BAYESIAN DECISION-FEEDBACK TECHNIQUES FOR DECONVOLUTION

Recently, there has been interest in reduced complexity suboptimal MAP symbol-by-symbol estimation for digital communications. In the known channel case, the MAP estimator can be simplified using decision feedback [1], resulting in a family of Bayesian decision feedback estimators (BDFE's); in the unknown channel case, recursive channel estimation can be combined with the BDFE, yielding an algorithm [8] that clearly outperforms blind linear equalizers. However, these algorithms can only be used with estimation lags less than the equivalent channel length and have a complexity that is exponential in the estimation lag. In this paper we propose a new suboptimal estimator suitable for both known and unknown channels. In the known channel case, the MAP estimator is simplified using a form of conditional decision feedback, resulting in a family of Bayesian conditional decision feedback estimators (BCDFE's); in the unknown channel case, recursive channel estimation is combined with the BCDFE. The BCDFE's are indexed by two parameters: a ''chip'' length and an estimation lag. These algorithms can be used with estimation lags greater than the equivalent channel length and have a complexity exponential in the chip length but only linear in the estimation lag. The BCDFE's are derived from simple assumptions in a model-based setting that takes into account discrete signaling and channel noise. Extensive simulations characterize the performance of the BDFE and BCDFE for uncoded linear modulations over both known and unknown (nonminimum phase) channel with severe ISI. The results clearly demonstrate the significant advantages of the proposed BCDFE over the BDFE in achieving a desirable performance/complexity tradeoff. Also, a simple adaptive complexity reduction scheme can be combined with the BCDFE resulting in further substantial reductions in complexity, especially for large constellations. Using this scheme, we demonstrate the feasibility of blind 16QAM demodulation with 10(-4) bit error probability at E(b)/N-o approximate to 18.5 dB on a channel with a deep spectral null.