Localization of objects using stochastic tunneling

This paper presents a novel wireless localization scheme in the three dimensional domain that employs stochastic optimization with tunneling transformation to recursively estimate the location of wireless tags in a network from pair wise signal strength measurements. Spatially co-located wireless tags, receiving signals from a common transmitter, exhibit correlation in their Received Signal Strength Indicator (RSSI) values. Hence in a network of wireless tags, with pair wise correlation coefficients available, posterior distribution of the unknown tag separation is used to relatively localize them using maximum a posteriori (MAP) Estimator. However, due to the non-convex/non-tractable nature of this posterior distribution, deterministic optimization methods will end in one of the many local maxima unless the initial guess is close to the region of attraction of the global maximum. In this paper, a novel stochastic localization method called LOCalization Using Stochastic Tunneling (LOCUST) is proposed which utilizes constrained simulated annealing with tunneling transformation to solve this non-tractable posterior distribution. The tunneling transformation allows the optimization search operation to circumvent or “tunnel” through ill-shaped regions in the posterior distribution resulting in faster convergence to global maximum. Finally, simulation results of our localization method are presented.