Optimising the barrier coverage of a wireless sensor network with hub-and-spoke topology using mathematical and simulation models

Abstract

Deploying wireless sensor networks (WSNs) to form a barrier that provides surveillance against illegal intruders (or targets) is a fundamental sensor-allocation problem. In this paper, we consider a WSN barrier coverage problem with hub-and-spoke topology in which communication hubs form central hubs that are connected to sensors via spokes. We develop and compare three models. The first two are mathematical models: an Integer Non-Linear Program (INLP) and an Integer Linear Program (ILP). The third is an Optimisation-via-Simulation (OvS) model, which comprises agent-based simulation and heuristics. We consider the following factors in the models: multiple sensor and target types, probabilistic detection function, sensor reliability, communication range, communication interference, network topology and budget constraints. The experiment shows that the INLP solutions are close to ILP global optimum solutions. The ILP model is only practical for a small problem, while the INLP and OvS models can cope with bigger ones. OvS can handle more realistic assumptions more easily, such as intelligent targets that can move freely across the barrier, and the movement speed depends on the path taken. To solve this using ILP, we need to reformulate the ILP model but it is very challenging. Hence, our contributions are twofold: (1) our models are more elaborate than other models in the literature and (2) this is the first work that demonstrates how OvS is used to solve a barrier coverage problem and demonstrates its benefit of handling more realistic assumptions.