Distributed particle filter using pairwise K-selective gossip for cognitive radio positioning

The cognitive radio positioning system requires a distributed localization algorithm, since the secondary uses (SUs) can not communication with primary users (PUs) directly. The localization algorithm should also be robust to the topology changes of SUs due to the opportunistic access nature. In this paper, we propose a distributed particle filtering method to locate the PU's position. The ranging measurement of PU is obtained through the extended sensing strategy (ESS). The particles for each node is generated locally and transmitted via a K-selective gossip algorithm. Our algorithm is implemented without a topology management scheme and the communication overhead is lower than other distributed particle filter schemes. Simulations demonstrate that the estimation performance of our algorithm is accurate and approach to the centralized scheme with low communication cost.