Optimal pre- and postfilters for multichannel signal processing

The authors present closed-form solutions to the problem of optimal pre-filtering and postfiltering for transmission of an N-dimensional signal though a communications channel composed of M identical subchannels with uncorrelated additive noise. The optimal system may include both Hadamard and Karhunen-Loeve transforms as factors of the prefilters and postfilters. The error improvement due to optimal prefiltering and postfiltering can be as high as 6 dB, as compared to optimal postfiltering only, for a first-order Gauss-Markov signal, for example. >