Practical design and performance evaluation of dynamic channel estimation based log-MAP turbo decoder for DS-CDMA systems in multipath fading environment

Excellent forward error correction capability of turbo codes in conjunction with DS-CDMA multiple access format can lead to highly power and bandwidth efficient mobile communication systems. Practical implementation of turbo-coded DS-CDMA systems is impeded by the extremely high computational complexity of optimal turbo decoding algorithms and their need for accurate channel information in the complex time-varying frequency-selective multipath fading channels encountered in such systems.; To obtain optimal performance turbo codec a thorough examination of convolutional turbo codes, their encoding and decoding structure, performance decisive factors, existing advanced performance enhancement techniques and comparison with alternate turbo code configurations is performed. Additionally, several innovative performance optimization ideas have been proposed. Log-MAP is established as the most efficient SISO decoding algorithm for turbo decoding after detailed comparisons with other popular turbo decoding options such as MAP, max-log-MAP and SOVA.; A novel block-wise noise variance estimation technique utilizing the correlation characteristics of the link-specific spreading codes employed in DS-CDMA systems, and an adaptive grouping based dynamic fading amplitude tracking technique, are proposed for highly accurate channel estimation in correlated multipath fading channels. The proposed channel estimation based turbo decoder performs within 0.15 dB to the known channel information based decoder in a multi-user frequency-selective Rayleigh fading channel with maximal-ratio-combining.; For DS-CDMA based mobile communication applications with MRC based Rake receiving, a low-complexity high-efficiency all-digital design for the proposed adaptive channel estimation based optimal log-MAP turbo decoder is presented. The hardware-complexity and decoding-performance determinant issues of quantization and data flow for the iterative log-MAP turbo decoder as well as the pre-decoder signal processing modules are prudently addressed to ensure highly efficient hardware implementation. The efficient design strategies applied confine the decoding complexity while leading to an excellent performance within 0.2 dB of the software decoder.