Detection and Coding Techniques for Fourth Generation Air-Interfaces based on Multicarrier Modulation

This thesis investigates a number of multiple-access, multiuser detection (MUD) and channel coding methods for the downlink of 4G wireless systems based on multicarrier modulation. Two joint Orthogonal Frequency Division Multiplexing (OFDM) and Code Division Multiple Access (CDMA) schemes have been characterised and compared in respect of their performance in different environments and configurations. The first scheme is Multicarrier Direct Sequence CDMA (MCDS-CDMA), based on time domain spreading. The second is Multicarrier CDMA (MC-CDMA), based on frequency domain spreading. The results demonstrate that, though MC-CDMA benefits from frequency diversity even in the absence of channel coding and antenna diversity, MC-DS-CDMA performs better than MC-CDMA with channel coding and antenna diversity.

MC-CDMA becomes subject to MAI enhancement in frequency selective fading channels. To reduce the effects of the MAI and exploit the available frequency diversity, MC-CDMA requires MUD. However, the optimal Maximum Likelihood (ML) detector is prohibitively complex. A novel near-ML MUD algorithm based on the Chase algorithm has been proposed in this thesis. The proposed algorithm offers significant improvement to the performance of MC-CDMA using much less complexity compared to ML-MUD. A gain of 1dB to 4dB over the non-MUD performance was achieved depending on the spreading factor and the number of error patterns.

Finally, a novel symbol-level Chase-based decoding algorithm has been proposed for non-binary Block Turbo Codes (BTC) with application to OFDM. The conventional bit-level decoding algorithm requires the components of a non-binary code to be represented by binary bits, which is inconsistent with the use of non-binary codes; whereas the proposed symbol-level decoding algorithm does not require such a binary representation. The superiority of the new decoding algorithm has been confirmed through simulation results. Moreover, the bit-level decoding algorithm works only with Grey coded constellations. The new symbol-level algorithm is not limited by this condition. This enables any mapping constellation to be used with BTC, especially those which enhance power efficiency and reduce peak-to-average-power-ratio effects.