Rutgers University

Department of Electrical and Computer Engineering

Ph.D. Thesis Abstract

*Power Control, Multiuser Detection and Interference Avoidance in
CDMA Systems*

by

Sennur Ulukus

In CDMA systems, users create interference to each other to the extent of the correlation between their codes and the ratio of their received powers. Consequently, near-far effects in which strong users with higher received powers can overwhelm the communication quality of the weak users with lower received powers must be mitigated. This thesis presents techniques that lessen the effects of near-far problem through power control, multiuser detection, and interference avoidance.

For general signature sequence sets, the optimum multiuser detector has exponential computational complexity in number of users. This thesis identifies a class of signature sequence sets for which the optimum multiuser detector has a polynomial complexity. Because of the exponential complexity of optimum multiuser detector, many suboptimum multiuser detectors have been developed. Among them, the decorrelating detector is a linear suboptimum detector which suppresses the multiaccess interference totally. This thesis presents a blind adaptive implementation for the decorrelating detector.

Power control is a resource allocation technique that balances the received powers of the users so that no one user creates excessive interference. Deterministic power control, which assumes the availability of perfect measurements of parameters such as signal to interference ratio, received interference power, or bit error rate, has been well studied; however, almost no effort has been made to analyze practical power control algorithms that must extract measurements from the received signal. This thesis analyzes the convergence properties of a class of stochastic power control algorithms that use readily available random observations.

Even when perfect power control is applied, the interference can be further suppressed by multiuser detection. Similarly, even when optimal linear multiuser receivers are used to maximize the signal to interference ratio, power control can be used to further diminish the near-far effect. This thesis introduces a combined power control and multiuser detection algorithm in which communication between the users and the base stations are optimized by controlling both the linear receiver lters and the power levels of the users.

This thesis introduces an iterative and distributed interference avoidance algorithm that operates by updating the signature sequences of the users. When users perform signature updates sequentially, the signature sequence set as a whole converges to an optimum set which maximizes the capacity of a single cell CDMA system.

Ph.D. Dissertation Director: Professor Roy D. Yates

September, 1998