Unscented Kalman filter with fading wireless channel

This paper considers the error behavior of the discrete-time unscented Kalman filter connected to the general nonlinear system via a wireless sensor. The measurements are transmitted through a noisy fading wireless channel where both transmission failure and signal fluctuation are considered. Sufficient conditions of statistical convergence property for the estimation error covariance is established. It is shown that the expected error covariance with respect to fading process is bounded and converges to a steady state value if the system satisfies some assumptions. Numerical example is given to illustrate the effectiveness of the developed techniques.