随着网络和视频传输技术发展，HTTP流媒体具有广泛应用前景。但由于终端设备的多样性和带宽的时变性，如何实现动态自适应HTTP流媒体传输(Dynamic Adaptive Streaming over HTTP, DASH)已成为国内外的一个研究热点。 本项目将针对DASH码率控制所面临的共性基础问题，在已有的控制理论和视频质量模型等相关工作基础上，研究基于非线性控制的DASH码率动态自适应理论和方法。在模型层面，提出并研究基于缓冲区的码率控制模型，包括非线性缓冲区控制方法、基于主观质量的码率量化方法和公平性控制模型，以解决动态码流切换导致的质量平滑、码率抖动和带宽效率等问题；在方法层面，研究基于非线性控制模型的自适应码流控制方法，特别是针对多服务器并发流传输，研究多流调度算法、多流控制模型和控制优化等问题；在此基础上，开展多服务器DASH关键技术实现和验证。

With the advance of video coding and networking, HTTP video streaming is recently emerging as a promising video delivery method in Internet. But due to diversified bandwidth and time-varying network conditions, it is a great challenge for researchers to stream video with high quality by Dynamic Adaptive Streaming over HTTP (DASH). In this proposal, we will focus on the fundamental issue of DASH, and investigate non-linear control-theoretic models and approaches to rate adaptation for DASH. The proposal consists of threefold: models, algorithms and implementations. In the modeling researching, we will propose and investigate a buffer-based rate control model, which consists of non-linear buffer control, subjective-quality-based rate quantization and fairness control, to achieve rate switching smoothly, less rate oscillation and high bandwidth efficiency. In the algorithm researching, we will investigate non-linear control-theoretic rate control approach. Especially for multiple-server HTTP video streaming, we will address on multi-server scheduling algorithm, multi-streaming rate adaptation modeling and control system optimization. Besides these, we will also carry out implementations and validations of multi-server DASH, to verify the proposed models and algorithms.