Research On Precoding Schemes In MU-MIMO Systems

In downlink of MU-MIMO systems, it is revealed that the sum rate scales with the active users, which guides the MIMO system design in theoretical. Meanwhile, an inevitable problem, which is how to cancel the multi-user interference, is introduced. Generally, precoding is thought as the best answer. However, this approach lies on the CSIT much, and the perfect CSIT is hard to acquire in practical. In this dissertation, the precoding design based on the imperfect channel information in MU-MIMO downlink channel is our research direction. In details, two points are considered. The first is the robust precoding scheme design based on dynamic CSIT model and the other is the performance analysis of limited feedback in MISO system.Since the CSIT plays a key role in the precoding scheme, the CSIT acquisition model is studied firstly. Suppose that CSIT is acquired by the feedback link, where the feedback delay is fixed and zero BER, and the mobile terminal moves with a certain speed. The channel statistics can be obtained by the training sequence. Due to the channel temporal and spatial correlations, the current CSIT can be estimated utilizing the accurate outdated feedback channel information and channel statistics jointly by the MMSE estimation theory. The channel information obtained by dynamic CSIT is imperfect channel information which introduces the channel uncertainty.The traditional precoding is designed based on the perfect CSIT, however it is hard to implement. Based on the dynamic CSIT model, varies of robust precoding schemes are studied in this dissertation, including robust linear ZF precoding, robust linear MMSE precoding and robust THP precoding. Through the Lagrangian multiplier method, the closed-form expressions of robust schemes are derived. The simulation results show that the performance of robust schemes is affected very much by the feedback delay and mobile speed. The robust system degrades when the feedback delay or mobile speed increases, which can be understood easily.By the moment generating function (MGF) approach, the symbol error rate (SER) of multi-input single-output (MISO) system with limited feedback is studied. The performance analysis of limited feedback has a guiding significance to the codebook design. However, it is difficult to compute the closed-form expression of SER. As a result, it is meaningful to investigate its lower bound. Exploiting the relation between MGF of the received signal-to-noise ratio (SNR) and SER, the lower bound of SER is obtained by utilizing the approximation of complex cap surface in this dissertation. The derived lower bound indicates the system performance is affected by the codebook size, transmit antennas and modulation. Meanwhile, the codebook size can be determined through the bound when the SER requirement is given in practical environment. The SER bound is demonstrated by numerical simulations in which the error is less than 0.1dB.