Channel Coding and Multiuser Detection for CDMA and Wireless Communications

Doktorsavhandling, 1999

Next generation wireless communication systems will require support for many different applications as well as improved quality compared to existing second generation systems like GSM (Global System for Mobile communications). Since there will be more users and higher data rates, the capacity must also be increased. This thesis studies channel coding and multiuser detection, two techniques that are of vital importance for designing a flexible system with high capacity and high quality. In the first part of the thesis optimum distance spectrum (ODS) convolutional encoders are presented. These encoders generate maximum free distance codes and have superior information error weights. Results show that the ODS codes outperform previously published maximum free distance codes at no extra cost. We next use the ODS encoders as parent encoders to find large and flexible families of multirate codes. These multirate codes are obtained both by puncturing and nesting of the parent ODS codes, and are shown to perform almost as well as unpunctured codes, and to be well suited for rate-matching in direct-sequence code-division multiple-access (DS-CDMA) systems. A study of the performance of a number of multiuser detectors shows that such advanced receiver structures are quite sensitive to errors in phase and timing estimates. With parameter estimation errors the performance of all investigated detectors approach that of the conventional matched filter detector. This emphasizes even more the important role of channel coding in CDMA systems. A family of powerful very low-rate maximum free distance convolutional codes is found by nesting of rate 1/4 ODS codes. These codes are proposed for combined coding and spreading in CDMA systems, and are shown to be both more flexible and more powerful than previously known very low-rate convolutional codes. When applied in a CDMA system, we obtain significantly better performance than with a conventionally coded and spread scheme. Detailed investigations of parallel and successive interference cancellation in combination with channel code decoding show performance very close to the single user bound for the very low-rate code. All results with interference cancellation assume that decisions are taken before regeneration and subtraction of the signal. Accurate analysis of such non-linear interference cancellation is difficult. We therefore present iterative approximative analysis techniques that give results close to simulation results. Sequential decoding of convolutional codes is a suboptimum technique that makes it possible to decode very long constraint length codes and thus obtain very low error rates. We evaluate the performance of sequential decoding theoretically and by simulations for Rayleigh fading channels. A tailbiting coding scheme together with sequential decoding is proposed and shown to give improved throughput for wireless packet data transmission.