返回
类型 基础研究 预答辩日期 2017-11-18
开始(开题)日期 2015-06-05 论文结束日期 2017-09-06
地点 东南大学健雄院二楼会议室 论文选题来源 973、863项目     论文字数 9 (万字)
题目 多层异构无线通信网络中的资源分配方案研究
主题词 多层异构网络,资源分配,非合作博弈,分布式优化
摘要 随着智能移动终端数量的不断增多和新颖的移动数据服务的不断涌现,对无线通信流量的需求正在以一个爆炸式的速度增长。多层异构网络的出现虽然在一定程度上满足了急速增加的移动终端对数据的需求,但也使得对频谱等资源的使用越来越捉襟见肘。无线资源越来越有限同人们对移动业务要求越来越严格的矛盾日益突出,同时资源的重复利用会导致严重的干扰和资源的不合理分配又弱化了网络的性能。因此,亟需对无线资源进行高效的管理、控制和调度,以满足各种业务需求,保证服务质量(Quality of Service,QoS),提高资源利用率和缓解资源紧缺的情况。随着5G的到来,多种网络的融合和多种新技术的融入使得干扰的场景更加复杂,需要分配的资源类型更多,资源分配方案的研究将会更加迫切。本论文深入研究了“多层异构无线通信网络中的资源分配方案研究”这一课题,针对含小小区和端到端(Device-to-Device,D2D)通信的多层异构网络,考虑基于吞吐量、能效、中断性能、用户关联和回程节省等资源分配问题,研究了多层异构无线通信网络中的高效资源分配方案。本论文研究的内容包括面向异构网络上行吞吐量的联合基站选择和资源分配方案、面向异构网络下行中断性能的资源分配方案、部署 D2D和小基站通信的异构网络中联合信道分配和功率控制方案、面向火车车厢异构通信网络能效的联合子载波和功率控制方案、异构云接入网络中联合用户关联和功率控制方案和空地通信网络中基于主动缓存技术的联合优化方案和部署方案。具体研究内容和主要贡献如下:(1)针对基于正交频分多址接入的异构蜂窝网络(Heterogeneous Cellular Network,HCN),提出了面向网络吞吐量最大化的联合基站关联和资源分配方案,设计了两种基于势博弈的分布式学习算法,并理论证明和分析了收敛到两种均衡解的性能。系统吞吐量最大化问题是一个NP-hard问题,首先引用权重变量将原优化问题转化为一个易于处理的形式,再将转化后的问题建模成一个完全势博弈并证明了存在一个最优纳什均衡点是原问题的一个近似最优解。为了获得原问题的最优解,引入状态空间将原问题重新建模为一个带有状态变量的普通势博弈,并证明了存在一个最大化系统吞吐量的循环状态均衡。其次,分别提出了Max-logit学习算法的两个变体进而获得这两种势博弈的均衡点:一个是用户间较少信息交换的同步学习算法并被证明了能够收敛到完全势博弈的最优纳什均衡点;另一个有效的学习算法被用于带有状态变量的普通势博弈模型中并证明其能够收敛到全局最优解。最后,通过仿真实验,相关理论结果得到了验证。(2)在推导出HCN下行中断概率表达式的情况下,针对基于系统中断性能的优化问题,提出了一个分布式资源分配方案,设计了基于匹配博弈的分布式算法以减轻网络干扰和提高系统性能。在HCN中,部署的小小区基站与宏基站共享频谱资源,导致了严重的跨层干扰和系统性能的损失。具体来说,首先推导出系统中断概率的闭合表达式。然后,提出了一个资源分配问题,进一步降低网络的中断概率。为了实现用户和资源块之间有效的匹配,资源分配问题被建模为一个匹配博弈,其稳定的匹配被认为是资源分配问题的解。最后,为了求解这个匹配问题,提出了一个可以收敛到稳定匹配的分布式算法。分析和仿真结果表明,理论推导的正确性得到了验证和算法的应用进一步提高了HCN的整体性能。(3)针对蜂窝网络中部署有D2D通信和小小区基站的上行通信网络,提出了面向网络吞吐量最大化的联合功率控制和子信道分配方案,推导出了最优功率解的闭合表达式并设计了两种博弈学习算法求解耦合的子信道分配问题。对于多层异构网络吞吐量最大化的资源分配问题,采用基于博弈论的学习方法来求解。然而,该资源分配问题是一个混合有整数的非凸问题,求解其最优解具有高挑战性。为了解决这个难题,首先通过预先设定子信道分配策略,推导出了各种情况下最优功率的闭合表达式。其次,把基于最优功率控制的子信道分配问题构造为一个博弈模型,并且定义了一个能够反映优化目标最优性能的福利函数。为了最大化福利,提出了一种分布式的试错学习算法,并证明其收敛到一个随机稳定状态。由于收敛到的稳定状态并不能保证是最优解,所以将这个子信道分配问题重新构造为一个完全势博弈模型,并提出了另一种分布式学习算法,其能够找到最优的纳什均衡且是问题的最优解。为了加速算法收敛,通过剔除不可用的策略组合,文中进一步改进了这两种算法。最后,仿真结果验证算法具有极快的收敛速度。(4)针对铁路运输通信网络的车厢通信部分,在车厢异构通信网络中含有能够收集能量的D2D通信的场景下,提出了联合子载波分配和功率控制方案,设计了基于博弈论的学习方法求解联合优化问题以优化网络的加权能量效率(Energy Efficiency,EE)。具体来说,首先提出了一个新的能效指标来评估网络EE性能并优化其下界。然而,联合资源分配问题的可行域中混合有整数,这是一个难以处理的棘手问题。为此,通过交替固定子载波和功率分配策略,将联合优化问题分解为两个资源分配子问题。这两个子问题被分别建模为两个完全势博弈,并且分析了它们均衡解的存在性和最优性。接下来,分别提出了一个用于求解功率控制问题的虚拟分布式迭代学习算法,该算法具有较快的收敛速度并推导出了均衡唯一性的存在条件,确保其能够获得全局最优解(即,最优的纳什均衡),以及提出了一个用于求解子载波分配问题的完全分布式Max-logit 迭代算法,并且验证了该算法仅需要交互局部信息就能以任意高概率获得最优的纳什均衡解。通过两种算法的交替迭代操作,最终能够得到联合资源分配问题的最优分配策略。最后,数值仿真结果表明该方案与一些现有的资源分配方案相比具有显著的性能优势。(5)在保证异构云接入网络中用户QoS要求的情况下,提出了面向和速率最大化的联合用户关联和功率分配方案,根据构造的Stackelberg博弈模型设计了基于变分不等式理论的分布式算法求解耦合的功率分配问题。然而,QoS限制的存在使得功率控制的策略空间之间是耦合的。基于此,该联合优化问题被建模为一个广义的Stackelberg博弈,设计了一个集中- 分布式算法,并且理论上证明了所提出的算法收敛到资源分配问题的最优解。具体来说,通过云端强大的计算能力求解用户的关联问题,并基于变分不等式理论提出了一个分布式功率分配算法。最后,仿真结果验证了算法的收敛性,以及所提方案与穷举方法具有相近的性能。(6)在降低传输时延和提供超高吞吐量方面,主动缓存受欢迎内容到基站端已经成为了一个有前景的解决方案。在地面HCN中,通过规划联合服务提升和缓存内容放置策略,提出了有效和速率和回程节省之间的最佳权衡方案,同时保证用户终端的QoS要求和基站的回程流量约束。然而,在这个非线性优化问题中存在有整数可行域这一棘手问题。为此,通过交替固定三类变量中的两个(即地面终端关联,功率控制和内容放置),将资源优化问题分解为三个子问题。针对这三类子问题,分别将它们转化为凸问题或松弛为易于处理的形式,并提出了相应的迭代算法。进而,提出了一个三层迭代算法,用于求解联合地面终端关联、功率控制和缓存放置问题。另外,当地面用户终端关联不到地面基站时,装置有具有缓存功能基站的无人机被部署,为这些终端提供通信服务。在保证QoS要求的情况下,提出了无人机的多目标部署方案(例如传输功率、无人机数目、位置和缓存内容)。具体地,提出了一个基于强化学习的方法,求解多目标的部署问题同时保持功率消耗和回程节省之间的最优折中。最后,数值仿真结果验证了所提方案的有效性。
英文题目
英文主题词 Multiple layer heterogeneous networks, Resource allocation, Non-cooperative game, Distributed optimization
