Constrained Cross Entropy Localization Technique for Wireless Sensor Networks

Abstract

Challenges in wireless sensor networks (WSNs) localization are diverse. Addressing the challenges in cross entropy (CE) localization utilizing cross entropy optimization technique in turn minimizes the localization error with a reasonable processing cost and provides a balance between the algorithmic runtime and error. The drawback of such minimization commonly known as flip phenomenon introduces errors in the derived locations. Beyond CE, the whole class of localization techniques utilizing the same cost function suffers from the same phenomenon. This paper introduces constrained cross entropy (CCE), which enhances the localization accuracy by penalizing the identified sensor nodes affected by the aforementioned flip phenomenon in the neighborhood through neighbor sets. Simulation results comparing CCE with both simulated annealing- (SA-) based and original CE localization techniques demonstrate CCE's superiority in a consistent and reliable manner under various circumstances thereby justifing the proposed localization technique.