Adaptive implementation of turbo multi-user detection architecture

Abstract

MULTI-access techniques have been adopted widely for communications in underwater acoustic channels, which present many challenges to the development of reliable and practical systems. In such an environment, the unpredictable and complex ocean conditions cause the acoustic waves to be affected by many factors such as limited bandwidth, large propagation losses, time variations and long latency, which limit the usefulness of such techniques. Additionally, multiple access interference (MAI) signals and poor estimation of the unknown channel parameters in the presence of limited training sequences are two of the major problems that degrade the performance of such technologies. In this thesis, two different single-element multi-access schemes, interleave division multiple access (IDMA) and code division multiple access (CDMA), employing decision feedback equalization (DFE) and soft Rake-based architectures, are proposed for multi-user underwater communication applications. By using either multiplexing pilots or continuous pilots, these adaptive turbo architectures with carrier phase tracking are jointly optimized based on the minimum mean square error (MMSE) criterion and adapted iteratively by exchanging soft information in terms of Log-Likelihood Ratio (LLR) estimates with the single-user’s channel decoders. The soft-Rake receivers utilize developed channel estimation and the detection is implemented using parallel interference cancellation (PIC) to remove MAI effects between users. These architectures are investigated and applied to simulated data and data obtained from realistic underwater communication trials using off-line processing of signals acquired during sea-trials in the North Sea. The results of different scenarios demonstrate the penalty in performance as the fading induces irreducible error rates that increase with channel delay spread and emphasize the benefits of using coherent direct adaptive receivers in such reverberant channels. The convergence behaviour of the detectors is evaluated using EXIT chart analyses and issues such as the adaptation parameters and their effects on the performance are also investigated. However, in some cases the receivers with partial knowledge of the interleavers’ patterns or codes can still achieve performance comparable to those with full knowledge. Furthermore, the thesis describes implementation issues of these algorithms using digital signal processors (DSPs), such as computational complexity and provides valuable guidelines for the design of real time underwater communication systems.