【作者】 黄秋萍；

【导师】 杨鸿文；

【作者基本信息】 北京邮电大学 ， 通信与信息系统， 2013， 博士

【摘要】 无线网络无论怎样发展,都不可能超越理论上的极限——网络在信息论意义下的容量。研究无线网络的容量问题可以使我们了解技术发展的潜力,发现容量的数学特性能启发我们找出趋近容量的必要条件和充分条件,为实际技术的设计指出探索的方向。虽然香农公式早已给出了点到点通信的容量公式,但扩展到多用户系统时还需要考虑许多新的问题,包括多用户调度及功率控制、用户间干扰、信道边信息反馈设计等等。同时,关于容量的度量也从单一的标量扩展出了容量域、和容量这样一些新的概念。本文主要研究多用户系统的和容量及功率控制问题,包括蜂窝系统的下行和上行两种场景。主要工作如下：(一)对于单小区下行多用户场景,研究了量化信道信息条件下的系统和容量问题。论文首先对单小区多用户系统的下行进行建模,然后结合Max C/I调度策略研究了量化信道信息条件下的系统和容量,针对有无功率控制、是否给定量化边界等情形对最优解进行了理论推导,给出了最优的功率、速率及量化边界的联合设计方案。(二)针对单小区上行多用户场景,研究了能使系统和速率最大的功率控制问题。首先研究了可以达到系统容量的最优的功率控制策略,证明了在用户的峰值功率限制下可以达到系统容量的最优的功率控制为二值功控,即所有被调度的用户都以峰值功率发送；根据这性质,论文进一步推导出了最优的功率控制向量的形式及最优的功率控制为时分控制的充分条件；在此基础上,论文给出了最优的功率控制的求解算法,并提出了一个低复杂度的次优算法。(三)针对单小区上行多用户场景,分别研究了服务质量约束条件和用户差异化约束条件下的系统和容量及最优的功率分配问题,推导了功率分配最优解的数学性质,然后根据这些数学性质分别给出了服务质量约束条件以及用户差异化情景下的最优功率分配。(四)研究了多小区多用户场景下的系统容量及功率控制问题。首先讨论了使多小区多用户系统上行和速率最大的最优的功率控制,基于最优解的数学性质,给出了次优的功率控制及和容量的求解算法,然后对多小区多用户系统的下行容量进行了建模和分析。论文在上行的研究模型包含了采用了连续干扰删除技术的场景,针对在多小区上行系统中二值功率控制可以用来近似最优的功率控制的场景,论文提出了一些次优功率控制算法以及考虑了用户公平性的算法。更多还原

【Abstract】 No mater how fast the evolution speed of the wireless network is, the rate of the wireless network can not transcend the theoretical limits-system capacity of the network under the perspect of information. Study the network theorical capacity can help us to know the developing goal of possible techniques and to explore the potential of the current networks. Besides, study the mathematic characters of system capacity can inspire us to find the neccesary condition and sufficient condition to reach the capacity, which indicates the directors for the design of the practical techniques.The information capacity for point to point communications was originally arised by Shannon. When extend the Shannon formula to multiuser systems, many problems have to be considered, e.g. multiuser schedule, power control, interferences among users and channel side information feedback designs. At the same time, the measurement of capacity is changed from a single scalar to many new concepts such as capacity region and sum capacity. This paper is focus on sum capacity and power control of multiuser systems. Both downlink and uplink are considered:(1) For the downlink of single cell multiuser systems, sum capacity with quantized channel side information is studied. The system capacity with quantized channel side information under Max C/I scheduler is studied. The optimal solutions for the scenarios with/without power control, with/without fixed quantization boundaries are derived, and the optimal joint design of power, rate and quantization boundaries are provided.(2) For the uplink of single cell multiuser systems, the optimal power control for sum rate maximization is studied. It is proved that under the users’ peak power constrains, binary power control is optimal, i.e. all scheduled users should transmit with peak power. Based on this property, the structure of the optimal power control is derived, the sufficient condition for time divison scheduler to be optimal scheduler is provided. With these conclusions, the optimal power control algorithm and a low complexity suboptimal algorithm are proposed.(3) For the uplink of single cell multiuser systems, sum capacity and its optimal power allocation for users with Quality of Service(QoS) requirement and users with experience differentiations are studied respecitively. Mathemetic properties of the optimal power allocation are derived. Based on these properties, the power control for users with QoS constraints and users with different experience are provided respectively.(4) For multicell multiuser systems, sun capacity and its optimal power control are studied. The optimal power control for uplink systems are analysised fisrtly, based on the properties of the optimal power control, some suboptimal algorithms are provided. Then we discuss the sum capacity of the downlink systems. The system model of uplink includes the cases of base stations with/without successive interference cancelation(SIC) technique. For the scenarios that binary power control can achieve similar performance as the optimal power control, many suboptimal algorithms as well as a fairness algorithm are proposed.更多还原