Adaptive Signal Processing for Multi-User Wireless Communications

This dissertation is concerned with signal processing for communications. The first part of the dissertation considers the problem of enhancing data throughput and radio link reliability with limited resources (including time, frequency and power) through multiple-user multiple-antenna systems. By allowing multiple data streams (to multiple receivers) to occupy the same channel, multi-user systems can potentially provide significant system throughput improvement. In this dissertation, we developed multi-user multi-antenna techniques to suppress various severe interferences in multi-user systems. We also developed user selection schemes to enhance link reliability and improve system throughput, by efficiently exploring the multi-user diversity and spatial isolation between any two users which is measured by the (normalized) inner product of their channel vectors. The results in this dissertation show that, through advanced interference suppression and proper user selection, data throughput and link quality in wireless systems can be improved significantly. Both single-carrier and multi-carrier schemes are considered here for multi-user multi-antenna systems. For single-carrier systems, frequency-domain multi-antenna signal processing techniques are investigated to suppress both multi-user CCI (co-channel interference) and ISI (inter-symbol interference). Our results show that the combination of frequency-domain signal processing and multi-antenna technique provides an effective and low-complexity scheme for suppressing severe interference in multi-user systems. As such, with a given BER performance constraint, the throughput of a multi-user system can grow linearly with the number of users if multiple data stream transmission over the same channel is allowed . For multi-carrier multi-user systems, our research focused on the issue of user selection. The results show that we can exploit wireless channels more efficiently through optimal user and subchannel allocation according to the time-varying nature, multi-user diversity and spatial isolation. Specifically, we show that, one can improve the system throughput of wireless systems by up to 100% through two-user schemes with reasonable complexity. The second part of this dissertation is concerned with parameter estimation through adaptive filtering. Set-Membership Filtering (SMF) is studied and frequency-domain SMF is developed. Adaptive frequency-domain SMF algorithms are derived and applied to frequency-domain equalization in broadband wireless communications. Results show that frequency-domain SMF can effectively improve the performance and reduce the implementation complexity of wireless communications. The problem of designing robust filters in impulsive noise is also investigated with the employment of SMF. We developed an ellipsoidal SMF algorithm to detect and suppress the outliers due to impulsive noise, which usually leads to major distortion in adaptive filtering. The problem of SMF is also formulated in the framework of general M-estimation and SMF-type of M-estimation algorithms are proposed. Simulation results show that the proposed algorithms offer robust filtering in impulsive noise and outperforms conventional least-squares methods.