New approaches to multiuser detection in DS/CDMA systems

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Electrical Engineering and Computer Science


Pramod K. Varshney


Decorrelators, Log-likelihood ratio quantizer, Multiuser detection, DS-CDMA

Subject Categories

Electrical and Computer Engineering | Signal Processing


The conventional optimal detector for DS/CDMA systems consists of a bank of matched filters. This detector suffers from severe degradation if the interfering signal is not orthogonal to the desired one. The optimum multiuser detection method is based on a maximum likelihood sequence detection formulation. Its intensive computational complexity makes this detector impractical. A variety of sub-optimum detection methods have been proposed over the years, such as decorrelating detection, MMSE detection, successive interference cancellation detection and decision feed-back detection. In addition, coding techniques have been considered for improved performance, such as convolutional coding and Trellis-coded modulation.

The goal of this dissertation is to develop improved suboptimal multiuser detection algorithms. We concentrate on three distinct approaches to the multiuser detection problem. One approach is based on the use of equi-correlated signature codes. The proposed receiver for the equi-correlation-based multiuser communication (ECBMC) scheme has low computational complexity. By using the equality of cross-correlations, the proposed ECBMC scheme can completely eliminate multi-access interference in a synchronous single-path DS/CDMA network. System performance is independent of the number of active users. The knowledge needed at the receiver is only the spreading sequences of the user being demodulated. The scheme has been extended to include multipath situations.

Another approach extends the decorrelator. It is based on the fact that the decorrelator outputs corresponding to each user have different reliability. These decorrelated signals for each user are first tested to determine whether or not a final decision can be made with high reliability. For users with high reliability, a hard decision is made at this stage. For the remaining users, a conditional quantization (CQ) algorithm is employed before a decision is made. Significant performance enhancement can be achieved over the conventional decorrelator at the expense of a modest increase in computational complexity.

We also present a multiuser detection scheme based on cell diversity for DS/CDMA systems. This scheme is employed when multiple base stations can detect the desired signals. Each cooperating base station computes the decision statistic for each user based on multiuser detection methods, quantizes it and transmits it to a central processor. The quantizer is a MMSE log likelihood ratio quantizer. Based on these received samples, the central processor determines the received bit for each user based on mutisensor signal detection theory. This system with cell diversity is shown to perform better than system without diversity.


Surface provides description only. Full text is available to ProQuest subscribers. Ask your Librarian for assistance.