英文摘要 Due to the proliferation of smart mobile equipment and innovative mobile data services, the wireless traffic demands are increasing at an unprecedented pace. Although multi-layer heterogeneous networks partially meet the data requirements of mobile terminals with a speed of rapid increase, the use of wireless resources (i.e., spectrum) will become even more stretched. The limited wireless resources cannot possibly meet the continuous increased requirements of various new mobile services and application cases. In addition, the reuse of wireless resources can lead to serious interference and the unreasonable resource allocations weaken the network performance. Therefore, it is urgently required to manage and allocate the wireless resources efficiently to meet the various application requirements, guarantee the quality of service (QoS), maximize the utilization of wireless resources and alleviate the shortage of resources. Since the integration of multiple networks and the emergence of new technologies make the interferences more complex in the fifth generation mobile communication (5G) network, the research on resource allocation schemes with multiple types of resources will be more urgent. This thesis focuses on resource allocation schemes in multiple layer heterogeneous wireless communication networks. For multiple layer heterogeneous networks with small cell and device-to-device (D2D) nodes, considering sum-rate, energy efficiency (EE), outage performance, user association and backhaul saving, the research of the thesis includes game theoretic max-logit learning approaches for joint base station selection and resource allocation in heterogeneous networks, resource allocation for outage performance in heterogeneous networks, resource optimization for D2D and small cell uplink communications underlaying cellular networks, energy-efficient resource allocation for energy harvesting-based D2D communication, resource optimization in heterogeneous cloud radio access networks, joint service improvement and content placement for cache-enabled heterogeneous cellular networks (HCNs) and the efficient deployment of unmanned aerial vehicle (UAV) for UAV-assisted communication networks. The main contributions of this thesis are listed as follows: (1) We investigate the problem of joint base station selection and resource allocation in an orthogonal frequency division multiple access (OFDMA) HCN. The original throughput maximization problem is NP-hard and we propose to solve it by using game theoretic stochastic learning approaches. To this end, we first transform the original problem into a tractable form which has a weighted utility function. Then we prove that an exact potential game applies and it exists the best Nash equilibria which is a near optimal solution of the original problem. In order to obtain the optimal solution, we redesign the utility function by leveraging a state space to formulate the original problem into an ordinal state based potential game, which is proved that it exists a recurrent state equilibrium point that maximizes system throughput. Furthermore, we propose two different variants of Max-logit learning algorithm based on these two games respectively: one is a simultaneous learning algorithm with less information exchange which achieves the best Nash equilibria point of the exact potential game, the other is an efficient learning algorithm for the ordinal state based potential game which can converge to the global optimization solution. Finally, numerical results are given to validate that theoretical findings. (2) We investigate the downlink outage performance of HCNs and its resource allocation problem to mitigate interferences. In HCNs, the deployed small cell base stations share the same spectrum resource with that of macro base station, which causes cross-tier interferences and the performance depreciation of the system. Specifically, we first derive closed-form expressions of overall outage probability of the system. Then, the resource allocation problem is proposed to reduce the outage probability of HCNs. This problem is formulated as a matching game, and a stable matching is considered to be the solution. Finally, to address this matching problem, a distributed algorithm is proposed which can find a stable matching. Both analytical results and simulations show the accuracy of