Node localization using Particle Swarm Optimization

Recent advances in developing sensor nodes have made the use of Wireless Sensor Networks (WSNs) possible in many different scenarios, most of them require each node to have some information about its geographical location. This paper presents a two-step distance-based algorithm to sensor network localization carried out in a centralized architecture. The first phase of the algorithm utilizes an improved version of the DV-distance method used to provide coarse position estimates for all the nodes. During the second phase, Particle Swarm Optimization (PSO) is performed to fine tune and obtain accurate estimation of the locations. In the second phase, several techniques are also used to address the main problems of localization such as flip ambiguity, collective translation and error propagation. To evaluate the performance of the algorithm, numerical simulations were performed and the results were compared with similar distance-based methods, namely one-phase simulated annealing (SA), trilateration and simulated annealing (TSA) and semi-definite programming localization (SDP). Results demonstrate that our proposal achieves a significant performance improvement in comparison to other related methods especially in networks with low connectivity.