Impacting the coding-spreading trade-off in multiuser CDMA with iterative MMSE demodulation

Abstract

Iterative minimum mean square error (MMSE) demodulation of coded direct sequence code division multiple access (CDMA) uplink channels suppresses multiuser interference over a wide range of user signal powers, by iteratively exchanging soft information between an MMSE multiuser detector and a bank of single-user decoders. The iterative feedback of soft information from the decoders allows the demodulator to suppress (via the MMSE paradigm) and subtract (as in parallel subtractive cancellation) interference from other users in the same cell. It also displays remarkable near-far resistance by functioning like a successive interference subtractor across iterations. Using multiple sensors at the demodulator improves its baseline performance and provides increased user capacities now via both array processing and refinements across iterations. Most remarkably, iterative and array processing allow increasing the ratio of the bandwidth expansion incurred by coding to that incurred by spreading. This is a very useful property in practical systems where user populations typically vary with time. Lower rate codes (with correspondingly lower spreading factors, to satisfy restrictions on the allowable total bandwidth expansion) are obviously helpful in lightly loaded systems. With the proposed (nonlinear) demodulation technique they also aid in improved in-cell user separation via soft-decoded feedback to the detector in the second and subsequent iterations, along with the additional initial spatial separation provided by multisensor arrays. They also serve to mitigate the effects of residual in-cell and out-of-cell interference in the final decoded estimates.