Low complexity landmark-node tracing in WSNs using multi-agent random walks

Abstract

Mobile sinks are generally used in wireless sensor networks for energy-efficient data collection. Scheduling of mobile sink in a large scale network is a challenging problem. Tracing of some nodes in the network as landmark-nodes, via clustering can lead to efficient mobile sink scheduling. In this paper, a novel method for landmark-node tracing based on multi-agent random walks on network graphs is proposed. This method ensures low complexity while maintaining the clustering efficiency, especially over large WSN. Additionally, it is energy-efficient and improves the lifetime of a network. Low complexity of the proposed method is illustrated by analysing the cover time and hitting time of multi-agent random walks. Extensive experimental results obtained on Intel Berkeley Research Lab data indicate large improvements in energy-efficiency and computational complexity when the proposed method is used for landmark-node tracing.