Acoustic target localization in sensor networks with FUZZYART

A low cost target localization algorithm is developed for wireless sensor networks based on acoustic signals. When an object is detected, each sensor will start delay calculations to estimate the TDOA (time difference of arrival) of the target in hand with respect to each sensor. Each sensor will exchange the delay information between each other. Then the sensors will reach a general consensus of where the target is located in the sensor field using triangulation, assuming each sensor is aware of its own location and that the data transmission is synchronized. Furthermore a FUZZYART neural network model is introduced at the cluster head to detect errors and assign probabilistic weight to each estimate to cope with spatial correlation and consensus vote. The algorithm proposed performs triangulation with low overhead in computations. Moreover simulation results show that the error in target localization averages around 23 cm and the maximum is 35 cm which is acceptable in sensor network environment.