Localization in wireless sensor networks by constrained simultaneous perturbation stochastic approximation technique

Localization of sensor networks poses an immense challenge and is considered as a hot research topic in recent days. To address the accuracy on localization this paper proposes constrained simultaneous perturbation stochastic approximation (SPSA) based localization techniques for wireless sensor networks. A simple centralized localization of the non-anchor nodes based on minimizing the summation of the estimated error of all neighbors is the basic building block of the proposed localization technique. This category of localization technique incurs errors often referred as flip ambiguity. The improvement of the simple SPSA based localization is made by modifying the algorithm to a constrained optimization technique using penalty function method where the correction on the flipped node is made by penalizing the identified flips by the penalty function. Simulation results demonstrate the superiority of the proposed SPSA algorithm compared to its closest counterpart, namely, the simulated annealing (SA) based localization algorithm.