Scalable Video Coding(SVC)是延伸於H.264/AVC影像編碼技術，不但擁有低位元輸出率、高抗錯性能和網路友善性等優點，還採用分層編碼概念將不同特性的bit-stream送到IP網路上。因為擁有網路友善性機制，使之非常適合應用在異質網路。另外，SVC提供三種可調性：空間、時間與訊噪比可調性，因此可以根據網路狀況與用戶端的裝置動態調整最適當的bit-stream。 即使H.264視訊封包的標頭會告知其封包的重要性資訊，但現存的OSI分層模型中MAC層和PHY層無法辨別應用層視訊封包的重要性，因此當網路流量開始壅塞，容易發生因遺失對影片影響度較大的視訊封包而導致視訊品質劇烈下降的情形。 因此，本文提出在IEEE 802.16e網路之跨多層機制使應用層中不同重要度的視訊封包能映射到網路層、MAC層，使MAC層能辨別視訊封包的重要性而給予相對應的服務品質保證。由實驗數據得知，如此架構可以大大提升用戶端的視訊品質。 The Scalable Video Coding (SVC) is extended from the H.264/AVC video coding standard. It is very suitable for deployed on diverse access networks because of its network-friendly nature. Besides, by means of the spatial, temporal and quality scalabilities of SVC, a terminal station would receive the best suitable combination of these layer representations according to the radio channel condition and the capacity of the specific terminal station dynamically. Under the concept of layer-separation of the OSI layer model, however, existing access network techniques would be upper layer unaware, that is, the PHY and MAC layer could not recognize the importance of dependency implied to each NAL Unit in decoding process. Thus, while the traffic loading in the access network becomes heavy, the video quality would degrade dramatically due to the upper layer-unaware nature of the existing systems. Here, we propose a cross multi-layer mechanism to let the MAC layer of a IEEE 802.16e network system become aware to the importance of decoding dependency of each NAL Unit and make a suitable decision to the corresponding resource allocation. According to the simulation results, it is obvious that the video quality would be improved significantly.
