmission phase of a round is divided into frames which are further divided in time slots. Each member node has one time slot in each frame 7,17,18 . At the end of each frame cluster head send data to base station. As a result, communication between member nodes and cluster heads is more as compared to communication between cluster heads and base station. So, intra-cluster communication distance is more than inter-cluster communication distance. Intra-cluster communication distance 16 can be defined as in Eq. (4): Intra − cluster communication distance =
N
Dist(i, CH)
(4)
i=1
where n is the number of member nodes of a cluster and Dist(i,CH) is the function that gives distance of node i to cluster head (CH). 16 also shows that efficiency of cluster depends upon position of cluster head in the cluster. Clusters having position of cluster head near to the center of clusters have less intra-cluster communication distance and are energy efficient. 4.2. Heterogeneous Networks Heterogeneous sensor networks are gaining ample interest because of not much increase in the overall cost of network 19,20 . Various clustering algorithms are proposed in the literature that capitalizes the presence of heterogeneous nodes to increase the lifetime of network 10,11,12,13 where tradition clustering algorithms are not handling these nodes 7,21,22,23 . Heterogeneous clustering algorithms handle heterogeneous nodes by giving preference to heterogeneous nodes over normal nodes in cluster head selection. Different epochs and selection probabilities are set for heterogeneous nodes and normal nodes. Heterogeneous nodes are selected more times than the normal nodes during the complete lifetime of network. Therefore in these algorithms, role of cluster head is rotated among all nodes but heterogeneous nodes are selected more times than the normal nodes. As discussed in the above section, performance of cluster relies on the position of cluster head in the cluster therefore, so performance of heterogeneous clustering algorithms depends on the position of heterogeneous nodes. Work of this paper analyzes the effect of position of heterogeneous nodes on the performance of existing clustering algorithms. 5. Simulation Results and Analysis A network topology of 100 nodes deployed randomly over an area of 100×100m2 is generated for analyzing the effect of position of heterogeneous nodes on the performance of clustering algorithms. Topology has two types of nodes advance nodes and normal nodes. Normal nodes have 0.5J initial energy. 10% of total nodes are heterogeneous nodes which have equal amount of energy at the time of deployment and is 1.0J (double of normal node energy). Data packet is of size 6400 bytes while control packet is of 200 bytes. Base station is positioned at (125,50). Three different scenarios for deployment of heterogeneous nodes are considered Random, Worst and Best. In random scenario, heterogeneous nodes are deployed randomly along with the other nodes. Worst scenario can have heterogeneous nodes positioned near to border of the deployed area or all these nodes have position nearby in the field. In the work of this paper first scenario is considered because in that case clusters are not a complete circle. In best case scenario, heterogeneous nodes are positioned such as cluster of these nodes are almost complete circle. Following simulation metrics are considered for the analysis of effect of heterogeneous nodes on the performance of clustering algorithms.
1046
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
Random
Worst
Best
NumberofAliveNodes
120 100 80 60 40 20 0 400
450
500
550
600
650
700
750
Rounds
Fig. 2. Node Death Rate of LEACH
• Node Death Rate: It demonstrates number of alive nodes over rounds. A lower node death rate happens because of load balanced network. The region of the node death rate is divided as stable region and unstable region. All nodes are alive in the stable region while unstable region is rest of the region. • Network Lifetime: It can be defined as the working period of network. 24 defines classifies network lifetime in three parts First Node Death (FND), Half Node Death (HND) and Last Node Death (LND). In our work, LND for death of 90% nodes because there are 10% advance nodes. Figure 2 demonstrates a detailed view of node death rate of LEACH for the three different scenarios. Figure shows alive nodes over rounds. LEACH does not capitalize the presence of node heterogeneity so does not get affected. In all three scenarios, node death rate is almost same and stable and unstable regions are also almost same. Figure 3 shows performance of SEP for all three scenarios. SEP has different epoch for advance nodes and normal nodes. Advance nodes are selected more times than normal nodes. In case of best scenario, advance nodes have better cluster formation than random scenario and worst scenario so have better load balance than the others. So, best case has prolonged stable region than random and worst case.
