Adaptive joint multiuser detection and channel estimation in multipath fading CDMA channels

The problem of joint multiuser detection and channel estimation in frequency-selective Rayleigh fading CDMA channels is considered. First the optimal multiuser detector for such channels is derived, which is seen to have a computational complexity exponential in the product of the number of users and the length of the transmitted data sequence. Two suboptimal detectors are then developed and analyzed, both of which employ decorrelating filters at the front-ends to eliminate the multiple-access interference and the multipath interference. The symbol-by-symbol detector uses a Kalman filter and decision feedback to track the fading channel for diversity combining. The per-survivor sequence detector is in the form of the Viterbi algorithm with the trellis updates being computed by a bank of Kalman filters in the per-survivor fashion. Both suboptimal detectors require the knowledge of all waveforms of all users in the channel and the channel fading model parameters. Adaptive versions of these suboptimal detectors that require only the knowledge of the waveform of the user of interest are then developed. The adaptive receivers employ recursive-least-squares (RLS) minimum-mean-square-error (MMSE) filters at the front-end to mitigate the interference, and use a bank of linear predictors to track the fading channels. It is shown that the front-end RLS-MMSE filters can be implemented using systolic arrays to exploit massively parallel signal processing computation, and to achieve energy efficiency. Finally, the performance of the suboptimal detectors and their adaptive versions are assessed by simulations.