The Modified Bayesian Cramer-Rao Bound for MIMO Channel Tracking

Abstract

The modified Bayesian Cramer-Rao bound (MBCRB) to the mean squared error (MSE) of multi-input and multi-output (MIMO) channel estimation under discrete nuisance parameters are studied in this work. A recursive formula is provided for the computation of the MBCRB in time-varying MIMO Rayleigh fading channels. Under the assumption that channel coefficients of the MIMO system are independent and identically distributed, the MBCRB for MIMO channel estimation is shown to be the same to the single-input and single-output (SISO) system. Based on the MBCRB, the MSE of Kalman channel tracking is investigated for various MIMO channel configurations with different orders of the channel model.