An EnergyEfficient VotingBased Clustering Algorithm for Sensor Networks

An Energy-Efficient Voting-Based Clustering Algorithm for Sensor Networks Min Qin and Roger Zimmermann Computer Science Department, Integrated Media Systems Center 5/24/2005

An Energy-Efficient Voting-Based Clustering Algorithm for Sensor Networks Presentation outline: §Clustering in sensor networks §Issues of traditional clustering protocols §Voting-based clustering algorithm (VCA) §Performance evaluation §Conclusions

Clustering in sensor networks §The network is divided into many clusters §Each cluster has one cluster head §Data collected from sensors are sent to the cluster head first, and then forwarded to the sink §Cluster head is capable of aggregating data from all sensors

A sample cluster Cluster head sink

Issues of traditional clustering protocols §Based on local weights or probabilities §Often result in undesirable cluster formations §Many of them do not consider topology information §Load balancing is hard

Example: a network with 4 sensors Desirable cluster formation C D B A

Clustering by traditional algorithms (HEED) C 0. 6 D 0. 7 B 0. 9 1 A

Resulting cluster formation (HEED) C D B A

Voting-based clustering algorithm (VCA) §Main ideas: §A sensor’s importance should be reflected from all its neighbors (including itself) rather than from itself §Use voting to reflect the importance of different neighbors § Topology and residual energy are two primary factors in selecting cluster heads §Assumptions about sensors: §Energy-aware §Quasi-stationary

VCA : An example D vote for all its neighbors (including itself) C 0. 25 B D 0. 25 A

VCA : An example D collect votes from all its neighbors 0. 5 C 0. 5 D 0. 5 B 0. 5 A 0. 5

VCA : An example Each node calculates the total vote it has got C 0. 75 D 1. 75 0. 75 B A 0. 75

Rules for voting § The sum of the votes a node gives to all its neighbors (including itself) is 1 § A neighbor with high residual energy should get more votes than a neighbor with low residual energy

Load balancing in VCA §If a sensor is covered by multiple cluster heads, the following two load balancing strategies are used : §Node degree v Join the head with the minimum node degree v. Balance the size of all clusters §Fitness v. Join the cluster head with the highest fitness v. Balance energy distribution of cluster heads

Procedures of VCA 1. Each sensor calculates its votes to all its neighbors 2. Sensors calculate and broadcast the total vote they have got from their neighbors. 3. Cluster heads are elected from those nodes that have the highest votes in their neighborhood 4. Sensors that are covered by at least one cluster head withdraw from voting 5. the remaining sensors restart from step 2 by ignoring the votes from those sensors that have withdrawn from the voting

Head Election in Multiple Rounds In the 1 st round 0. 5 C 0. 5 0. 25 0. 5 B D 0. 25 0. 33 A 0. 33 E 0. 5

Head Election in Multiple Rounds After the 1 st round 0. 75 C D 1. 58 B 0. 75 E 0. 83 A 1. 08

Head Election in Multiple Rounds In the 2 nd round, E ignores vote from A C D 0. 5 B A 0. 33 E 0. 5

Head Election in Multiple Rounds After the 2 nd round 0. 75 C D 1. 58 B 0. 75 A is covered by 2 cluster heads, it chooses E since it has lower degree E 0. 5 A 1. 08

Properties of VCA §Message complexity: O(N) §Time complexity: O(N) • Normally finishes within 2 -5 iterations §Cluster heads are well distributed, No two cluster heads covers each other §High-degree nodes tend to get more votes, and give less votes to others

Simulation settings Parameter Value S [0, 100]2 Sink location (50, 200) Cluster radius 15 m Data packet 250 bytes Clustering packet 30 bytes WITHDRAW packet 10 bytes Network phase 5 TDMA frames operation Energy for data fusion 5 n. J/bit/signal Initial Energy 2 J Threshold distance (d 0) 100 m

Network lifetime (when the first node dies) §A sensor joins the cluster head with the highest fitness value in VCA-fitness §A sensors joins the cluster head with the minimum node degree in VCA-Min degree §Average result from 100 independent simulations

Network lifetime (when the last node dies)

Conclusions §VCA is completely distributed, energy-efficient and location unaware §Using fitness can balance the energy across the network, sensors tend to die at a similar time §Using Min-degree can balance the size of all clusters, some nodes may live much longer than others §Democracy can be very helpful for sensor networks.

Conclusions §Questions? §Comments? §http: //dmrl. usc. edu/publications/ Thank you!