This dissertation was presented to the Faculty of the Graduate School of The University of Texas at Austin in partial fulfillment of the requirements for the degree of

Ph.D. in Electrical Engineering

A downlink wireless system features a centralized basestation communicating to a number of users physically scattered around the basestation. The purpose of resource allocation at the basestation is to intelligently allocate the limited resources, e.g. total transmit power and available frequency bandwidth, among users to meet users' service requirements. Channel-aware adaptive resource allocation has been shown to achieve higher system performance than static resource allocation, and is becoming more critical in current and future wireless communication systems as the user data rate requirements increase. Adaptive resource allocation in a multichannel downlink system is more challenging because of the additional degree of freedom for resources, but offers the potential to provide higher user data rates. Multiple channels can be created in the frequency domain using multiple carrier frequencies, a.k.a. multicarrier modulation (MCM), or in the spatial domain with multiple transmit and receive antennas, a.k.a. multiple-input multiple-output (MIMO) systems. This dissertation aims to study the system performance, e.g. total throughput and/or fairness, in multiuser multicarrier and multiuser MIMO systems with adaptive resource allocation, as well as low complexity algorithms that are suitable for cost-effective real-time implementations in practical systems.

The first contribution of this dissertation is a general framework for adaptive resource allocation in multiuser multicarrier systems that maximizes the total throughput subject to fairness constraints to enforce arbitrary proportional data rates among users. Whereas the global optimality is computationally intensive to obtain, a low complexity algorithm that decouples the subchannel and power allocation is proposed.

The second contribution concerns precoding using block diagonalization (BD) for single-carrier downlink multiuser MIMO systems. The contribution is twofold. First, it is shown that BD, as a practically realizable precoding technique, can achieve a significant part of the sum capacity achieved by dirty paper coding (DPC), which is optimal. Practical coding schemes that approach the DPC sum capacity, however, are still largely unknown. Second, an upper bound on the ergodic sum capacity gain of DPC over BD in Rayleigh fading channels is derived.

The third contribution concerns low-complexity BD precoding algorithms. Due to the zero inter-user interference requirement imposed by BD, the maximum number of simultaneously supportable users is limited. The brute-force search for the optimal user set, however, is computationally prohibitive for systems with a large number of users. The dissertation proposes two suboptimal user selection algorithms for BD that have linear complexity in the number of users, yet achieve total throughput close to the optimal.

A common characteristic of the resource allocations for multiuser multicarrier and multiuser MIMO systems is that the limited resources shall be allocated among multiple users as well as multiple parallel subchannels. As MCM and MIMO have been widely adopted in various standards, the research in this dissertation contributes to a better understanding of the system performance, and bridges the theory to practical implementations with the proposed low complexity algorithms.