Research Of Detection Methods In Multi-Antenna System

MIMO technology which adopts multi-antennas can improve the data rate in communication system effectively, so it is known as one of the key technologies in the next generation wireless communication system. By now, MIMO has been the focus of both academic and industrial circles.The computation complexity of non-linear algorithm is increasing non-linearly with the increase of antenna number, such as ML algorithm whose complexity would be increasing in an exponent level with the antenna number in the transmitter increases. Although ML’s performance is the best in SER, the extreme complexity makes ML unfit for the industry implementation.Sphere detection method, a quick searching algorithm, is designed for decreasing the complexity of ML through reducing the searching area. And it has a polynomial computation complexity on average compared with ML detection. Sphere detection converts the MIMO detection to the following process:obtain the Euclidean distance between y and Hx, and then compare this distance with the radius of sphere detection. Sphere detection only searches the symbol in the radius through confining the searching area.The contents of this paper are detection methods of receiver in wireless communication system. Based on the learning from the achievement in this area, I probed into some hard problems in MIMO detection and then proposed some detection algorithms. The problem of MIMO sphere detection is that the complexity of MIMO sphere detection is extremely high when the channel condition is bad, which brings so many problems in the industry implementation.This paper focuses on the contradiction between SER performance and computation complexity of the algorithms and it proposes some technology solutions. The proposed methods would achieve a nearly ML performance in a wide SNR range, and they have the lower complexity as well. The proposed methods may have the huge significance to the industry implementation. There are two research directions in this paper: combination of sphere detection and linear detection, and Tree-pruning sphere detection.As to the combination of sphere detection and linear detection, we proposed3detect methods which can be classified into two parts. The first part proposed an improved method which combines K-Best sphere detection with OSIC, and then improved this method. K-Best mechanism can fix the computation complexity which makes the implementation much easier. The other part proposed a combination method of promising path sphere detection and MMSE. Increasing Radius Search (IRS) is one of the branches of sphere detection, and it proposed the Improved Increasing Radius Search (IIRS) which used the promising path mechanism to improve IRS in this branch.As to the Tree-Pruning sphere detection, this paper gives an improved tree-pruning sphere detection which has a stable complexity expectation. The new detection algorithm would set the tree-pruning radius which is used in detection process. This algorithm has a stable complexity expectation which stands a stable electrical consuming.