Research On Random Beamforming And Adaptive Schedule For MIMO Systems

The development of the wireless communication system has the growing demand of the transmission rates and the guaranteed Quality-of-Service(QoS).The MIMO technique can multiple the transmission rate and improve the quality of the data link,and no additional power and spectrum resource are needed.It can also supply the high-speed data service.MIMO technique has become one of the key technologies in the next generation communication systems.Most of former researches focused on the point-to-point transmission.The research on multiuser MIMO systems,especially the multiuser diversity gain drew more and more interest in recent years.Our research work concentrates on the adaptive transmission strategy and multi-user scheduling mechanism of the broadcast downlink channel of the multiuser MIMO systems.Although the Dirty Paper Coding(DPC) has been proved to be the optimal strategy to achieve the capacity region of the MIMO downlink broadcast channel,it can't be applied in practical systems.The former studies have shown that by scheduling the multiple users with the feedback of part channel status information,orthogonal random beam-forming technology can achieve the same scaling law with the DPC scheme when the user number in the system is large enough.But if the user number is limited,the performance of the algorithm has suffered serious losses.By using multiple beams to communicate with multiple users simultaneously,ORBF scheme achieves spatial multiplexing gain and improves the system capacity.However,during data-transmitting over each beam,it will interfere with other beams,especially in high Signal-to-Noise-Ratio(SNR) condition,and the system is "interference limited".In this paper,we firstly analyze the optimal beam number in different conditions of various user number and various system average SNR by the theoretical analysis and numerical simulation.And then we propose a strategy of beam selection and multi-user scheduling by user-sorting and using the additional feedback of channel status information of part users.In each slot,the proposed strategy dynamically selects some of the optimal beams for data transmission.It can improve the system capacity by reducing the inter-beam interference.And besides that,the scheduling method by user-sorting avoids the traversal search of beams and users,greatly reducing the complexity of the algorithm.This paper also gives a novel beam selection and multiuser scheduling algorithm combined with lookup tables.The simulation results show that the capacity of the proposed algorithm is slightly less than the traversal search algorithms,but the complexity of the proposed algorithm is much lower.The weight matrixes used in ORBF are randomly generated in each time slot, whether the weight matrixes match the users' channel matrixes will limit the system performance of the sum rate,so choosing the best-matched weight matrix becomes a new issue.Koutouris and wan's research has proved that choosing the best-matched weight matrix for each time slot is efficient for improving the systems' sum rate.However,their algorithms require each user to investigate the channel and feedback the CSI many times, which leads to a complex process of implementation and too much feedback.We propose a novel method which needs only one training slot to choose the best-matched one from multiple weight matrixes,the proposed method can reduce the number of training slots and give more time for data transmission and improve the spectrum efficiency.In our method users need evaluate the channel status only once,they can use the rotation of unitary matrix to construct several virtual equivalent channels.Through the measurement and feedback the part CSI of the equivalent channels,the base station choose the best one from the unitary matrixes which construct the equivalent channel,and then the data is multiplied with the selected matrixes and transmitted.Our method can get the multi-matrix diversity gain by using only one training slot.A majority of the researches on multiuser scheduling of the downlink channel of MIMO system focus on improving the sum rate performance of the system,but ignore the fairness over all users.Proportional Fairness Schedule(PFS) gets a tradeoff between the system capacity and fairness when the users in the system experience the homogeneous channel conditions and the fading of the channel varies rapidly.However it ignores the fairness between the users in different conditions.In this paper,we denote a new parameter to describe the dynamic behavior of each user's channel condition.The parameter uses the request rate of the present time slot and the statistical information of the request rate.Two adaptive PFS algorithms with the new parameter are proposed.They can offer more chance to the users who is falling into the bad channel conditions and guarantee their quality of service.