Statistical Precoding and Detection Ordering in MIMO Multiple-Access Channels with Decision Feedback Equalization

We present a novel approach for joint transmitter-receiver design in the uplink of a wireless multiple-input multiple-output communication system. It applies to, e.g., a fast-fading frequency-division duplexing system with periodic pilot signaling from each user - a scenario hindering transmitter optimization based on channel state information (CSI), while CSI-based receiver optimization is possible. Each user multiplexes data onto several, independently coded subchannels processed by a linear precoder, and detected at a base station (BS) employing zero-forcing decision feedback (DF) equalization, eliminating all interference prior to detection. We target the problem of jointly designing fixed linear precoders for all users as well as a fixed detection order for the DF receiver based on long-term channel statistics. We propose an efficiently implementable alternating-minimization technique that is verified numerically to converge fast, and to outperform the popular V-BLAST scheme - a computationally more complex ordered-DF receiver with limited applicability by requiring equal-rate subchannels in the system.