the derivation and the overall performance improvement of HCNs by using the proposed algorithm. The numerical results finally show the validity of our analysis and that the system performance is further enhanced by using our algorithm. (3) We investigate the joint power control and subchannel allocation problem for D2D and small cell uplink communications in a cellular network, with the aim of improving the network throughput. For this throughput maximization problem, we leverage a game-theoretic learning approach to solve it. However, there is an intractable issue for the feasible region of the optimization problem, which is intermixed with the integer nature. To tackle this problem, we first present the closed-form expressions of the optimal power under different scenarios by presetting the subchannel allocation profile. Based on the optimal power control, we then formulate the subchannel allocation problem as a game theoretical framework, and define a welfare function which has the same optimum as the optimization objective. To optimize the welfare function, a distributed trial and error learning algorithm is proposed to converge to a stochastically stable state. Since the achieved stable state cannot guarantee to be the optimal solution, we reformulate this problem as an exact potential game model and propose another distributed learning algorithm to find the best Nash equilibrium which is the global optimum. Moreover, to accelerate the convergence, these two algorithms are modified by getting rid of these unavailable strategic profiles. Finally, numerical results verify that the proposed algorithms have high convergence rate with a less computation. (4) We address the downlink resource (subcarriers and power jointly) allocation problem for energy harvesting-based D2D communication in a railway carriage communication network to improve the EE of the system. The considered problem is formulated as maximizing the weighted EE and is solved by leveraging a game-theoretic learning approach. Specifically, we first propose a new performance metric for evaluating the EE and optimize its a lower bound. However, there exists an intractable issue of mixing the integer nature into the feasible region. To this end, we decompose the optimization problem into two subproblems by fixing the subcarrier and power allocations alternately. These two subproblems are formulated as two exact potential games, and the optimal properties of their solutions are analyzed. Accordingly, we respectively design a virtual distributed learning algorithm for the power control to find the global optimum (i.e., the best Nash equilibrium) based on the derived conditions of uniqueness of NE which can effectively accelerate convergence, and a fully distributed Max-logit algorithm for the subcarrier allocation to obtain the best NE with an arbitrarily high probability in which only local information needs to be exchanged. Through the alternation of two algorithms and iterative operation, the optimal solution of the problem is achieved. Finally, numerical results show that the proposed schemes have superior performance than the existing ones. (5) We tackle the problem of sum-rate maximization via joint user association and power allocation in heterogeneous cloud radio access networks while guaranteeing the QoS requirements of all users. A generalized Stackelberg game is applied to this problem with the coupled strategy set of power allocation, and a centralized-distributed method is designed to achieve the optimal solution. Specifically, user association can be efficiently solved by cloud computing and a distributed power allocation algorithm is proposed by using variational inequality theory. Finally, numerical results show that the proposed algorithm is convergent and gets a performance closed to the exhaustive method. (6) Caching popular contents in the storage