Exponential-weighting-basedmaximum likelihood for determining measurement random latency probability in network systems

The standard nonlinear Gaussian approximation (GA) filter used for randomly delayed measurement in network systems, usually assume that the measurement random latency probability (MRLP) is known a priori. However, practically, the MRLP may be unknown or even be time-varying, causing the standard nonlinear GA filter to certainly fail. Motivated by the above situation, this paper is concerned with the application of maximum likelihood based on exponential weighting for adaptively determining the MRLP. Furthermore, an adaptive version of the nonlinear GA filter is proposed for joint state estimation and MRLP determination. Finally, simulation results demonstrate the performance of the new adaptive GA filter compared with the standard one.