Linear MMSE MIMO Channel Estimation with Imperfect Channel Covariance Information

Abstract

Abstract-In this paper, we investigate the effects of imperfect knowledge of the channel covariance matrix on the performance of a linear minimum mean-square-error (MMSE) estimator for multiple-input multiple-output (MIMO) channels. The estimation mean-square-error (MSE) is analytically analyzed by providing both a very tight lower bound and an upper bound. The proposed analysis is useful for the understanding of how estimation accuracy of the channel covariance matrix impacts on system performance, depending on the average signal-to-noise ratio (SNR) and specific propagation conditions. Conclusions are fully supported by numerical results.

On the Effect of Imperfect Channel Estimation upon the Capacity of Correlated MIMO Fading Channels

“…We also note that in the medium SNR regime the higher the correlation the smaller the difference C CSIR − C lo . This behavior is a consequence of the LMMSE estimator that provides smaller estimation errors over channel with higher correlation [12].…”

Mean square error analysis and linear minimum mean square error application for preamble‐based channel estimation in orthogonal frequency division multiplexing/offset quadrature amplitude modulation systems

“…From (26), it can be noted that the product I 2 m,n (I * m,n+1 ) 2 yields a quadruple sum with intrinsic interference terms k gl m,n p,q raised to the fourth power. The expectation of all the cross-factors [i.e.…”

“…Since the noise component in B m,n = I m,n + W m,n is zero mean and uncorrelated with I * m,n+1 , therefore E{Re{B 2 m,n (I * m,n+1 ) 2 }} in (25) is equal to E{Re{I 2 m,n (I * m,n+1 ) 2 }}. From (26), it can be noted that the product I 2 m,n (I * m,n+1 ) 2 yields a quadruple sum with intrinsic interference terms k gl m,n p,q raised to the fourth power. The expectation of all the cross-factors [i.e.…”

Pilot Contamination Mitigation for Wideband Massive MMO: Number of Cells vs Multipath

“…γ pr kis is expressed in (26) whereZ = (m,j) g mji X * mj X T mj + σ 2 I. Specifically, when g kii g mji which is the case in practice and N → ∞ (or N LKN c ), the proposed approach achieves the SINRγ pr kis given in (27). Furthermore, when γ pe kis 1 and γ pr kis 1, we will have…”

“…We would like to point out here that such a problem can potentially be addressed by leveraging the techniques used in the robust beamforming designs (see for example[25]-[27]). …”