Research On Network Caching Mechanism Of Immersive Video

Immersive video is a new type of media that enables users to freely explore a full range of perspectives,interact with video content,and create a real sense of immersion.In recent years,immersive video has developed rapidly.Immersive video is delay sensitive and hunger for bandwidth when transmitted over the current network architecture.It costs over 300 times more bandwidth to transmit a full immersive video than an ordinary 2D video,and less than 20 ms delay in order to prevent causing dizziness in the client side.These transmission requirements exceed the capabilities that the existing network can support.For instance,the LTE system has a peak downstream rate of 100 Mbps at 20 MHz bandwidth,which is far from the 400 Mbps required for smooth playback of the immersive video.The cache system can help bridge the distance between data files and users,reducing the amount of data transmitted by the backbone network and reducing the transmission delay,which is a powerful tool to solve the above challenges.This paper studies the cache system of immersive video and analyzes the viewing behaviors of immersive video users.Furthermore,this paper proposes an immersive video version caching algorithm and a hybrid tile caching strategy,on the basis of exhaustive analysis of two different encoding methods of immersive video.The main contents of this dissertation are as follows:(1)This paper proposes a user-aware immersive video version caching scheme.In the existing research,the immersive video version of Quality Emphasis Center(QEC)is uniformly fixed on the spherical surface.However,in the real world,the user's viewpoint is unevenly distributed on the sphere,and the distribution will change during the playback.Based on this observation,this paper clusters historical user viewpoints to determine the position of the video version QEC,and obtains the relationship between the number of clusters/version and the user's viewing quality.On this basis,this paper builds an optimization problem and proposes the corresponding heuristic algorithm to allocate the number of versions to each frame of the video when given cache space.The numerical results show that,compared with the existing scheme,our scheme can improve the users' viewing quality to some extent under different cache spaces.(2)This paper proposes a user-aware hybrid tile caching scheme.Existing immersive video tile caching schemes decide cache policy in terms of the popularity of tiles.In view of the characteristics of the tile cache,this paper improves the existing indicators that measure viewing quality that version encoding brings to the users.Compared with the traditional user request probability,the improved measure proved to be more effective in distinguishing tile importance.On this basis,combined with the compression ratio of the tiles,this paper proposes a hybrid tile caching scheme which builds an optimization model to maximize the user's viewing quality,this model caches tiles with different cut granularity when given a certain cache size.At the same time,this paper gives a heuristic algorithm to solve the optimization model.The numerical results show that,compared with the scheme that uses request probability to cache tiles,our scheme can effectively improve users' viewing quality when the cache size is insufficient,and reduce the amount of data that needs to be transmitted from the video source server due to cache misses.When the cache size is sufficient,the performance of our scheme is similar to current schemes.