MMSE interference suppression for joint acquisition and demodulation in CDMA systems

Minimum mean squared error (MMSE) demodulation for direct-sequence CDMA systems eliminates the near-far problem, and can be implemented adaptively (i.e., without explicit knowledge of the parameters of the multiple-access interference), given a training sequence for the desired transmission. However, prior to timing acquisition, the receiver does not know the phase of the training sequence, i.e., it does not know, for a given observation interval, which bit of the training sequence contributes the most signal energy. Conceivably, this timing information could be obtained using conventional acquisition techniques by correlating over long enough intervals and applying enough power control to resolve the near-far problem. In this paper, however, we present an adaptive approach to the problem of near-far resistant joint acquisition and demodulation. Our method is to use a training sequence with a short period P, and run P adaptive algorithms either serially or in parallel, one for each assumed phase of the training sequence. The adaptive algorithm that yields the least mean squared error (MSE) corresponds to the correct phase, and yields in addition an MMSE correlator that can be used for continued training or for decision-directed adaptation. Thus, acquisition results in a near-far resistant demodulator that implicitly accounts for the timings and amplitudes of all the transmissions without explicitly estimating even the timing of the desired signal. We note that a method for joint acquisition and demodulation that does not require a training sequence has also been devised