Low complexity equalization approaches in wireless communications

In digital communications, channel equalizers are effective means of combating inter-symbol interference (ISI) introduced by the linear distortive channels. As data rate is growing rapidly in modern wireless systems to support various applications from voice to multimedia, design of high speed low complexity channel equalizers has become more and more important. This dissertation focuses on several low complexity high performance channel equalization approaches for wireless communications.; We first propose a bi-directional decision feedback equalizer (Bi-DFE) with contradictory block arbitration (CBA). Our approach resolves the contradictory blocks in the hard output sequences from the forward and reverse DFEs based on the principle of maximum likelihood sequence estimating. While remaining at linear computational complexity, Bi-DFE under our arbitration strategy can achieve significant BER improvement over the classical single-directional DFE as well as the Bi-DFE with linear soft output combining.; Next, we study channel equalization strategies for a specific 3G mobile system, namely, EDGE. After investigating linear equalization, decision feedback equalization and suboptimal MLSE equalization, we adopt turbo equalization to further improve the detection performance for EDGE signals. Delayed decision feedback (DDF) algorithm is used to reduce the complexity of equalizer trellis. A non-binary DDF-SOVA equalizer is developed to calculate the soft values of EDGE's 8-PSK symbols. Bi-directional processing is also employed to enhance the output reliability of the DDF-SOVA equalizer.; The third contribution of this dissertation is a new differential correlation approach (DCA) to blind DPSK symbol estimation based on second order statistics. This method simplifies Tong's Viterbi algorithm based scheme, and is applicable to a class of time-varying channels. Besides, we show that Tong's scheme suffers from error propagation.; Finally, we present an additional work on CDMA multiuser detection. Semidefinite programming with relaxation (SDR) algorithm has been reported recently as a promising alterative to ML multiuser detector for its low complexity. In our study, we extend the BPSK/QPSK originated SDR to more generic QAM signal detection. An example of 16-QAM multiuser signal detection is provided to demonstrate the superior performance of SDR.