Approximate ML Serial Detector Based on Tomlinson-Harashima Pre-Equalization

In this paper, we propose a novel and simple approximate maximum likelihood detector (A-ML-D) for single input and single output (SISO) systems over frequency-selective fading channels based on Tomlinson-Harashima pre-equalizer. By assuming full channel state information (CSI) at the transmitter side, the pre-equalizer can remove some inter-symbol interference (ISI) at the transmitter. At the receiver, we implement a Gaussian approximation, a pre-whitening filter and a matched filter to realize a set of parallel SISO schemes, and thus, facilitate single symbol detection. The proposed scheme can obtain full multi-path diversity and achieve near-optimal performance with a complexity lower than linear MMSE and MMSE-DFE. Analytical symbol error rate (SER) is derived to further justify the proposed detector.