Node placement for connected target coverage in wireless sensor networks with dynamic sinks

Abstract

Target coverage and connectivity are two fundamental and critical issues in wireless sensor networks. The former is for providing sufficient monitoring quality where all points of interest in the network are covered by sensor nodes. The latter is for guaranteeing satisfactory communicating capability where all sensor nodes can connect to at least one sink via nodes (i.e., sensor nodes and relay nodes). Though considerable efforts have been devoted to optimize the placement of sensor nodes and relay nodes under connected target coverage constraint (i.e., guaranteeing both target coverage and connectivity), all of the existing works in this area address only networks with one static sink. In the meanwhile, although there are many works in literature considered the networks with dynamic sinks, none of them studies how to optimize the location of sensor and relay nodes. In this article, we focus on wireless sensor networks with dynamic sinks, which consist of multiple sinks and the sinks’ positions may change periodically, and study how to place a minimum number of nodes for connected target coverage. Specifically, we decompose the problem into two sub-problems. The first one, named as target coverage problem, is to place sensor nodes for covering all targets. The second one, named as network connectivity problem, is for placing relay nodes to connect sensor nodes to the sinks. We first formulate the target coverage problem under an integer linear programming model and present an exact algorithm to determine the optimal solution. We then propose a constant-approximation algorithm based on partitioning and shifting scheme. For the network connectivity problem, we first prove its NP-hardness and then propose two approximation algorithms. The first one exploiting the minimum group Steiner tree to minimize the placed relay nodes, while the second one is a time-efficient algorithm based on clustering and spanning tree approaches. The experiment results show the superiority of our proposed algorithms in terms of both the number of the required nodes and the time complexity.