Citation

Abstract

The topic of this dissertation is a digital multiuser communications system whose structure is identical to a single cell of a typical cellular telephone system. It consists of several spatially distributed portables (users) which communicate simultaneously with a central base station. The general objective is to investigate and improve the central receiver which detects the signals of all in-cell portables.

In order to support several users, a multiple access scheme combining frequency diversity (SSMA - spread spectrum multiple access) with receive antenna diversity (SDMA - space division multiple access) is considered. The system is designed for high data rates, frequency selective and quasi-stationary radio channels.

A vector model of the system is developed, which incorporates frequency diversity as well as multiple receive antennas. In addition to simplifying the mathematical analysis, it provides precious insight into the behavior of the system.

A new approach to approximate and bound the error probability in linear systems is described, which is accurate, numerically very efficient and easy to use. The results show that it compares very well with the existing state-of-the-art approximations.

Promising detectors for multiuser systems are numerically efficient and take the signals of all users into account for the estimation process. One of the candidates strongly considered for a practical implementation in future systems is the equalizer family with multiple inputs (from multiple receive antennas) and multiple outputs (one for the signal of each user). This detector type is analyzed here. Although significantly less complex than the optimum detectors, multiple-input multiple-output (MIMO) equalizers require a total number of operations which is too heavy a burden for present systems. For that reason, numerically more efficient approaches for the optimization of the equalizer coefficients are considered.

In addition, an extension of the standard MIMO equalizer with decision feedback is analyzed. This detector addresses the special situation of cellular systems in which the received signal strengths differ significantly. It is shown that the inclusion of simple delay elements may strongly reduce the requirements for power control, a technique that adds considerable complexity but crucial to current cellular systems.