Random
Worst
Best
NumberofAliveNodes
120 100 80 60 40 20 0 200
400
600
800
1000
1200
Rounds
Fig. 3. Node Death Rate of SEP
DEEC protocol also takes care of network dynamics and considers residual energy of node and average energy of network for cluster head selection. So, advance nodes are selected more times and also have preference over normal nodes in round. Figure 3 gives a detail view of node death rate for three cases. As can be seen from the figure, Stable region of best case is prolonged as compared to worst case and random scenario. DEEC also got affected most in worst case, because clusters with less efficiency occur frequently.
1047
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
Random
Worst
Best
NumberofActiveNodes
120 100 80 60 40 20 0 200
400
600
800
1000
1200
1400
Rounds
Fig. 4. Node Death Rate of DEEC
Table 1 shows comparison of network lifetime of LEACH, SEP and DEEC protocols for all three scenarios. LEACH does not consider node heterogeneity in cluster head selection so have almost same network lifetime (different values happens because of random nature of protocol and simulation environment). In best case, network is better load balanced than the other scenarios so lifetime of network get extended. In case of SEP protocol, FND decreases by 7.5% for worst case while increases by 8.3% for best case over random deployment. Table 1. COMPARISON OF NETWORK LIFETIME OF LEACH, SEP, DEEC. Protocol
Topology
FND
HND
LND
LEACH
Random Worst Best
509 494 531
612 592 618
677 669 688
SEP
Random Worst Best
541 500 586
703 648 735
808 762 852
DEEC
Random Worst Best
570 520 605
773 723 819
975 952 1030
6. Conclusion Node heterogeneity prolong the network lifetime. Heterogeneous clustering approaches capitalize the presence of heterogeneous nodes for cluster head selection. Efficiency of cluster depends upon position of cluster head nodes. Performance of heterogeneous network is affected by the position of heterogeneous nodes. Work of this paper, anatomizes the effect of position of heterogeneous nodes on the performance of traditional clustering approaches. 7. References References 1. Akyildiz, I., Su, W., Sankarasubramaniam, Y., Cayirci, E.. Wireless sensor networks: a survey. Computer Networks 2002;38(4):393 – 422. 2. Anastasi, G., Conti, M., Di Francesco, M., Passarella, A.. Energy conservation in wireless sensor networks: A survey. Ad Hoc Netw 2009; 7(3):537–568.
1048
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
3. Karl, H., Willig, A.. Protocols and architectures for wireless sensor networks. John Wiley & Sons; 2007. 4. Abbasi, A.A., Younis, M.. A survey on clustering algorithms for wireless sensor networks. Comput Commun 2007;30(14-15):2826–2841. 5. Tyagi, S., Kumar, N.. A systematic review on clustering and routing techniques based upon {LEACH} protocol for wireless sensor networks. Journal of Network and Computer Applications 2013;36(2):623 – 645. 6. Rajagopalan, R., Varshney, P.K.. Data aggregation techniques in sensor networks: A survey. Comm Surveys and Tutorials, IEEE 2006; 8:48–63. 7. Heinzelman, W.R., Chandrakasan, A., Balakrishnan, H.. Energy-efficient communication protocol for wireless microsensor networks. In: Proceedings of the 33rd Hawaii International Conference on System Sciences-Volume 8; HICSS ’00. Washington, DC, USA: IEEE Computer Society. ISBN 0-7695-0493-0; 2000, p. 10 – 20. 