Study On MU-MIMO Technology In Very High Throughput Wireless Local Area Network

In order to improve the throughput performance of wireless LAN, multi-usermultiple-input multiple-output orthogonal frequency division multiplexingtechnology has been adopted by IEEE802.11ac standard. Downlink multi-user MIMOchannel capacity is the key in determining the throughput performance, but thechannel capacity optimization problem is a complex non-convex problem. Faced withthis situation, we proposed a step-by-step optimization scheme, which decomposedthe optimization problem into three sub-problems, namely user grouping problem,spatial steam mapping problem and power allocation problem respectively. A detailedstudy of these three sub-problems was conducted, and a sub-optimal solution schemeto the channel capacity optimization problem was determined ultimately.In the user grouping problem, we discussed four user grouping metrics. Amongthem, the channel capacity criterion is a widely used metric, but it is difficult to beused in practice due to its computational complexity. Therefore, the singular valuedecomposition based null space projection criterion has been proposed, which hasrelatively low complexity, and can approximate the performance of the channelcapacity criterion. Orthogonal procedures based null space projection criterion andweighted sum of user attenuation and user correlation criterion has been used in thesingle-antenna user situation. In this paper, we extended these two metrics, so thatthey can be applied to the case of multi-antenna users. On this basis, we discussed theexhaustive search grouping algorithm and sequential adding grouping algorithmwhich can be applied to all the user grouping metrics, and the quadratic optimizationgrouping algorithm which can be applied to the weighted sum of user attenuation anduser correlation criterion. In addition, we also discussed a randomly selectinggrouping algorithm.In the spatial stream mapping problem, we adopted the assumption that there isno interference between the users. On this basis, we discussed four spatial streammapping approaches. Singular value decomposition based spatial stream mapping isthe optimal mapping approach which has been proposed, but it cannot be used inpractice because it requires that users have the knowledge about the channel stateinformation of other users. Therefore, the first class channel matrix pseudo inversebased spatial stream mapping has been proposed. This spatial stream mappingapproach can be used in practice, but its performance is relatively poor. In this paper,we propose the second class channel matrix pseudo-inverse based spatial streammapping approach, which is equivalent to the first class channel matrix pseudoinverse based spatial stream mapping approach in performance, but has greatlyreduced computational complexity. We also proposed an eigen vector matrix pseudoinverse based spatial stream mapping approach which is suitable for wireless localarea network, and has better performance than the channel matrix pseudo inversebased spatial stream mapping approaches.In the power allocation problem, we adopted the water filling approach, which isthe optimal power allocation approach when the user grouping approach and thespatial stream mapping approach has been determined. Through the simulation experiments, we compared the various user groupingapproaches and spatial stream mapping approaches, and identified a combination ofuser grouping approach and spatial stream mapping approach as the solution schemefor wireless LAN downlink multi-user transmission channel capacity optimizationproblem. In addition, we designed a physical layer multi-user transmission framestructure which is suitable for this solution scheme.