无线接入网QoS保证的节能控制建模与协同策略优化

With the integration of telecom, broadcasting and Internet networks, wireless access networks are proliferating deployed to support multiple integrated services. Energy consumption and QoS provisioning are becoming crucial constraints in the design and operation of wireless access networks. This research investigates the modeling, performance analysis and cooperative optimization method of energy saving control with QoS guarantee for addressing this issue. First, an energy saving control scheme that combines dynamic power management (DPM) and adaptive bandwidth allocation (ABA) is presented. DPM adaptively adjust the energy consumption level/communication capacity with the fluctuation of workloads by conducting frequency carriers switching on/off and base stations sleep/awake, while ABA reallocates the bandwidth of ongoing calls to adapt to the variation of available communication capacity. The cooperation of DPM and ABA aims to reduce energy consumption while to guarantee QoS simultaneously. Then, an analytical model based on semi-Markov switching control processes is introduced to formulate the energy saving control problem. Under this framework, events are flexibly defined to driven DPM and ABA to interact with environment promptly, and the system dynamic is captured accurately with a hierarchical structure. The optimum of energy saving under multiple QoS constraints is achieved by cooperative optimization of the event-driven policies defined at different layers. Finally, adaptive algorithms are developed based on event-based optimization approach to perform the cooperative optimization online. The feature of event-driven policies and the information of dynamic hierarchy are exploited to release the dependence on the knowledge of environment and system parameters, and to reduce the computation complexity considerably. This research proposes a high energy-efficient control scheme for adaptive resource management of wireless accesses networks, and provides a feasible solution to a class of common and key technology problems in energy saving control of network communication systems.

面向三网融合背景中无线接入网低能耗、高性能的设计和运行需求，研究具有可靠QoS保障的节能控制建模、性能分析及多策略协同优化方法。采用动态功耗管理、适应带宽配置等资源管控策略对基站的能耗级/通信容量、业务呼叫使用带宽进行自适应调整，驱动系统资源与波动业务需求的有效匹配。构建层级化半Markov切换控制过程模型，建立基于事件反馈和层级协同的控制结构和分析优化框架。引入大系统理论方法进行策略解耦，通过并行协同优化实现QoS性能约束下的能耗最优化。研究多策略协同的自适应优化算法及其在线分布式并行实现，利用模型的层次化动态结构信息和事件驱动的策略特性，摆脱优化算法对环境和系统参数的依赖，大幅降低计算复杂度，提高计算实时性和在线实现的灵活性。项目研究为无线接入网提供高能效的自适应资源管理控制决策方案，促进网络通信系统节能控制中一类具有共性的关键技术问题的解决。

随着通信技术的发展和应用需求的推动，提供多媒体业务泛在接入的无线接入网日益普及，无线网络流量急剧增长带来的高能耗问题引发对绿色无线通信的迫切需求。由于应用环境的复杂性和业务服务的多样性，提高能效并提供各类业务可靠QoS保证成为无线接入网设计和运行的一项颇具挑战性的研究课题。本项目面向无线接入网高能效的自适应资源管理，从系统与控制的角度，开展QoS保证的节能控制建模、性能分析和协同策略优化研究。（1）结合系统级的动态功耗管理、传输层的可伸缩视频编码以及链路层的传输速率控制，提出跨层协同的具有QoS保证的节能控制机制。（2）针对无线接入网系统的动态特性，提出事件驱动的涵盖半Markov过程、离散时间Markov过程和部分可观Markov过程的切换控制系统分析模型，对多种类型的层次化系统动态具有广泛的描述能力，为复杂应用环境中无线接入网具有可靠QoS保证的协同控制提供了有效的性能分析和策略优化框架。（3）引入事件性能势的概念，推导出切换控制过程的性能灵敏度公式，提出基于策略梯度和策略迭代的在线自适应优化算法，给出算法的计算复杂度分析和收敛性证明。提出梯度矢量合成法和约束策略空间在线辨识法直接求解具有随机型性能指标约束的优化问题。所提算法具有收敛速度快、计算量小、自适应性强的特点，适用于未知且时变环境中的在线自适应优化。（4）对基站休眠和载频开启控制、可伸缩视频编码流媒体跨层协同的认知无线传输、视频流媒体编码层的选择传输控制、自适应传输速率控制、移动云计算随机任务序列的执行调度、高能效的无线中继传输等无线接入网的研究热点问题提出了切实可行的优化控制方案。项目研究有助于促进网络通信系统节能控制中一类具有共性的关键技术问题的解决，推动相关理论探索和应用研究的深入开展。