8. Pal, V., Singh, G., Yadav, R.. Network adaptive round-time clustering algorithm for wireless sensor networks. In: Advances in Computing, Communications and Informatics (ICACCI), 2013 International Conference on. 2013, p. 1299–1302. 9. Bajaber, F., Awan, I.. Adaptive decentralized re-clustering protocol for wireless sensor networks. Journal of Computer System and Science 2011;77(2):282–292. 10. Smaragdakis, G., Matta, I., Bestavros, A.. SEP: A Stable Election Protocol for clustered heterogeneous wireless sensor networks. In: Second International Workshop on Sensor and Actor Network Protocols and Applications (SANPA 2004). Boston, MA; 2004, . 11. F. A. Aderohunmu, J.D.D.. An Enhanced Stable Election Protocol for Clustered Heterogeneous Wireless Sensor Networks. ????, . 12. Qing, L., Zhu, Q., Wang, M.. Design of a distributed energy-efficient clustering algorithm for heterogeneous wireless sensor networks. Comput Commun 2006;29(12):2230–2237. 13. Kumar, D., Aseri, T.C., Patel, R.. EEHC: Energy efficient heterogeneous clustered scheme for wireless sensor networks. Computer Communications 2009;32(4):662 – 667. 14. Liu, M., Cao, J., Chen, G., Wang, X.. An energy-aware routing protocol in wireless sensor networks. Sensors 2009;9(1):445–462. 15. Rappaport, T.. Wireless Communications: Principles and Practice. Upper Saddle River, NJ, USA: Prentice Hall PTR; 2nd ed.; 2001. ISBN 0130422320. 16. Pal, V., Singh, G., Yadav, R.P.. Schs: Smart cluster head selection scheme for clustering algorithms in wireless sensor networks. Wireless Sensor Network 2012;4:273. 17. Wang, A., Yang, D., Sun, D.. A clustering algorithm based on energy information and cluster heads expectation for wireless sensor networks. Computers & Electrical Engineering 2012;38(3):662 – 671. 18. Wu, X., Xiong, Y., Huang, W., Shen, H., Li, M.. An efficient compressive data gathering routing scheme for large-scale wireless sensor networks. Computers & Electrical Engineering 2013;39(6):1935 – 1946. 19. Mhatre, V., Rosenberg, C.. Design guidelines for wireless sensor networks: communication, clustering and aggregation. Ad Hoc Networks Journal, Elsevier Science 2004;2:45–63. 20. Yarvis, M., Kushalnagar, A., Singh, H., Liu, Y., Singh, S.. Exploiting heterogeneity in sensor networks. In: in Proc. of the IEEE Infocom. 2005, . 21. Heinzelman, W., Chandrakasan, A., Balakrishnan, H.. An application-specific protocol architecture for wireless microsensor networks. Wireless Communications, IEEE Transactions on 2002;1(4):660 – 670. 22. Heinzelman, W.B.. Application-specific protocol architectures for wireless networks. Ph.D. thesis; 2000. AAI0801929. 23. Jang, K.Y., Kim, K.T., Youn, H.Y.. An energy efficient routing scheme for wireless sensor networks. In: Computational Science and its Applications, 2007. ICCSA 2007. International Conference on. 2007, p. 399–404. 24. Haase, M., Timmermann, D.. energy adaptive clustering hierarchy with deterministic cluster-head selection. In: In IEEE Conference on Mobile and Wireless Communications Networks (MWCN) (2002. 2002, p. 368–372.
