Adaptive Bandwidth System with SCTP for Scalable Multi-view Video Streaming

• Main Authors: Chen-An Liu, 劉哲安
• Other Authors: Yuan-Sun Chu
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/6em2vk

碩士 === 國立中正大學 === 電機工程研究所 === 101 === Due to the development of image technology, the amount of data of High-definition 2D video, full view video, and 3D video are more than traditional video, the bandwidth to transfer video data also have to increase. However bandwidth is limited .When bandwidth is not insufficient, it may result in delay and loss screen because video data can't be immediately sent to the client. When the bandwidth is sufficient, if the default is too small amount of data transferred, it will result in poor video quality. Although H.264/SVC has the characteristics of the multi-layer coding and provides spatial, temporal and quality scalability, it focuses on single view video and has the question of compression ratio, decoding times and bandwidth allocation extended to multi-view video. H.264/MVC compresses multi-view video efficiently, can get all of view video by decoding once. Compares with H.264/SVC, H.264/MVC lacks spatial, temporal and quality scalability and has poor adaptability to unstable changes of network bandwidth. SCTP protocol has multi-homing、multi-streaming、partial-reliable three characteristic. Using multi-streaming separate I、P、B frame to different logic channel. Set different lifetime to different logic channel. Smart phone and notebook have tow network card nowadays .We can use this characteristic to establish multi-homing improve the quality of video. The system in this paper is based on multi-view video coding, and joins temporal and quality scalability in scalable video coding, improves the bandwidth adaptability of multi-view video streaming. SCTP establish multi-homing that we can use the backup channel when the primary channel is disconnect. Experiments show that, it provides more than 33.8dB of PSNR and doesn’t influence user watching fluency when the primary channel disconnect.