Intelligent genetic algorithms for next-generation broadband multi-carrier CDMA wireless networks

Abstract

This dissertation proposes a novel intelligent system architecture for next-generation broadband multi-carrier CDMA wireless networks. In our system, two novel and similar intelligent genetic algorithms, namely Minimum Distance guided GAs (MDGAs) are invented for both peak-to-average power ratio (PAPR) reduction at the transmitter side and multi-user detection (MUD) at the receiver side. Meanwhile, we derive a theoretical BER performance analysis for the proposed MC-CDMA system in A WGN channel. Our analytical results show that the theoretical BER performance of synchronized MC-CDMA system is the same as that of the synchronized DS-CDMA system which is also used as a theoretical guidance of our novel MUD receiver design. In contrast to traditional GAs, our MDGAs start with a balanced ratio of exploration and exploitation which is maintained throughout the process. In our algorithms, a new replacement strategy is designed which increases significantly the convergence rate and reduces dramatically computational complexity as compared to the conventional GAs. The simulation results demonstrate that, if compared to those schemes using exhaustive search and traditional GAs, (1) our MDGA-based P APR reduction scheme achieves 99.52% and 50+% reductions in computational complexity, respectively; (2) our MDGA-based MUD scheme achieves 99.54% and 50+% reductions in computational complexity, respectively. The use of one core MDGA solution for both issues can ease the hardware design and dramatically reduce the implementation cost in practice.