【作者】 李鹏；

【导师】 蒋雁翔； 康绍莉；

【作者基本信息】 东南大学 ， 电子与通信工程（专业学位）， 2017， 硕士

【摘要】 移动互联网和物联网的高速发展,对未来移动通信系统的频谱效率、能量效率和设备连接数量等指标提出了更高的要求。大规模天线技术通过增加天线数目,大幅提高了系统谱效,成为未来移动通信系统的关键技术之一。然而,大规模天线技术中天线数目的增加,不仅造成系统能耗增加,使系统能效提升受限,还引起导频干扰严重,使多小区资源优化更加复杂。新型多址接入技术通过资源复用和非线性检测,大幅提高了系统谱效和设备连接数量,成为未来移动通信系统的关键技术之一。然而,新型多址接入技术中资源复用可能引入更多资源域,使资源优化面临更多优化对象和复杂干扰。针对这些问题,本论文就大规模天线与新型多址接入技术的资源优化方法展开研究。首先,本论文研究了大规模天线技术中的资源优化方法。针对多小区场景上行链路的高能效功率控制问题,利用约束马尔可夫决策过程表征能效模型,将高能效功率控制优化问题构造为基于约束马尔可夫决策过程的优化问题,并提出一种实现全局最优功率控制策略的优化算法。在所提优化算法中,首先通过拉格朗日乘子法将约束优化问题变换为无约束优化问题,然后利用改进值迭代算法得到全局最优功率控制策略。在改进值迭代算法中,传统值迭代算法负责内层循环迭代,强化学习算法负责外层循环迭代。对于所提优化算法,一方面,可以利用得到的最优决策规则建立离线查询表,从而避免频繁计算;另一方面,可以通过调整算法中的相关参数,实现系统性能与计算复杂度之间的均衡,提供更好的灵活性。仿真结果表明,所提优化算法可以实现理论最优性能。其次,本论文研究了新型多址接入技术的发射机接收机联合设计方法,提出了一种联合大规模天线技术的图样分割多址接入技术(Pattern Division Multiple Access with Large-scale Antenna Array,PDMA-LSA),并给出了 PDMA-LSA的发射机接收机联合设计方案。大规模天线技术与图样分割多址接入技术的结合点在于空域中的波束资源。PDMA-LSA中图样的实现是基于功率域和波束域的联合复用。发射机利用包含功率控制和波束分配的图样映射,实现多用户信号叠加。接收机利用包含空域滤波和顺序干扰消除的混合检测,实现多域叠加信号的检测。PDMA-LSA可极大提升设备连接数量,有效降低非线性接收检测的计算复杂度。此外,PDMA-LSA实现了大规模天线技术与多址接入技术的结合,有利于充分发挥两种技术的潜力,从而区别于其他新型多址接入技术。仿真结果表明,PDMA-LSA相比正交多址接入技术和功率域非正交多址接入技术,即使在未进行资源优化的情况下,也具有一定的性能提升。最后,本论文研究了新型多址接入技术中的资源优化方法,针对提出的PDMA-LSA进行功率控制和波束分配研究。对于功率控制问题,通过推导证明了高能效优化问题的凸性质,从而得到全局最优发射功率控制方案。对于波束分配问题,首先分析波束分配的物理意义,然后证明了发射机中波束赋形与接收机中空域滤波之间的联合特性,并利用此特性对低干扰优化问题实现降维,从而大大降低其求解复杂度。仿真结果表明,优化功率控制对PDMA-LSA的性能具有一定提升,优化波束分配对PDMA-LSA的性能具有显著提升。更多还原

【Abstract】 The rapid development of mobile Internet and Internet of Things puts forward high demands on the spectral efficiency,energy efficiency,access capacity and other indicators of the future mo-bile communication system.As one of the key technologies of the future mobile communication system,the large-scale antenna array technology achieves a substantial increase in the spectrum efficiency by increasing the number of antennas.However,on the one hand,energy consumption of the system grows as the number of antennas grows and then the energy efficiency is impaired.On the other hand,pilot contamination is aggravated as the number of antennas grows and then resource allocation in multi-cell get more complex.As one of the key technologies of the future mobile communication system,the novel multiple access technology greatly improves the spectrum efficiency and access capacity,by resource multiplexing and non-linear detection algorithm.Howev-er,resource multiplexing may reuse more resource domains and introduce more optimization objects and complex interference to resource allocation.In view of these problems,the thesis studies the resource allocation of large-scale antenna array and novel multiple access technologies.Firstly,the thesis investigates the resource allocation method of large-scale antenna array tech-nology.As for the high energy efficiency power control problem in multi-cell uplink scenario,the constrained Markov decision process method is introduced to characterize the energy efficiency,and then the high energy efficiency power control problem is formulated as a constrained Markov decision process based optimization problem.A optimization algorithm is proposed to solve the problem and obtain the globally optimal power control scheme.As for the proposed algorithm,Lagrange method is first utilized to transfer the constrained Markov decision process based optimization problem to Markov decision process based optimization problem,and the improved value iterative algorithm is proposed to achieve the globally optimal power control strategy.In the improved value iterative algorithm,value iterative algorithm is in charge of the inner loop,and Q-learning is in charge of the outer loop.For the proposed algorithm,on the one hand,the achieved optimal decision rules can be utilized to establish an off-line lookup table,so as to avoid frequent computation.On the other hand,relevant parameters in the algorithm can be adjusted to achieve the trade-off between system performance and computational complexity.The simulation results show that the proposed algorithm can achieve the globally optimal performance.Secondly,the thesis investigates the novel multiple access technology,and proposes pattern division multiple access with large-scale antenna array(PDMA-LSA).Large-scale antenna array and Pattern division multiple access are bonded based on the beam resources in the spatial domain.In the proposed scheme,pattern is designed in both power domain and beam domain in a joint manner.Pattern mapping at the transmitter utilizes power allocation and beam allocation to superpose user signals,while hybrid detection of spatial filter and successive interference cancellation is employed at the receiver to separate the superposed multiple-domain signals.The PDMA-LSA reduces the computational load of the non-liner detector while greatly enhancing the access capacity.In addition,the PDMA-LSA spontaneously achieves the integration of large-scale antenna array into multiple access,and can give full play to the potential of the two technologies,which makes it significantly different compared to the other non-orthogonal multiple access schemes.Simulation results show that the PDMA-LSA achieve a slight performance gain compared with orthogonal multiple access and power-domain non-orthogonal multiple access scheme,even in the absence of resource allocation.Finally,the thesis investigates the resource allocation of novel multiple access technology.For the proposed PDMA-LSA,power control and beam allocation are studied.In the research on the power control problem,the convexity of the energy efficiency optimization problem is proved,and then the globally optimal transmit power control scheme is obtained.In the research on the beam allocation problem,the physical significances of the beams are first analysed,and then the low interference optimization problem is reduced dimension by means of technical characteristics of beamforming and spatial filter,which greatly reduces the computation complexity.Simulation results show that the power control method contributes slightly to performance gain,and the beam allocation method achieves significant performance improvement.更多还原