Unscented Particle Filter for Channel Estimation of OFDM Communication Systems

The paper estimates the nonlinear, non-Gaussian channel response for OFDM wireless communication system. To achieve this, we adopted a simplified version of the unscented particle filter which is basically a particle filtering algorithm that incorporates an unscented Kalman filter for the generation of its importance density function. For linear and Gaussian state and measurement models, Kalman filter is the most suitable one. For nonlinear and Gaussian models unscented Kalman filter can be used to obtain the best solution. But systems with linear and Gaussian assumptions do not hold for real-life applications. The unscented particle filter that we introduce can give the closest approximation of the channel response for any nonlinear, non-Gaussian OFDM system.