ScienceDirect Procedia Computer Science 57 (2015) 1042 – 1048
Third International Conference on Recent Trends in Computing (ICRTC 2015)
Effect of Heterogeneous Nodes Location on the Performance of Clustering Algorithms for Wireless Sensor Networks Vipin Pala,∗, Yogitab , Girdhari Singhc , R P Yadavc a Vivekananda
Global University, Jaipur - 302012, India Panjab University, Chandigarh - 110061, India c Malaviya National Institute of Technology, Jaipur - 302017, India b UIET,
Abstract Improved network lifetime without much increase in the cost contributes popularity to heterogeneous wireless sensor networks. Clustering algorithms designed to utilize advantage of heterogeneity of nodes allow these nodes to be cluster head more times than normal nodes to have load balanced network. Cluster head selection is pivotal for the performance of clustering algorithms as cluster quality in terms of communication distance depends upon the location of selected head in the cluster. Work of this paper analyzes the effect of location of heterogeneous nodes on the performance of clustering algorithms. Worst case, average (random) case and best case for location of heterogeneous nodes are considered for analyzing the effect on the performance of clustering algorithms. © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of organizing committee of the 3rd International Conference on Recent Trends in Computing 2015 (ICRTC-2015)
Keywords: Clustering Algorithm; Heterogeneous; Network Lifetime; Wireless Sensor Network
1. Introduction Wireless sensor networks 1,2 consist of various densely deployed sensor nodes. Sensor nodes sense the application area and send the sensed data to base station vie single hop or multi-hop. In-network processing is done either by the node itself or by other relying nodes to reduce the data. Sensor nodes work in-collaboration in the application area to complete the task. So, sensor node is the basic entity of wireless sensor networks. Advancement in MicroElectro-Mechanical-Systems (MEMS) provides low cost yet powerful sensor node that consists of sensing and data processing unit, wireless communication transceiver, battery and memory 3 . Battery power of sensor nodes is limited and harsh/remote application area makes it quite impossible to recharge or replace battery of these nodes. Battery power of sensor nodes is consumed by sensing, processing and communicating the data and also in other operations performed by nodes. So, energy carried by nodes is prime constraints to complete the task successfully and timely therefore there is more emphasis on efficient and economical energy consumption of sensor nodes from circuitry of nodes to protocols of network to application level. ∗
Corresponding author. Tel.: +91-7737701804. E-mail address: [email protected]
1877-0509 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of organizing committee of the 3rd International Conference on Recent Trends in Computing 2015 (ICRTC-2015) doi:10.1016/j.procs.2015.07.376
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
1043
Clustering approach 4,5 is considered energy efficient for wireless sensor networks and also provides scalability to the network. Clustering in wireless sensor networks is the idea to group nodes into clusters and selecting a cluster head node for each cluster to take responsibility of all nodes. Clustering approach is cross-layer approach that has impact on layers of network. Nodes sense the phenomenon and send data to cluster head node. Cluster head node is the point of in-network processing and applies data aggregation/fusion technique 6,3 to reduce that collected data to some meaningful information. Control on radio transmission power, reduced collisions, data aggregation/fusion, and TDMA scheduling in clustering approach provides energy efficiency to wireless sensor networks. Clustering algorithms seem to be attractive for achieving desired performance of wireless sensor network while consuming energy of sensor nodes efficiently. There are various clustering algorithms in literature for homogenous as well as heterogeneous wireless sensor networks. Heterogeneous wireless sensor networks are gaining importance because heterogeneity of nodes improves performance of network without demanding much increase in cost. Clustering algorithms for heterogeneous networks take advantage of these nodes in cluster head selection and give more chances to be selected as cluster head. Intra-cluster communication distance is the measure of quality of clusters which depends upon the position of cluster head in the cluster. As heterogeneous nodes are selected more times as cluster head therefore location of these nodes affects performance of clustering algorithms. Work of this paper analyzes the effect of location of heterogeneous nodes on the performance of clustering algorithms. Rest of paper is organized as: literature review is done in section 2. Section 3 describes network model. Section 4 describes effect of cluster head position on quality of cluster. Analysis of simulation results is done in section 5 and section 6 concludes the work of paper.
