Research And Design Of 360-Degree Video Streaming Scheduling Algorithms Based On SVC Coding And DASH

In recent years,with the continuous development of mobile networks,the increasing number of video devices,and the emergence of video applications,video traffic is becoming the predominant of global Internet traffic.Due to the advent of panoramic cameras and the popularity of cheap interactive display devices,360-degree panoramic video streaming has become a popular Internet service.However,providing 360-degree video streaming services on today's Internet still faces severe challenges: high bandwidth and low Motion-to-Photon(MTP)latency(latency between head movement and video rendering)requirements.Besides,a typical dynamic adaptive streaming system uses an MPD file to describe the average bitrates of different quality representations.However,video bitrates vary significantly along the temporal dimension for VBR videos.Thus,the high dynamics of video bitrates that cause bitrate overestimation/underestimation will also affect the efficiency of ABR algorithms.In order to alleviate the negative effect caused by bitrate overestimate/underestimate to ABR algorithms,this thesis discovered that with SVC encoding and tiling,strong video chunk size correlation can be observed on both the layer and spatial dimensions.Quantitative measurement revealed that the enhancement layer chunk size is strongly correlated with the base layer chunk size of the same corresponding tile,and horizontally adjacent tiles are also strongly correlated.Based on these observations,this thesis further proposed to enhance the 360-degree scheduling framework with a chunk size predictor.The chunk size predictor can be used to predict chunks' size in the following two cases:(1)In the case of the same tile,the enhancement layer chunk size can be predicted based on the historical data and the corresponding base layer chunk size;(2)the corresponding video chunk size can be predicted based on historical data and the horizontally adjacent chunk size.Experiments showed that ABR algorithms augmented by a chunk size predictor could significantly improve a user's perceived video playback quality.