Near ML And Lattice Reduction Research On Massive MIMO Detection

Large MIMO makes a clean break with current practice through the use of a large excess of service-antennas over active terminals and time division duplex operation. We consider multiple-input multiple-output(MIMO) systems, which are known to have high spectral efficiency in rich scattering environments and high link robustness. A major difficulty in the implementation of MIMO systems is the detection(signal separation) problem, which is generally computationally expensive to solve. This problem can be especially pronounced in large MIMO systems. Many different methods have been proposed during the past two decades that aim to achieve, with reduced computational complexity, the performance of the optimal detector.There are two main categories of MIMO detectors. The first consists of detectors whose complexity(run-time) depends on the particular channel realization. This category includes, in particular, methods that perform a tree-search. Unfortunately, these methods have an exponential worst-case complexity unless a suboptimal termination criterion is used. The other category of detectors consists of methods that have a fixed(deterministic) complexity that does not depend on the channel realization. These methods are more desirable from an implementation point of view, as they eliminate the need for data buffers and over-dimensioning(for the worst-case) of the hardware. The main work of this thesis is to put forward two types of different detection methods.One kind of detection method which this thesis proposed is genetic likelihood ascent search algorithm,which rose on the likelihood ascent search algorithm to further improve the performance of large MIMO receiver system via genetic approach. The genetic likelihood ascent search algorithm is an improved method of likelihood ascent search algorithm. The algorithm combine genetic search method with likelihood ascent search algorithm to achieve better performance.Another detection method proposed in this thesis belongs to the second category detection technology, whose complexity is polynomial form.As the second class detection techniques,this thesis proposes a Lattice Reduction(LR) aided listed interference elimination algorithm and LR aided recursive interference elimination algorithm,which are both according to the view that subspace matrix of matrix is more orthogonal generated by LR.LR aided recursive interference elimination algorithm is a part interference cancellation algorithm.it can achieve better performance the LR aided linear detection.while its complexity which is related to the number of recursice is higher than the LR aided linear detector.LR aided listed interference elimination algorithm is designed to Massive MIMO decomposed into multiple low dimensional MIMO and use the LR aided linear detection on each low dimensional MIMO system.this method provides a well defined tradeoff between computational complexity and performance.