Filterbank transceivers optimizing information rate in block transmissionsover dispersive channels

Optimal finite impulse response (FIR) transmit and receive filterbanks are derived for block-based data transmissions over frequency-selective additiv e Gaussian noise (AGN) channels by maximizing mutual information subject to a fixed transmit-power constraint. Both FIR and pole-zero channels are con sidered, The inherent flexibility of the proposed transceivers is exploited to derive, as special cases, zero-forcing (ZF) and minimum mean-square err or receive filterbanks, The transmit filterbank converts transmission over a frequency-selective fading channel, affected by additive colored noise, i nto a set of independent flat fading subchannels with uncorrelated noise sa mples. Two loading algorithms are also developed to distribute transmit pow er and number of bib across the usable subchannels, while adhering to an up per bound on the bit error rate (BER), Reduction of the signal-to-noise rat io (SNR) margin required to satisfy the prescribed BER is achieved by codin g each subchannel's bit stream. The potential of the proposed transceivers is illustrated and compared to discrete multitone (DMT) with simulated exam ples.