2. Review of Literature Low Energy Adaptive Cluster Hierarchy (LEACH) 7 is fully distributed clustering algorithm. In set-up phase, cluster head selection, cluster formation and TDMA scheduling of nodes are performed. In steady phase, nodes send data to cluster head and cluster head aggregates the data. Aggregated data is send to base station. Re-clustering is done over regular time periods to rotate role of cluster head among all nodes that makes network load balance. LEACH does not consider heterogeneity of nodes for cluster head selection, i.e. all nodes have equal probability of cluster head. 8 addresses problem of fixed round-time in LEACH and proposed a network adaptive round-time for LEACH. Round-time is adaptive to number of nodes alive in network. Adaptive Decentralized Re-clustering Protocol (ADRP) 9 is base station assisted protocol that selects a set of cluster heads for current round and another set of cluster heads for few next rounds. In the initial phase, nodes send their energy and location information to base station. Base station divides the network into different clusters and selects one cluster head for each cluster, depending upon energy and location. After cluster formation and cluster head selection, base station selects sets of cluster head nodes for few next rounds according to energy consumption approximation. Base station then broadcasts the cluster information, current cluster heads and next cluster heads. Nodes send data periodically to the cluster head and cluster heads send data after aggregation to base station. At the end of current round, new cluster heads are selected from the set of next cluster heads if list is not empty; otherwise initial phase is executed. Nodes save energy by avoiding re-clustering for few rounds as cluster heads are known in advance. LEACH and its variant are not able to handle heterogeneous nodes present in network. Nodes with extra resources are also treated as normal nodes. Stable Election Protocol (SEP) 10 is a clustering algorithm capable of handling heterogeneous nodes to prolong the stable region of network lifetime. SEP divides nodes into advance nodes (having high energy) and normal nodes. Probability of selection of a node as cluster head is different for advanced nodes and normal nodes. Epoch, number of rounds in which all nodes are selected as cluster head once, of the nodes is increased to let advanced nodes to become cluster head for more rounds. SEP-E 11 extends the level of heterogeneity of nodes up to level three. There are three types of nodes Advance nodes, Intermediate nodes and Normal nodes. SEP and its variant does not consider residual energy of nodes or residual network of network. Distributed Energy Efficient Clustering (DEEC) 12 selects cluster heads according to residual energy of node and average energy of network. Two types of nodes - Advance nodes and normal nodes, are considered for heterogeneous network. Election probabilities are different for advance nodes and normal nodes which also takes account residual energy of nodes. High residual energy nodes have more chances for role of cluster head over other nodes in a round. Energy Efficient
1044
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
Heterogeneous Clusters (EEHC) scheme 13 extends the work of DEEC and SEP-E. Three level of nodes are considered Super node, Advance node and Normal node. Energyaware Routing Protocol (ERP) 14 exploits heterogeneous energy of nodes for cluster head selection. ERP minimizes the energy consumption of nodes by area coverage problem. A node with high residual energy to the average of all the neighbor nodes in cluster range has high probability of being cluster head. In the beginning of each round, nodes exchange information message with nodes in their cluster range. With updation in neighbor table, each node calculates average energy of cluster range nodes. For completing cluster head selection, broadcasting delay time is calculated according to residual energy of node and average energy of cluster range nodes. Nodes with high residual energy are elected as cluster heads eventually. Plain nodes join nearest cluster head node. By solving the area coverage problem according to the requirements of application, few nodes are selected for active state while other remaining nodes transit to sleep state to save energy. Heterogeneous clustering algorithms extend the network lifetime but do not analyze the effect of position of these heterogeneous nodes on the performance. 3. Network Model Following network assumptions are considered: • All sensor nodes and base station are stationary once deployed in the field. • There is single base station located outside the field. • The nodes are considered to die only when their energy is exhausted. Most of the energy of nodes is dissipated due to communication between two nodes and it depends on the distance between them. Both sending and receiving of data consumes energy 15 . Energy dissipation model is shown in Fig. 1 and explained next.
Fig. 1. Radio Energy Model
For sending m bit data over a distance d, the total energy consumed by a node is as follows: ET X (m, d) = ET X−elec (m) + ET X−amp (m, d) ET X (m, d) =
mEelec +m f s d2 mEelec +mmp d4
(1)
d < dcrossover d ≥ dcrossover
(2)
where dcrossover is crossover distance, while the energy consumption for receiving that message is as follows: ERX (m) = mEelec For considered network model, values considered for Eelec is 50nJ/bit, The crossover distance dcrossover is considered 87m.
(3) fs
is 10pJ/bit/m2 , and amp is 0.0013pJ/bit/m4 .
