Performance assessment of polynomial expansion detectors in MIMO channels with measured data

Polynomial expansion detectors introduced by S. Moshavi et al. (see Int. J. Wireless Inf. Networks, vol.3, no.1, p.1-17, 1996) combined with recent results in random matrix theory were found to be powerful low-complexity tools to mitigate interference on MIMO communication channels (see Muller, R.R. and Verdu, S., IEEE J. Selected Areas in Comm. vol.19, no.8, p.1429-41, 2001). The random matrix model of communication via antenna arrays characterizes the channel in terms of its asymptotic eigenvalue distribution by a single parameter. Muller (see Proc. European Personal Mobile Comm. Conf., 2001) generalises the polynomial expansion of Muller and Verdu to MIMO communication channels promising very good performance at complexity per bit that grows only linearly with the number of antennas. The reliability of Muller's results strongly depends on the accuracy of the used asymptotic random matrix model to describe the antenna array channel. We provide measured performance results for the receiver in terms of bit error rate and signal-to-interference ratio for communication over real-world indoor MIMO channels. Computer simulations show that, for more than 5 antennas, performance with the measured data is close to the prediction found with the asymptotic random matrix models used by Muller and only a little behind the performance of far more complex receivers, such as the linear MMSE receiver, that involve matrix inversions to combat crosstalk.