Performance Analysis of Single and Multiple Estimation in MIMO Rician FrequencyFlat Fading Channels

Abstract

In this paper, the performance of the single-estimation (SE) and multiple-estimation (ME) is investigated in multiple-input multipleoutput (MIMO) Rician frequency-flat fading channels using the maximum likelihood (ML) technique, the new shifted scaled least squares (SSLS) estimator, and the minimum mean square error (MMSE) estimator. The closed form equations are obtained for mean square error (MSE) of the estimators in SE and ME cases under optimal training. Analytical and numerical results show that the ML estimator has lower error in the case of ME than SE. Moreover, it is seen that the performance of SSLS and MMSE channel estimators in the ME case is better than SE particularly at high signal to noise ratios (SNRs) and for weak line of sight (LOS) propagation paths. At low SNRs and for strong LOS propagation paths, and also for small numbers of sub-blocks used for channel estimation, the SSLS estimator is better than ML as well as the MMSE estimator is better than both estimators but in other cases the ML estimator is better than SSLS (MMSE). The un-equal power
allocation is also investigated analytically and numerically. Simulation results show that exponential power allocation is an appropriate method in ME case.