4. Problem Definition Clustering algorithms are energy efficient approach for wireless sensor networks. Cluster head selection plays vital role on the performance of clustering algorithms. This section shows the effect of selection of cluster heads.
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
1045
4.1. Effect of Position of Cluster Head Clustering algorithms groups the nodes in clusters and selects one cluster head for each cluster to have the hierarchy in the network. Nodes send the information about the environment to respective cluster head and cluster heads send the collected data to the base station after performing data aggregation/ data fusion. So, communication involved in clustering algorithm can be classified as: Inter-cluster communication communication between cluster heads and base station, and intra-cluster communication communication between member nodes and cluster head 16 . Consequently, distances involved in the communication can also be classified as: Inter-cluster communication distance and intracluster communication distance. Most of clustering algorithms operate in rounds. Data transmission phase of a round is divided into frames which are further divided in time slots. Each member node has one time slot in each frame 7,17,18 . At the end of each frame cluster head send data to base station. As a result, communication between member nodes and cluster heads is more as compared to communication between cluster heads and base station. So, intra-cluster communication distance is more than inter-cluster communication distance. Intra-cluster communication distance 16 can be defined as in Eq. (4): Intra − cluster communication distance =
N
Dist(i, CH)
(4)
i=1
where n is the number of member nodes of a cluster and Dist(i,CH) is the function that gives distance of node i to cluster head (CH). 16 also shows that efficiency of cluster depends upon position of cluster head in the cluster. Clusters having position of cluster head near to the center of clusters have less intra-cluster communication distance and are energy efficient. 4.2. Heterogeneous Networks Heterogeneous sensor networks are gaining ample interest because of not much increase in the overall cost of network 19,20 . Various clustering algorithms are proposed in the literature that capitalizes the presence of heterogeneous nodes to increase the lifetime of network 10,11,12,13 where tradition clustering algorithms are not handling these nodes 7,21,22,23 . Heterogeneous clustering algorithms handle heterogeneous nodes by giving preference to heterogeneous nodes over normal nodes in cluster head selection. Different epochs and selection probabilities are set for heterogeneous nodes and normal nodes. Heterogeneous nodes are selected more times than the normal nodes during the complete lifetime of network. Therefore in these algorithms, role of cluster head is rotated among all nodes but heterogeneous nodes are selected more times than the normal nodes. As discussed in the above section, performance of cluster relies on the position of cluster head in the cluster therefore, so performance of heterogeneous clustering algorithms depends on the position of heterogeneous nodes. Work of this paper analyzes the effect of position of heterogeneous nodes on the performance of existing clustering algorithms. 5. Simulation Results and Analysis A network topology of 100 nodes deployed randomly over an area of 100×100m2 is generated for analyzing the effect of position of heterogeneous nodes on the performance of clustering algorithms. Topology has two types of nodes advance nodes and normal nodes. Normal nodes have 0.5J initial energy. 10% of total nodes are heterogeneous nodes which have equal amount of energy at the time of deployment and is 1.0J (double of normal node energy). Data packet is of size 6400 bytes while control packet is of 200 bytes. Base station is positioned at (125,50). Three different scenarios for deployment of heterogeneous nodes are considered Random, Worst and Best. In random scenario, heterogeneous nodes are deployed randomly along with the other nodes. Worst scenario can have heterogeneous nodes positioned near to border of the deployed area or all these nodes have position nearby in the field. In the work of this paper first scenario is considered because in that case clusters are not a complete circle. In best case scenario, heterogeneous nodes are positioned such as cluster of these nodes are almost complete circle. Following simulation metrics are considered for the analysis of effect of heterogeneous nodes on the performance of clustering algorithms.
