Research On Signal Detection Techniques For MIMO Communication Systems

Multiple-Input Multiple-Output (MIMO) is an important technology of smart antenna in the field of wireless communications. It breaks through the channel capacity bottleneck of traditional Single-Input Single-Output (SISO) system, and can increase the capacity of wireless channel greatly without increasing the bandwidth via using the technology of multiple antennas at the transmitting terminal and the receiving end. Because of the lacking frequency resource and the narrow channel bandwidth in today, MIMO becomes the hot research topic for scholars. What’s more, signal detection is the key of MIMO systems, the performance and complexity of the receiver in MIMO systems has a direct impact on the quality of communications and the recovery of the signals.This thesis introduces the research of the detection techniques in MIMO wireless systems. At first, the channel model and the channel capacity of MIMO systems are introduced. Secondly, presents the principles of OFDM and MIMO-OFDM. Then introduces some kinds of traditional detection algorithms of MIMO systems, including the zero forcing (ZF) algorithm、minimum mean square error (MMSE) algorithm、maximum likelihood (ML) algorithm、QR decomposition algorithm、successive interference cancellation (SIC) algorithm, and the ZF-SIC algorithm which composed by ZF and SIC, the MMSE-SIC algorithm which composed by MMSE and SIC. Simulate these traditional detection algorithms by computer, and then compare the performance, analysis the advantage and disadvantage of every algorithm. The emphasis is the sphere decoding (SD) algorithm. SD algorithm is the improved and simplified algorithm of ML, and it has the same performance as the ML algorithm in detection with a lower computation complexity.This paper proposes four kinds of improved algorithm base on researching the traditional detection algorithms.(1) A novel algorithm which consists of ML algorithm and SIC algorithm is proposed in this paper. The proposed algorithm utilize the ML algorithm to detect the first level of SIC algorithm, and utilize the SIC algorithm to detect the other levels. Because of the good BER performance of ML, the novel algorithm can reduce the error propagation efficiently and can get a better detection performance than SIC, and the complexity is lower than ML algorithm. The improved algorithm takes a trade-off between BER performance and computation complexity.(2) Propose an improved sphere decoding detection algorithm which combined with MMSE. This novel algorithm modifies the parameter k, which can constrict the sphere radius, to be a constant, meanwhile combine with MMSE rule. The new algorithm can reduce the computation complexity efficiently without change the BER performance obviously.(3) A pruning sphere decoding algorithm based on the MMSE is proposed in chapter4. Pruning sphere decoding algorithm can estimate the enlargement quantities by the statistical expectation, and cancel the enlargement quantities at the same time. This method will decrease a lot of complexity to the traditional algorithm. Meanwhile, introduce the MMSE rule to the pruning SD algorithm, can also decreases a part of complexity.(4) It also proposes an ordered sphere decoding detection algorithm. To apply the sorting method which comes from sorted QR decomposition algorithm to the sphere decoding algorithm, sort the signals with the order of SNR from high to low. The novel algorithm not only keep the same performance as the traditional sphere decoding algorithm, but also can decrease the calculation complexity of the sphere decoding algorithm efficiently.In this thesis, a lot of simulation experiments are carried out, and the results of the simulation are been analyzed and compared. Experimental results show that the improved methods are effective.