Investigation of Sign Of Innovation - Particle Filter tracking performance in ad hoc noisy binary WSN

Abstract

This paper deals with target tracking in a binary Wireless Sensor Network (WSN) context. In particular, we propose to study the performance of the Sign Of Innovation Particle Filter (SOI-PF) algorithm in a noisy context. This parallel algorithm based on the exchange of only one bit by instant between the sensors, has shown its efficiency for target tracking in a highly non linear and noiseless framework. For the proposed purpose we consider that the communication between sensors is done via noisy and fading channels, thereby placing the study in a realistic communication context. The major difficulty in our case is that the sensors don't receive the same observation since the communication channel differ from a pair of sensors to an other. Based on exhaustive simulations, we notice that the SOI-PF preserve its good tracking ability for Signal to Noise Ratio (SNR) value upper than 15dB. For SNR below this value, we succeed to improve the SOI-PF's performance by assigning a fixed sensor to perform the scheduling task for selecting the active sensors.