Study On Cooperative Beamforming And Multiuser MIMO Technologies

Multi-antenna or multiple-input multiple-output (MIMO) technology has been widely adopted in modern wireless communication systems. Beamforming plays a key role in achieving the array gain in MIMO communication systems. As the emerging of cooperative communications and multi-user MIMO systems, cooperative beamforming and multi-user multi-group multicast beamforming attract great attentions in the recent research communities of communication and signal processing. However, a substantial increase in the number of users and also in the delivery of user data brings critical challenges to the conventional multiuser MIMO systems. Massive MIMO is one of the most effective ways to overcome these challenges. The spectral efficiency and also the energy efficiency of the wireless communication systems are significantly improved via implementing more antennas at the base stations. As a result, massive MIMO becomes a hot research topic in the literature recently.The paper mainly focuses on the investigation of the above-mentioned three important physical layer techniques. We study the design of cooperative relay beamforming in both dual-hop and multi-hop layered relay networks, the optimization of multi-user downlink beamforming in terms of user fairness and transmission robustness, and the design of uplink training scheme and downlink precoding scheme in massive MIMO systems. The main contributions can be summarized as follow:In the first part, we investigate amplify-and-forward based cooperative beamforming in dual-hop relay networks. Firstly, we consider a single source scenario. The optimal cooperative beamforming scheme is derived in terms of maximizing the received signal-to-noise ratio (SNR) at the sink when the multiple relays introduce correlated Gaussian noises and are subject to individual power constraints. It is proved that the noise correlation always improves the average transmission rate of this network when the relays adopt the obtained optimal cooperative beamforming scheme. Then, we extend our study to a dual-hop relay network with multiple sources. We propose a method to characterize the achievable rate region of this network where cooperative beamforming is employed by relays.In the second part, we study the cooperative beamforming in a multi-hop layered relay network. Under the individual relay power constraint, the joint multi-layer beamforming optimization problem in terms of maximizing the received SNR at the sink is formulated as a non-convex problem. Generally speaking, this problem is hard to solve. We propose two efficient algorithms to solve the beamforming design problem in a layer-by-layer manner. In the first approach, we adopt an alternating optimization technique to iteratively solve the joint beamforming problem. In each step, we solve a localized single layer beamforming optimization problem that can be equivalently converted into a convex second-order cone program (SOCP). Thus this restricted problem can be efficiently solved in polynomial time. We prove that this approach results in an asymptotic optimal solution of the original problem when the network is in the generalized high-SNR regime. In the second approach, we propose a per-layer sum-rate maximization problem to compute the beamforming vectors employed by the relays in each layer. We adopt a successive convex approximation (SCA) algorithm to approximately solve this problem, which converges to a local maximum point of the per-layer sum-rate maximization problem. Simulations confirm that our proposed beamforming schemes have superior performance than the conventional ones.In the third part, the focus is put on the design of downlink multi-group multicast beamforming for a multi-user MIMO system. When perfect channel state information (CSI) is available at the BSs, we propose an approach to maximize the minimum SINR among all users in different multicast groups. This is also referred to as user fairness design criterion. Two algorithms are proposed to approximately solve this NP-hard problem, namely the Dinkelbach-type algorithm and the successive convex approximation algorithm. Both the algorithms yield a lower computation complexity than the conventional bisection algorithm. Then a more practical scenario is considered when the channel estimation error exists. We formulate an SINR-constrained problem and devise a robust rank-two multicast beamforming scheme. We also investigate the sufficient condition under which the semi-definite relaxation method can provide rank-two solutions. The numerical results show that the feasibility ratio of the system under predefined SINR targets can be significantly improved by applying the proposed scheme.In the fourth part, we consider pilot contamination in massive multi-user MIMO systems, which is one of the most critical problems in such systems. The time division duplex operation can be employed to acquire downlink channel state information via uplink training using the channel reciprocity. However, the short channel coherence time due to user mobility significantly limits the number of orthogonal pilot sequences that can be applied. As a result, the same pilot sequence may be reused by different users, leading to pilot contamination in the uplink channel estimation. To tackle this problem, we propose a novel cell-specific uplink training scheme along with a downlink pilot contamination elimination precoding scheme. We show that the system will have an unbounded beamforming gain as the number of base station antennas approaches infinity. To further demonstrate the effectiveness of our proposed scheme, we also derive closed-form expressions of downlink SINR when the number of BS antennas is finite. We use simulations to show the performance improvement of our proposed uplink training scheme and PECP scheme.