Efficient disjoint boundary detection algorithm for surveillance capable WSNs

Abstract

Wireless Sensor Networks (WSN) have given a new approach for applications such as surveillance, tracking, and monitoring. In such monitoring systems, the sensor nodes at the network boundary require remaining vigilant to detect events like objects entering and leaving the area under surveillance. Nodes performing surveillance of target area can be classified as boundary node and others as an interior node. Due to continuous vigilance, boundary node suffers from quicker energy exhaustion (compare to internal nodes) which results to a shorter life span of the node. None of the reported works on boundary detection algorithms have considered the problem of maximizing the lifetime of boundary nodes. The problem of maximizing the lifetime of boundary nodes collectively addresses the problem of an optimal number of boundary nodes detection and maximization of boundary node’s lifetime. In this work, we formulate the problem of the maximum lifetime of the boundary as the problem of partitioning the boundary region of the network into disjoint dominating sets of a boundary region with the aim of maximizing the number of disjoint dominating sets of a boundary region in the network. This is the problem of the domatic partition of boundary region in graph theory, where each partition is dominating the set of boundary region (i.e., not the dominating set of entire network). Rotating the disjoint dominating sets of boundary region in the network resulting into maximizing the lifetimes of boundary nodes. We found that our disjoint dominating sets of boundary region algorithm have an approximation ratio ((4.8+ln5)opt+1.2), where opt is the minimum size dominating set which has the same approximation ratio as the minimum dominating set problem. The time complexity of boundary detection algorithm is O(mn2) where m=|Euk∖1|, n=|Nuk∖1| and k is the radius of graph. Simulation results reveal that using our proposed domatic partition of boundary region algorithm, the boundary node saves 35.71% energy lesser as compared to without domatic partition based boundary detection algorithm. To support our claim, we extended our work one step further and performed the experimental analysis of our proposed algorithm on telosb motes.