Uniform Cluster Distribution with Anchor Free Node Localization for Energy Efficient Data Gathering in Wireless Sensor Networks

Abstract

Data gathering is an important research area in wireless sensor networks, being it the core operation. As an efficient topology management approach clustering is shown to have advantage, over other network architectures, due to its capability to reduce the communication burden and utilization of the processing features of the sensor nodes for local data aggregation. The algorithms proposed for this purpose may be classified as centralized or distributed based on the place of decision making. Centralized algorithms are revealed to lack the scalability in terms node density and distributed algorithms have an edge over them in such scenario. For distributed algorithms, it is needed to distribute the clusters uniformly over the network area, for them to be more energy efficient. Acquisition of the knowledge about their locations, by the sensor nodes, is expected to prove to be useful for this distribution but due to their limited energy resources and the associated cost providing location finding capabilities, like GPS or anchor nodes, is not advisable. If provided such facility, the error if occurs, due to its own limitations, may get propagated over the entire network. So, a scheme is proposed in this paper, for uniform distribution of the clusters over the network area in which the sensor nodes, utilizing anchor free and distributed localization mechanism, estimate their relative positions and based on these estimates assign to different clusters with an expectation from RSSI to provide the necessary support for required distance estimation. Sensor nodes, with knowledge of their distances to the base station, estimate their locations through collaborative efforts of small percentage of total nodes, with minimum of communication burden. Results of the simulations carried out in NS2, show that the proposed scheme has a potential to distribute the clusters uniformly over network area.