【作者】 田春长；

【导师】 杨大成；

【作者基本信息】 北京邮电大学 ， 通信与信息系统， 2010， 博士

【摘要】 波束赋形技术的研究开始于20世纪60年代,最初广泛应用于雷达、声纳及军事通信领域,由于价格等因素一直未能普及到其它通信领域。随着移动通信技术和现代数字信号处理技术的发展,数字信号处理芯片处理能力的不断提高,使得利用数字处理技术在基带形成天线波束成为可能。以此代替模拟电路形成天线波束方法,提高了天线系统的可靠性与灵活程度,因此智能天线技术开始在移动通信中得到应用。波束赋形技术最早被用于3G标准之一的TD-SCDMA系统中。在下一代网络中也采用了该项技术,在目前版本的标准中确定了基于用户专用参考信号的单波束赋形技术。在最新的关于后续演进技术的讨论中,多波束赋形得到了极大关注,该技术在提高频谱效率方面显示出了非常大的优势。在下一代网络的众多特征中,本文重点关注如下三项技术的相互融合,分别是基于专用参考信号的波束赋形、下行功率控制技术、以及邻小区同频干扰的控制与协调。通过对仿真网络中的信号特征分析发现,在基本的满功率发射的网络中,由于波束成型技术的引入,信号与干扰之比得到大幅度的提高。特别是在距离基站较近的位置,接收信干噪比远高于最高阶调制编码方式正确接收所需的量。再考虑信干噪比很高时,有效信噪比同发送功率呈现非线性的关系,此时增加发送功率将不会正比增加有效信干噪比。这些现象说明蜂窝系统中存在发送功率过量的情况,即增加额外的发送功率不能带来用户速率的提高,反而对相邻小区造成了干扰。这一观察结果启发我们思考如何在网络中运用功率控制和功率分配的手段充分挖掘由波束赋形技术带来的潜在增益。本文从多个层面研究了同下行波束赋形技术相结合的功率优化问题。首先,我们研究了不需要基站间实时通信的干扰协调方法,分析了波束赋形同静态软频率复用相结合的技术方案。从信号覆盖角度讨论了静态的功率补偿以及划分至中心用户频带的用户数量比例问题。然后,运用博弈论的纳什均衡原理,我们研究了多基站博弈的分布式功率控制技术。本文讨论了两类功率控制基本问题,即确定目标SINR最小化发射总功率问题和联合优化目标SINR与发射功率的问题。针对这两类问题,共讨论了三种博弈,分别是保障误块率的博弈、线形功率定价博弈、最大化系统总容量博弈。然后,分别以基站间不通信、一次通信、两次通信以及三次通信为假设模型,给出了三种博弈在系统中实现的六种分布式功率控制算法。其中,本文重点研究了最大化总容量博弈问题,并对传统系统总容量效用作了改进,提出了一种有效SINR受限的系统总容量效用函数。文中还进一步讨论了软频率复用同分布式功率控制算法相融合的新技术,提出了一种基于分布式功率控制的智能软频率复用技术。最后,本文讨论了在理想条件下,具备集中式控制中心的算法。由于理论上对联合考虑多载波系统的用户调度、波束赋形和功率分配的问题还是一个未解决的开放问题,所以在本文中讨论一种接近最优且复杂度较低的算法。我们运用遗传算法,提出了一种双层资源分配模型,并以此来衡量分布式算法同理论最优算法的性能差距。文章最后一部分对所提出的多种算法使用系统仿真器作了验证。通过研究波束赋形技术结合静态软频率复用的算法可以发现,系统整体性能很难有明显提高,不论是调整功率补偿因子,还是调整中心频带的用户数量,都难以在扇区平均频谱效率和边缘用户频谱效率两方面同时得到改善。通过对确定目标SINR最小化发射总功率问题有关算法的验证发现,该类算法对提高系统速率的作用很小。即便是在获得比较准确的信干噪比预测信息,仍然没有明显的速率改善。与之相比,使用本文提出的有效信干噪比受限最大化总容量分布式功率控制算法能够获得明显的整体速率提升。相比于基本方案,在扇区平均频谱效率方面最高有15.2%的提升,同时边缘用户频谱效率提高了76.1%。进一步,基于分布式功率控制算法的智能软频率复用技术对基站间通信的速率要求较低,但是达到了比基本分布式功率控制算法更好的系统性能。从理想实现方案和简化实现方案的对比也可以看出,当基站间通信数据量降为原来1／4时,后者仍然能够极大改善边缘用户的速率。对集中式算法的验证表明通过理想最优的调度和功率控制都能大大提高系统的频谱效率。然而同时采用这两种技术并不是两部分增益的叠加。这说明采用集中式调度后,功率控制的优化空间就会变小。采用单项技术的增益已经接近于可获得的最大系统增益。最后,对比智能软频率复用技术和集中功率控制技术的系统性能,可以看到本文提出的智能软频率复用在扇区平均频谱效率方面最高可获得15.5%的提升,已经非常接近理想集中功率控制方案所能获得的20.5%的增益。边缘用户频谱效率方面,智能软频率复用技术相比基本方案提升了77.6%,同时也比较接近集中功率算法118%的增益。更多还原

