Iterative Nonlinear Detection of MIMO Signals using an MMSE- OSIC Detector and Adaptive Cancellation

We present an improved algorithm for iterative detection of the received signal in a convolutionally encoded wireless MIMO system. The method is a combination of an iterative linear MMSE equalization and an enhanced version of the nonlinear ordered successive interference (OSIC) cancellation. To improve regular OSIC we introduce thresholdbased adaptive cancellation. The threshold is switched and takes reliability of a precedent symbol decision into account. By this, the number of cancellations of wrongly decided symbols is reduced to a large extend. In addition, the MAP soft output demapper takes the so called Post Detection Interference and Noise (PDIN) into account. The variance of the PDIN is derived. As a result, we obtain an SNR gain of 3 dB at a BER of 10−4 compared to existing solutions for uncorrelated Rayleigh fading MIMO channels with additive white Gaussian noise. For a correlated channel we obtain a large improvement for the first pass of about 5 dB at a BER of 10−3.