Lattice-structure based adaptive MMSE detectors for DS-CDMA systems.

Abstract

There has been significant interest in the research community on detectors for DS-CDMA systems. The conventional detector, which detects users ' data bits, by using a filter matched to the users' spreading codes, has two major drawbacks. These drawbacks are (1) its capacity is limited by multiple access interference (MAl) and (2) it suffers from the near-far problem. The remedy to these problems is to use a multiuser detector, which exploits knowledge of users ' transmission and channel parameters to mitigate MAl. Such detectors are called multi user detectors (MUD). A number of these detectors have been proposed in the literature. The first such detector is the optimal detector proposed by Verdu. Following Verdu's work a number of suboptimal detector were proposed. These detectors offer better computational complexity at the expense of the bit error rate performance. Examples of these detectors are the decorrelating detector, the minimum mean squared error detector (MMSE), the successive interference cancellation and parallel interference cancellation. In this thesis, we consider the adaptive DS-CDMA MMSE detector, where lattice-based filter algorithms are employed to suppress MAl. Most of the work in the literature has considered the implementation of this detector using the Least Mean Square (LMS) algorithm. The disadvantage of using the LMS algorithm to implement the MMSE detector is that the LMS algorithm converges very slowly. The main aims of this thesis are as follows. A review of the literature on MUD is presented. A lattice based MUD is then proposed and its performance evaluated using both simulation and analytical methods. The results obtained are compared with those of the LMSMMSE detector. From the results obtained the adaptive Lattice-MMSE detector is shown to offer good performance tradeoff between convergence results and BER results.