【Abstract】 Beamforming technology began in the 20th century,60’s. It is first widely used in radar, sonar and military communications. As modern digital signal processing technology developed, it makes the use of beamforming in personal communication possible. Beamforming technology was firstly applied in the 3G standard TD-SCDMA system. In the next-generation network, it has been used again. In current version of the standards, UE-specific reference signal based single stream beamforming technology has been identified. Moreover, in the latest evolution of technology, beamforming has been great concerned that the technology highly improved spectrum efficiency.In so many features of next-generation networks, this article focuses on three technologies. They are UE-specific reference signal based beamforming, downlink power control, and inter-cell interference coordination. Through the analysis of network signal characteristic, it is showed that in the full power transmission mode, due to the introduction of beamforming, the signal interference ratio is greatly enhanced. Especially in the close position from the serving cell center, received SINR is much higher than the amount required by highest modulation level. Considering SINR is high, the received SINR will be in nonlinear relationship with the transmission power. That means increasing transmission power will not increase the effective SINR. This observation inspired us to think about how to use the network power control and power allocation methods to tap the potential gain. In this paper, we study on a series of the downlink beamforming technology with combination of power optimization problem. First, we studied interference coordination methods without real-time communication between the base stations. We make an analysis on beamforming combined with soft frequency reuse solution. Further discussions have been made on the perspective of static power compensation and user ratio of the cell-center frequency band. Then, using game theory, we study the distributed power control game of multiple base stations. This article discusses two types of power control problem, namely, fixed target SINR minimizing total power issues and the joint target SINR and transmit power optimization problem. To solve these problems, we discussed three games respectively. They are block error rate game, linear power pricing game, and maximizing total capacity game. Then, assuming not communicate between base station, one time of communication, two times of communications or three times of communications as the basic models, six kinds of distributed power control algorithm have been achieved. Among them, the paper focused on the maximum total capacity Game, and that game had been improved. We proposed a novel effective SINR constrained total system capacity utility function. Since then, the paper further discusses the integration of soft frequency reuse and distributed power control algorithm. A novel scheme namely smart soft-frequency reuse technology has been proposed. Finally, the article discusses the ideal conditions with centralized control center. For the optimal solution of joint considering user scheduling, beamforming and power allocation is still an unsolved open problem, we discussed a close to optimal but low complexity algorithms. We proposed a two-tier resource allocation model using the principle of genetic algorithm, so as to measure the performance gap between the Distributed algorithm and the optimal algorithm in theory.In the last part, the proposed algorithms were verified using the system simulator. For the combination of Beamforming and SFR, we found that it is difficult to improve overall performance significantly, whether to adjust the power compensation factor, or adjust the UE ratio. For the fixed target SINR minimize total power algorithms, we found that such algorithms improve system performance little. Even access to more accurate information of SINR, the rate is still not significantly improved. In contrast, the proposed distributed power control algorithms can achieve significantly improvement of the overall rate. Compared with the basic program, the cell-average spectral efficiency made up to 15.2% increase, while cell-edge users’ spectrum efficiency improved 76.1%. Further, the proposed smart soft frequency reuse algorithm demands a lower rate between base stations, but reached a little better system performance than common distributed power control algorithms. At last, the centralized algorithm was verified. It is observed that optimal scheduling and optimal power control both greatly improve the system spectrum efficiency. However, the gain by using both technologies is not the addition of individual gains. This indicates that after using centralized scheduling, the space of power control optimization will be decreased. Finally, the contrast of smart soft frequency reuse and centralized power control technology has been made. We can see that the proposed smart soft frequency reuse scheme achieved the highest enhancement of 15.5% in cell-average spectrum efficiency, which is very close to the ideal power control program’s 20.5% gain. For cell-edge user spectral efficiency, the proposed smart soft frequency reuse scheme achieved 77.6% gain compared to basic program, which is close to 118% gain of ideal power control program.更多还原