Multiple Node Localization in Cognitive Radio-Based Wireless Sensor Networks Using Grid Search

The knowledge of location information of multiple co-channel emitting nodes existing simultaneously is very common and important in wireless sensor networks as it can determine the exclusion region of the wireless networks and helps efficient spectrum utilization, thus providing more value-added services. The work in this paper explores the problem of using received signal strength (RSS) measurements taken by a network of mobile sensor nodes to estimate the locations of multiple emitting nodes in a given geographical area. For localization, we consider applying the maximum likelihood (ML) estimation algorithm based on grid search technique that requires an analytical expression to summands of log-normal random variables for likelihood calculations at each grid location. Since there is no closed form expression for the sum of log-normal distribution, we investigate log-normal distribution approximations studied in literature and propose our own probability density function approximation to the sum of log-normals based on distribution fitting to a set of simulated random variables. We evaluate localization performance of grid search technique using the proposed approximation and the well-known Fenton-Wilkinson’s approach which is known to fail at large dB spread values. Our simulation results show that the proposed approach addresses the problem of Fenton-Wilkinson’s approach and gives higher location estimation accuracies in all dB spread values.