1046
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
Random
Worst
Best
NumberofAliveNodes
120 100 80 60 40 20 0 400
450
500
550
600
650
700
750
Rounds
Fig. 2. Node Death Rate of LEACH
• Node Death Rate: It demonstrates number of alive nodes over rounds. A lower node death rate happens because of load balanced network. The region of the node death rate is divided as stable region and unstable region. All nodes are alive in the stable region while unstable region is rest of the region. • Network Lifetime: It can be defined as the working period of network. 24 defines classifies network lifetime in three parts First Node Death (FND), Half Node Death (HND) and Last Node Death (LND). In our work, LND for death of 90% nodes because there are 10% advance nodes. Figure 2 demonstrates a detailed view of node death rate of LEACH for the three different scenarios. Figure shows alive nodes over rounds. LEACH does not capitalize the presence of node heterogeneity so does not get affected. In all three scenarios, node death rate is almost same and stable and unstable regions are also almost same. Figure 3 shows performance of SEP for all three scenarios. SEP has different epoch for advance nodes and normal nodes. Advance nodes are selected more times than normal nodes. In case of best scenario, advance nodes have better cluster formation than random scenario and worst scenario so have better load balance than the others. So, best case has prolonged stable region than random and worst case.
Random
Worst
Best
NumberofAliveNodes
120 100 80 60 40 20 0 200
400
600
800
1000
1200
Rounds
Fig. 3. Node Death Rate of SEP
DEEC protocol also takes care of network dynamics and considers residual energy of node and average energy of network for cluster head selection. So, advance nodes are selected more times and also have preference over normal nodes in round. Figure 3 gives a detail view of node death rate for three cases. As can be seen from the figure, Stable region of best case is prolonged as compared to worst case and random scenario. DEEC also got affected most in worst case, because clusters with less efficiency occur frequently.
1047
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
Random
Worst
Best
NumberofActiveNodes
120 100 80 60 40 20 0 200
400
600
800
1000
1200
1400
Rounds
Fig. 4. Node Death Rate of DEEC
Table 1 shows comparison of network lifetime of LEACH, SEP and DEEC protocols for all three scenarios. LEACH does not consider node heterogeneity in cluster head selection so have almost same network lifetime (different values happens because of random nature of protocol and simulation environment). In best case, network is better load balanced than the other scenarios so lifetime of network get extended. In case of SEP protocol, FND decreases by 7.5% for worst case while increases by 8.3% for best case over random deployment. Table 1. COMPARISON OF NETWORK LIFETIME OF LEACH, SEP, DEEC. Protocol
Topology
FND
HND
LND
LEACH
Random Worst Best
509 494 531
612 592 618
677 669 688
SEP
Random Worst Best
541 500 586
703 648 735
808 762 852
DEEC
Random Worst Best
570 520 605
773 723 819
975 952 1030
6. Conclusion Node heterogeneity prolong the network lifetime. Heterogeneous clustering approaches capitalize the presence of heterogeneous nodes for cluster head selection. Efficiency of cluster depends upon position of cluster head nodes. Performance of heterogeneous network is affected by the position of heterogeneous nodes. Work of this paper, anatomizes the effect of position of heterogeneous nodes on the performance of traditional clustering approaches. 7. References References 1. Akyildiz, I., Su, W., Sankarasubramaniam, Y., Cayirci, E.. Wireless sensor networks: a survey. Computer Networks 2002;38(4):393 – 422. 2. Anastasi, G., Conti, M., Di Francesco, M., Passarella, A.. Energy conservation in wireless sensor networks: A survey. Ad Hoc Netw 2009; 7(3):537–568.
