Improved Distance Estimation with Node Selection Localization and Particle Swarm Optimization for Obstacle-Aware Wireless Sensor Networks

Abstract Sensor-node localization is among the greatest concerns in the field of wireless sensor networks. Range-based localization techniques generally outperform range-free techniques, particularly in terms of their accuracy. Range-based localization techniques depend on a popular distance estimation method, which requires conversion from a received signal strength indicator to distances. In a case where sensor nodes are in an area with obstacles, direct communication between certain pairs of nodes is impracticable; the data must be relayed over multihop (or detour) routes. One promising approach to improve the accuracy of sensor-node distance estimation is to segment (or cluster) sensor nodes to a restricted set of anchor nodes whose estimated distances to unknown nodes are not on a detour route. Some certain topologies can decrease the localization precision; e.g., when each group’s node density is low, large empty spaces (or gaps) might affect the localization precision. If an unknown node is close to another group, using only anchor nodes within its own group could reduce the estimation precision. When anchor nodes within the same group lie along a straight line, the approximation of the unknown-node location could be misinterpreted. Thus, to enhance the localization precision, we make use of anchor nodes in other nearby groups to estimate the locations of unknown nodes. We also apply particle swarm optimization (PSO) with an improved fitness function to estimate the locations of unknown nodes. The localization performance is intensively evaluated in obstacle-prone scenarios. The simulation results show that the proposed scheme achieves higher accuracy than recent state-of-the-art PSO-based methods.