Performance analysis of an mimo channel esimator based on superimposed training and first-order statistics

Channel estimation for multiple-input multiple-output (MIMO) time-invariant channels using superimposed training is considered. Recently in X. Meng and J.K. Tugnait (2004), the first-order statistics-based approach of J.K. Tugnait and Weilin Luo (2003) was extended to multiuser systems (where semiblind versions using linear MMSE equalizers or Viterbi detectors were also presented). In this paper we present a performance analysis of the approach of X. Meng and J.K. Tugnait (2004) to obtain a closed-form expression for the channel estimation variance. We then address the issue of superimposed training power allocation for complex Gaussian random (Rayleigh) channels for MIMO systems arising from spatial multiplexing of a single user signal. Illustrative simulation examples are provided