Vector Viterbi algorithm with structured noise covariance matrix estimation for smart antennas

The problem of improving the noise covariance matrix estimation with limited training data to improve the performance of the vector Viterbi algorithm (VVA) in the space-time equalizer for mobile communications is addressed. We exploit the array geometry to improve the noise covariance matrix by constraining the structure during the estimation procedure. A Toeplitz structured covariance estimate based on the periodogram of the noise residuals is used for the covariance matrix inversion in the VVA. Finally, computer simulation results are presented for a simple vector channel model neglecting the multipath effects. The Reed Mallet Brennan (1974) SNR loss factor is also calculated to justify how the VVA performance improves because of the structured noise covariance estimation