of base stations has emerged as a promising solution for reducing the transmission latency and providing extra-high throughput. We tackle the optimal tradeoff problem between the sum of effective rates and backhaul saving via joint service improvement and content placement in cache-enabled heterogeneous cellular networks while guaranteeing the QoS requirements of all ground user terminals (GUTs) and backhaul traffic constraints of all base stations. However, there exists an intractable issue of mixing the integer nature into the feasible region in the nonlinear optimization problem. To this end, we decompose the optimization problem into three subproblems by alternately fixing two of three class of variables (i.e., GUT association, power control and content placement). Aiming to these subproblems, we respectively convert them into the tractable forms and propose the corresponding algorithms. By combining them, we propose a three-tier iterative algorithm for jointly optimizing GUT association and cache placement. Moreover, when GUTs have failed to be associated with ground base stations, unmanned aerial vehicle (UAV) mounted cache-enabled base stations is used as an effective solution for providing wireless services to these GUTs. The efficient deployment of UAVs (such as transmitting power, number of UAVs, locations and caching) while guaranteeing the quality of service requirements is investigated. Specifically, we propose a reinforcement learning-based approach to solve the multi-objective deployment problem while maintaining the optimal tradeoff between power consumption and backhaul saving. Finally, numerical results have verified the effectiveness of our schemes.
学术讨论
主办单位时间地点报告人报告主题
信息学院数字信号处理实验室 2016.10.10 无线谷5216会议室 代海波 Joint User Association and Power Allocation in Heterogeneous Cloud Radio Access Networks
信息学院数字信号处理实验室 2017.3.20 无线谷5216会议室 代海波 Joint Service Improvement and Placement Optimization for UAV-assisted Ground Networks
信息学院 2015.4.1 无线谷1319会议室 Dr. Yong Zeng Lens Antenna Array Enabled 5G: Performance Improvement and Cost Reduction
信息学院 2015.5.14 无线谷6411会议室 Prof. Dapeng Oliver Wu Multimedia over Future Internet: Challenges and Opportunities
信息学院 2015.6.24 无线谷1319会议室 Prof. Xinyu Zhang Millimeter-wave Wireless Networking and Sensing: A Unified Perspective
信息学院 2015.6.26 无线谷1319会议室 IEEE Fellow Wei Zhang Cooperative Spectrum Sharing Between Cellular and Wireless Ad-hoc Networks
信息学院数字信号处理实验室 2015.11.2 无线谷5216会议室 代海波 Distributed Optimization for Heterogeneous Networks with Massive MIMO: A Generalized Stacklberg Equilibrium Problem Perspective
信息学院数字信号处理实验室 2016.1.15 无线谷5216会议室 代海波 Joint Power Control and Subchannel Allocation for Device-to-Device and Small Cell Uplink Communications Underlaying Cellular Networks
     
学术会议
会议名称时间地点本人报告本人报告题目
2016 IEEE Wireless Communications and Networking Conference (WCNC) 2016.4.4 卡塔尔多哈 Resource Allocation for Device-to-Device and Small Cell Uplink Communication Networks
7th International Conference on Wireless Communications and Signal Processing 2015.10.17 江苏南京 Energy Efficient Multi-Pair Transmission in Large-Scale Multi-Antenna Relay Systems
     
代表作
论文名称
Game Theoretic Max-logit Learning Approaches for Joint Base Station Selection and Resource Allocatio
Energy-Efficient Resource Allocation for Device-to-Device Communication with WPT
Distributed Optimization with Incomplete Information for Heterogeneous Cellular Networks
Energy-Efficient Optimization for Device-to-Device Communication Underlaying Cellular Networks
Resource Allocation for Outage Performance in Heterogeneous Networks: A Matching Game Approach
D2D通信系统中节能功率控制算法
Resource Allocation for Device-to-Device and Small Cell Uplink Communication Networks
 
答辩委员会组成信息
姓名职称导师类别工作单位是否主席备注
郑宝玉 正高 教授 博导 南京邮电大学
徐大专 正高 教授 博导 南京航空航天大学
王保云 正高 教授 博导 南京邮电大学
黄永明 正高 教授 博导 东南大学
李春国 正高 教授 博导 东南大学
      
答辩秘书信息
姓名职称工作单位备注
何世文 其他 讲师 东南大学