1048
Vipin Pal et al. / Procedia Computer Science 57 (2015) 1042 – 1048
3. Karl, H., Willig, A.. Protocols and architectures for wireless sensor networks. John Wiley & Sons; 2007. 4. Abbasi, A.A., Younis, M.. A survey on clustering algorithms for wireless sensor networks. Comput Commun 2007;30(14-15):2826–2841. 5. Tyagi, S., Kumar, N.. A systematic review on clustering and routing techniques based upon {LEACH} protocol for wireless sensor networks. Journal of Network and Computer Applications 2013;36(2):623 – 645. 6. Rajagopalan, R., Varshney, P.K.. Data aggregation techniques in sensor networks: A survey. Comm Surveys and Tutorials, IEEE 2006; 8:48–63. 7. Heinzelman, W.R., Chandrakasan, A., Balakrishnan, H.. Energy-efficient communication protocol for wireless microsensor networks. In: Proceedings of the 33rd Hawaii International Conference on System Sciences-Volume 8; HICSS ’00. Washington, DC, USA: IEEE Computer Society. ISBN 0-7695-0493-0; 2000, p. 10 – 20. 8. Pal, V., Singh, G., Yadav, R.. Network adaptive round-time clustering algorithm for wireless sensor networks. In: Advances in Computing, Communications and Informatics (ICACCI), 2013 International Conference on. 2013, p. 1299–1302. 9. Bajaber, F., Awan, I.. Adaptive decentralized re-clustering protocol for wireless sensor networks. Journal of Computer System and Science 2011;77(2):282–292. 10. Smaragdakis, G., Matta, I., Bestavros, A.. SEP: A Stable Election Protocol for clustered heterogeneous wireless sensor networks. In: Second International Workshop on Sensor and Actor Network Protocols and Applications (SANPA 2004). Boston, MA; 2004, . 11. F. A. Aderohunmu, J.D.D.. An Enhanced Stable Election Protocol for Clustered Heterogeneous Wireless Sensor Networks. ????, . 12. Qing, L., Zhu, Q., Wang, M.. Design of a distributed energy-efficient clustering algorithm for heterogeneous wireless sensor networks. Comput Commun 2006;29(12):2230–2237. 13. Kumar, D., Aseri, T.C., Patel, R.. EEHC: Energy efficient heterogeneous clustered scheme for wireless sensor networks. Computer Communications 2009;32(4):662 – 667. 14. Liu, M., Cao, J., Chen, G., Wang, X.. An energy-aware routing protocol in wireless sensor networks. Sensors 2009;9(1):445–462. 15. Rappaport, T.. Wireless Communications: Principles and Practice. Upper Saddle River, NJ, USA: Prentice Hall PTR; 2nd ed.; 2001. ISBN 0130422320. 16. Pal, V., Singh, G., Yadav, R.P.. Schs: Smart cluster head selection scheme for clustering algorithms in wireless sensor networks. Wireless Sensor Network 2012;4:273. 17. Wang, A., Yang, D., Sun, D.. A clustering algorithm based on energy information and cluster heads expectation for wireless sensor networks. Computers & Electrical Engineering 2012;38(3):662 – 671. 18. Wu, X., Xiong, Y., Huang, W., Shen, H., Li, M.. An efficient compressive data gathering routing scheme for large-scale wireless sensor networks. Computers & Electrical Engineering 2013;39(6):1935 – 1946. 19. Mhatre, V., Rosenberg, C.. Design guidelines for wireless sensor networks: communication, clustering and aggregation. Ad Hoc Networks Journal, Elsevier Science 2004;2:45–63. 20. Yarvis, M., Kushalnagar, A., Singh, H., Liu, Y., Singh, S.. Exploiting heterogeneity in sensor networks. In: in Proc. of the IEEE Infocom. 2005, . 21. Heinzelman, W., Chandrakasan, A., Balakrishnan, H.. An application-specific protocol architecture for wireless microsensor networks. Wireless Communications, IEEE Transactions on 2002;1(4):660 – 670. 22. Heinzelman, W.B.. Application-specific protocol architectures for wireless networks. Ph.D. thesis; 2000. AAI0801929. 23. Jang, K.Y., Kim, K.T., Youn, H.Y.. An energy efficient routing scheme for wireless sensor networks. In: Computational Science and its Applications, 2007. ICCSA 2007. International Conference on. 2007, p. 399–404. 24. Haase, M., Timmermann, D.. energy adaptive clustering hierarchy with deterministic cluster-head selection. In: In IEEE Conference on Mobile and Wireless Communications Networks (MWCN) (2002. 2002, p. 368–372.