A robust diffusion algorithm using logarithmic hyperbolic cosine cost function for channel estimation in wireless sensor network under impulsive noise environment

In a wireless sensor network (WSN), the performance of the error-squared based adaptive channel estimation algorithm degrades in the presence of impulsive noise. Robust methods are used to minimize the impulsive noise impact at the cost of a slow convergence rate. We propose a robust diffusion adaptive channel estimation algorithm using the logarithmic hyperbolic cosine cost function assuming that the received signal at each node is corrupted by impulsive noise. The sensor nodes experiencing the common channel with the base station are grouped using their initial channel estimates. After grouping, the robust diffusion algorithm is applied in each group to estimate the corresponding channel. The steady-state mean square deviation (MSD) and excess mean square error (EMSE) of the robust algorithm are derived. The algorithm's steady-state and convergence rate performance are simulated for Rayleigh fading channel corrupted by Bernoulli-Gaussian distributed impulsive noise. The simulation results demonstrate that the proposed algorithm gives faster convergence without compromising steady-state performance compared to other robust algorithms.