【作者】 陈军；

【导师】 冯穗力；

【作者基本信息】 华南理工大学 ， 信息与通信工程， 2014， 博士

【摘要】 分布式天线系统(Distributed Antenna System, DAS)作为一种新型的宽带无线网络结构，通常由拉远单元(Remote Antenna Unit, RAU)、中央控制单元(Central Antenna Unit)和用户组成。在分布式天线系统中，各天线单元之间地理位置相距遥远，彼此间通常通过光纤等有线方式相连。拉远单元仅负责信号的传输和接收，主要的信号处理在中央控制单元中完成。由于分布式天线系统能大大的缩短与用户之间传输距离，因此相对于传统的共址MIMO(Co-located multi-input multi-output, C-MIMO)有更好的覆盖效果。虽然早期的分布式天线系统的提出主要为解决室内的盲区覆盖问题，大量的研究证实，分布式天线系统相对于C-MIMO拥有更高的系统容量、更优的频谱效率、更大的分集增益和更加的节能。如今，基于分布式天线系统的多点协作传输技术已经被3GPP采纳成为LTE-Advance中一项关键技术。不仅如此，分布式天线系统与微小区和Femtocell相比，在节能和频谱效率上也拥有其自身的优势。目前，无处不在的无线宽带网络的普及与应用带给我们日常生活便利的同时，也带来了日益增长的高能耗和高的二氧化碳排放量。随着人们对于节能观念的增强，人们越来越重视绿色节能的无线通信。因此，无线通信的节能技术成为学术界和工业界一个十分活跃的课题。分布式天线系统的能耗相对于其他系统有着其自身天然的优势，本论文主要研究分布式天线系统中的节能技术。研究在满足用户的QoS（Quality of service）要求的前提下，最小化分布式天线系统中发射节点的传输功率，即功率的优化控制问题。本论文从以下几个方面对分布式天线系统进行研究，获得的主要创新成果包括以下几个方面：1.静态功率控制问题：在信道状态信息(Channel State Information, CSI)准静态（即变化不是很快）场景中，考虑获取信道状态信息存在误差的情形下（如估计误差或传输延时），对于研究的联合天线选择与功率控制问题，基于凸优化理论，提出了以下两种带鲁棒性的功率优化算法：第一种算法采用S-Lemma并结合半正定松弛将研究的问题转化为半正定规化问题;第二种算法借助Bernstein概率不等式和半正定松弛原理，将联合的天线选择和功率控制问题转化二次锥规划与半正定规化的混合问题。对于第二种鲁棒性的算法，我们在理论上证明，转化后问题的最优解是秩为1的矩阵，从而证明半正定松弛后所得解是紧的。仿真验证了所提算法具体较强的鲁棒性，同时证明了采用所有天线对用户进行发送信号的天线选择方式不仅具有最好的鲁棒性而且最节能。2.动态功率控制问题：对于信道状态信息快变的场景，在信道是连续变化的前提下，提出一种动态最优的功率控制方法，首先对信道采用随机微分方程进行建模。在此基础上，借助于半正定规划中的原-对偶内点法将动态功率控制问题转化为等价的随机微分方程的稳定性问题。仿真验证所提算法依概率稳定。结合实际应用，我们进一步将所提的算法扩展到信道延时的场景，结合信道预测算法（如：Berg和修正协方差的方法），所提出的算法相对于利用过时信道状态信息的算法的功率节能最大可以达5dBm，即相对于采用过时信道状态信息的算法，所提出的算法功率节能最多可达30%。3.多小区的动态分布天线系统中的性能分析：考虑接收端的信道状态存在延时和多小区间的同频率干扰以及信道的大尺度衰落（主要是路损），我们对分布式天线系统的两种场景（Noise-limted场景和存在同频率干扰的场景）的中断概率进行了理论上的分析，得到中断概率的闭式表达式。Monte Carlo仿真验证了我们所得到的闭式表达式的正确性。更多还原

【Abstract】 Distributed antenna system (DAS), is a new wireless network, consists of spatiallyseparated antenna nodes including contral antenna unit (CAU) and remote antenna unit(RAU). In DAS, the RAUs are distributed geographically apart and connected via opticalfiber, to the CAU which is responsible for the primary signal processing. Widely separatelyantenn units contributed to the shorter access distances between the transmitter and receiver.Moreover, DAS has the merits of better coverage in contrast with the co-locatedmultiplut-input and multiplut-output (C-MIMO). Although the DAS was originally proposedto cover the indoor dead point, recently studies identified that DAS had many advantagessuch as higher capacity and better spectral efficiency and higher diversity. Now, thecoordinated multipoint (CoMP) based on DAS has been adopted by3GPP in theLTE-Advance. Moreover, DAS shows better power and spectral efficient comparing with thestate of art solutions such as mirco cell and femtocell.Nowadays, the seamless and ubiquitous wireless communiation has enhanced the quality ofour everyday life. However, this blanket wireless service comes at the cost of ever-increasinghigh power consumption and CO2emission. The growing conerng over the powerconsumption has triggered the research on green wireless communication. Therefore thepower-efficient techniques in wirelss communications are receiving considerable researchattention. Motivated by these observations, this paper focus on the power efficient techniquesin DAS, specifically, we aim to the minmize the transmit power of all the antenna nodes inmulticell DAS, while fulfill the quality of service (QoS) of all the mobile users. It is alsoreferred to as power control problems in what follows. Compared with the existing studies onDAS, the main contributions of this paper are following:Quasi-static power control for multicell DAS:We propose two robust-beamforming schemes for downlink of multi-cell DAS. One isbased on S-procedure; the other is based on Bernstein ineqaulity, which is suitable for signalcell C-MIMO.We first formulate our problem as a SDP and then prove the solution of this SDP isalways rank-one which guarantees the relaxation is tight, i.e. the optimal solution of relaxproblem is equivalent to the original problem.Considering channel state information at transmitter (CSIT) error, the most robust andenergy efficient AS scheme subject to QoS guarantee for all the users in multiccell DAS, ismade through extensive simulations. Dynamic power control for multicell DAS:Assuming the channel changes continuously, CSI is firstly modeled by employingstochastic differential equation (SDE). The considered dynamic beamforming problem is thentransformed into stabilization for SDE using primal-dual interior point (PDIP) method forsemidefinite program (SDP). By exploiting the theory of stochastic stability, the proposedmethod can be shown to be asymmetrically stable in the sense of probability.Using the channel prediction methods (such as Berg, modified convriance), the proposedmethods can be extened to the delayed CSI cenarios.Dynamic system performance of multicell DAS:Considering the delay of CSIT in the presence of pathloss, we analysis the systemperformance of DAS. In particular, we derive the closed-form outage probability undernoise-limied and unequal power CCI cenarios. Monte Carlo simulations verify our analysis.更多还原