Blind SNR Estimation Based on Multistage Wiener Filter

In this paper, an algorithm based on multistage wiener filter (MSWF: Multistage Wiener Filter) is proposed for blind SNR estimation. Combined with the signal subspace decomposition theory, the noise contained subspace decomposition can be achieved by CSA-MSWF (CSA: Correlation Subtraction Algorithm) forward iterations, this avoid the complex decomposition of the signal autocorrelation matrix in classical methods, so the power of the signal and noise could be obtained in this way, also the SNR estimation values. Simulation in Additive Gaussian noise channel (AWGN) proved that the performance of this approach is slightly better than traditional methods, and while the true SNR is -7~25dB, the standard deviation of SNR estimation is less than 0.5dB. When the codes received are 100 and the actual SNR is 10dB, the biases of SNR estimation are smaller than 0.35dB for all modulation types simulated here, it shows the rapid ability of blind SNR estimation in the small samples supported environment of the estimator.