Signal Detection Algorithms In MIMO-OFDM Communication System

MIMO-OFDM wireless communication system, the core technology of Beyongd3G or4G, which combines the MIMO and OFDM technologies, can raise the system capacity andspectral efficiency, and also can effectively resist the interference and noise. The quality of acommunication system depends largely on the performance of the detection module.Therefore, the signal detection algorithm is the priority of the MIMO-OFDM communicationsystem's research, and also the core of the research in the paper.First of all, the paper introduced theoretical knowledge of MIMO and OFDMtechnologies, including the characteristics of wireless channel, the channel capacity of MIMOsystem and the model of V-BLAST MIMO system. It also analyzed the fundamentalprinciples, modulation and demodulation techniques and the performance of the OFDMtechnology. It provided the model of the MIMO-OFDM system. And then, it provided thesignal detection algorithms of the MIMO-OFDM system, including the ZF, MMSE, SIC,OSIC, QR, ML algorithms and sphere decoding (SD) algorithm. Finally, it analyzed thecomputational complexity and detection performance of these algorithms.The paper focused on a new algorithm, ZF-MLD detection algorithm, which combinedZF and ML algorithms. The detection performance of the ZF-MLD was better than ZFalgorithm, with almost the same computation. But the ZF-MLD algorithm had to dopseudo-inverse operation twice on4*4multi-antenna space-time multiplexing system, whichhad a high computation. To improve the ZF-MLD algorithm, the paper proposed a newalgorithm, which based on the idea of interference cancellation. It was named ZF-IC-MLD.Experiment and analysis showed that the ZF-IC-MLD algorithm had less computationalcomplexity than ZF-MLD algorithm, without reducing any detection performance.The detection performance of ZF-MLD algorithm was a little low. To avoid this, thepaper proposed another new algorithm, which based on ZF-MLD algorithm and the idea ofthe local search. It was called ZF-LR-MLD. Experiment showed that the ZF-LR-MLDalgorithm had better detection performance than ZF-MLD algorithm, with more computation.