Research On Multi-user Power Optimization Theory In Precoding-based Wireless Communication Networks

With the rapid development of Information Technology, all the existing informa-tion systems with different technical standards and service modes will be coordinated and turned to be a heterogeneous one, which will greatly change and affect the way of acquiring information. During this process, wireless communications have been play-ing a more and more important role in conveying the information bits. Then, more smart terminals, larger amount of data bits, and stricter requirement on the Quality of Service (QoS) will lead to higher demand on the transmission speed (hence capacity) of the future wireless communication systems.From the shannon information theory, this objective can be achieved through but not limited to the following three approaches.· More efficient utilization of the spectrum resource, e.g. based on the cognitive radio (CR) technology.· Increasing the number of the parallel independent channels to transmit informa-tion, such as using multi-antenna-related techniques.· Power controlling based on the optimization theory.These have greatly improved the system capacity and satisfied the demands on high speed transmission. However, on the other hand, the development of wireless commu-nication has brought two obstacles to be dealt with, i.e. the rapid increase of inter-ference and the excessive utilization of resource. So, many interference-management techniques are used to suppress or to mitigate the inter-stream, inter-user, and inter-cell interference, e.g., based on the precoding techniques. Afterwards, on the basis of the interference processing, the system performance can be further improved by some effi-cient or even optimal resource allocation schemes, such as power controlling. Through resource allocation, some performance metrics at the physical layer, e.g., capacity, and those QoS-related metrics at the application layer, e.g., Peak Signal to Noise Ratio (PSNR) for video quality evaluation can be improved or optimized.In this thesis, we investigated the problem of optimal power allocation in the precoding-based multi-user Multiple Input Single Output (MISO) downlink systems, and also in precoding-based cognitive radio networks. The objective is to maximize the physical-layer sum-capacity or the application-layer sum-PSNR through power al-location. The problem can be formulated as a multi-variable nonlinear nonconvex optimization problem which can not be solved by the traditional convex optimization techniques. Then, we design the optimal power allocation schemes based on some powerful mathematical theories including the Branch and Bound (B&B) framework, convex relaxation technique, Geometric Programming, monotone analysis, composition property of convexity and their combinations. The main contributions of this thesis arc as follows.1. We investigate the optimal power allocation scheme in the precoding-based multi-user MISO downlink systems. At the Base Station (BS), Signal-to-Leakage-and-Noise Ratio (SLNR) precoding is applied to mitigate the inter-user interference. On the basis of the designed precoding, we propose a globally opti-mal power allocation scheme to maximize the sum-capacity of all the users in the system. The power allocation problem is formulated as a nonlinear nonconvex optimization problem and then we design an optimal power allocation algorithm which can be adopted to all the channel conditions. Meanwhile, we find the op-timization problem could be solved by Geometric Programming when the system works in the high SINR region. Based on this observation, we further propose a judgement-decision algorithm to balance the optimality and computation complex-ity in which power allocation scheme is acquired by Geometric Programming in high SINR case and by the proposed sum-capacity-maximized algorithm in other cases. Through the simulation results, we evaluate the performance improvements in terms of the sum-capacity achieved by the proposed power allocation comparing with three existing power allocation schemes and then analyze the effects of the number of transmit antennas and of the SNR on the sum-capacity performance. This contribution corresponds to Chapter2and paper1in the paper list at the end of the thesis.2. We investigate the optimal power allocation in the CR network based on SLNR precoding and hybrid opportunistic spectrum access. SLNR precoding is applied at the secondary BS to combat the inter-user interference in the secondary system and the interference to the primary user. On the basis of the designed precoding vectors, we propose an optimal power allocation scheme to maximize the sum-capacity of the secondary user system under the interference temperature constraint on the subcarrier where primary user exists. The power al-location problem is formulated as a multi-variable problem with power constraints and interference temperature constraints, which can be proven neither convex nor linear. Then, based on a combination of the Branch and Bound framework and the convex relaxation technique, we design an optimal power allocation algorithm that can approach the global optimum with any predefined optimality precision. We utilize the convex relaxation technique to transform the problem to a relaxed, solvable one and then, under the branch and bound framework, partition the vari-able space to make the solution obtained by solving the relaxed problem approach the primitive global optimum at any predefined precision. Through simulations, we evaluate the capacity improvement brought by the proposed power allocation scheme. This contribution corresponds to Chapter3and paper2in the paper list at the end of the thesis.3. We investigate the optimal power allocation scheme that can maximize the sum-PSNR in the SLNR-precoding-based multi-user multimedia transmission system. We take the PSNR as the measurement of the multi-media quality and define it as the utility function. The SLNR precoding is used to mitigate the inter-user interference at the transmitter. On the basis of the precoding vectors designed, we optimize the sum-PSNR through power allocation. The formulated power allocation problem is nonlinear and nonconvcx. To deal with this, we design an optimal power allocation scheme adapted to high SINR case. Specifically, when the system works in the high SINR region, we transform the formulated problem into a convex one through monotonic analysis, Geometric Programming together with the composition property of convexity. Then, from the simulation results, we evaluate the effects of the proposed power allocation scheme on sum-PSNR and further discuss the effects of precoding and power allocation on improving the sum-PSNR performance of the system respectively. This contri-bution corresponds to Chapter4and paper3in the paper list at the end of the thesis.