Chaotic signal extraction in wireless sensor networks with unknown statistics

For solving the issue of the chaotic signal extraction in wireless sensor networks (WSNs) with unknown statistics, we present a new algorithm based on a cost reference cubature particle filter (CRCPF) in this paper. The CRCPF uses the cubature-points rule to obtain prediction particles before propagation and resampling, meanwhile, it realizes particle selection and propagation by the user-defined cost and risk functions. Simulation results confirm that the proposed algorithm can extract chaotic signal effectively in the situation of the probability distribution of the noise is unknown.