QoS保障的视频流媒体认知无线传输的控制建模与策略优化

面向高动态的复杂认知无线环境，研究可伸缩视频流媒体具有可靠QoS保障的传输控制建模、性能分析与策略优化方法。采用结合认知信道配置、发射功率控制、调制编码选择和视频编码调节的跨层协同的资源自适应管控，构成视频传输的QoS保障和性能优化机制。构建事件驱动的部分可观Markov控制过程的系统分析模型，有效描述信道感知和状态估计具有随机误差的认知无线视频传输系统的动态演化并保留可用于自适应优化的动态信息。基于事件性能势建立一阶矩和二阶中心矩测度的性能灵敏度分析体系，为多种类型的QoS保障提供严谨的性能分析和优化框架。结合在线辨识满足统计型约束的策略空间，提出基于随机学习的快速求解带约束优化问题的在线自适应算法，切实提高优化控制在线实现的可行性。项目研究为5G通信中需求快速增长的视频流媒体认知无线传输提供有效的优化控制方案，促进高动态异构无线网自适应资源管理中一类具有共性的关键技术问题的解决。

This research investigates the modeling, performance analysis, and policy optimization for QoS-guaranteed scalable video streaming over cognitive radio (CR) in complex dynamic environments. Combined with CR channel allocation, transmit power control, adaptive modulation and coding, and video encoding adjustment, a cross-layer resource adaptive management scheme is presented for CR video streaming. An analytical model based on event-driven partially observable Markov control processes is introduced to formulate the control of CR video streaming. It efficiently captures the dynamics evolution of CR systems with random errors in channel sensing and state estimation, and conserves more available dynamics information that could be exploited for adaptive optimization. Performance sensitivity analysis based on event potentials is established for mean and variance criterions, aiming to provide a rigorous analysis and optimization framework for provisioning multi-type QoS perceived to user experience. Combined with online identification of the subspace of policies satisfying the statistical constraints, stochastic learning based adaptive algorithms are developed to directly solve the constrained optimization problems. This research provides an efficient control and optimization scheme for QoS-guaranteed CR video streaming to address the dramatically increasing demands in 5G communications, and contributes to the solutions of a class of common and key technical problems in adaptive resource management of highly dynamic heterogenous wireless networks.