Evolutionary-Algorithm-Assisted Joint Channel Estimation and Turbo Multiuser Detection/Decoding for OFDM/SDMA

• Published in: IEEE transactions on vehicular technology 2014-03, Vol.63 (3), p.1204-1222
• Main Authors: Jiankang Zhang, Sheng Chen, Xiaomin Mu, Hanzo, Lajos
• Format: Article
• Language: English
• Subjects: Applied sciences; Channel estimation; Channels; Coding, codes; Decoders; Decoding; Detection, estimation, filtering, equalization, prediction; Differential evolution algorithm (DEA); Evolutionary algorithms; evolutionary algorithms (EAs); Exact sciences and technology; genetic algorithm (GA); Information, signal and communications theory; Iterative decoding; joint channel estimation (CE) and turbo multiuser detection (MUD)/decoding; Joints; Mud; Multiplexing; Multiuser detection; OFDM; Optimization; Orthogonal Frequency Division Multiplexing; orthogonal frequency-division multiplexing (OFDM); particle swarm optimization (PSO); repeated weighted boosting search (RWBS); Searching; Signal and communications theory; Signal, noise; space-division multiple access (SDMA); Swarm intelligence; Systems, networks and services of telecommunications; Telecommunications; Telecommunications and information theory; Transmission and modulation (techniques and equipments)
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2013.2283069

The development of evolutionary algorithms (EAs), such as genetic algorithms (GAs), repeated weighted boosting search (RWBS), particle swarm optimization (PSO), and differential evolution algorithms (DEAs), have stimulated wide interests in the communication research community. However, the quantitative performance-versus-complexity comparison of GA, RWBS, PSO, and DEA techniques applied to the joint channel estimation (CE) and turbo multiuser detection (MUD)/decoding in the context of orthogonal frequency-division multiplexing/space-division multiple-access systems is a challenging problem, which has to consider both the CE problem formulated over a continuous search space and the MUD optimization problem defined over a discrete search space. We investigate the capability of the GA, RWBS, PSO, and DEA to achieve optimal solutions at an affordable complexity in this challenging application. Our study demonstrates that the EA-assisted joint CE and turbo MUD/decoder is capable of approaching both the Cramér-Rao lower bound of the optimal CE and the bit error ratio (BER) performance of the idealized optimal maximum-likelihood (ML) turbo MUD/decoder associated with perfect channel state information, respectively, despite imposing only a fraction of the idealized turbo ML-MUD/decoder's complexity.