Research On Multi-view Video Caching Strategy In Edge Computing Environment

With the development of video coding technology,people's demand for watching video is not only stay in two-dimensional video,but also three-dimensional video and virtual reality video gradually enter people's life.As a common form of 3D video and virtual reality video,multi-view video is captured by multiple cameras at different positions in the same scene.Multi-view video is more in line with the needs of users and can provide better viewing experience for users,which is the development trend of video technology in the future.However,multi-view video needs a large amount of view data to support the user's demand of switching any view.Therefore,in order to make multi-view video as popular as two-dimensional video,we need to solve the problem of view data caching in local.Edge computing,as a technology to provide storage and computing resources and reduce latency near the user's network edge,is often used to solve the problem of video transmission and caching with huge amount of data.At present,there are three factors to consider when designing caching strategy for multi-view video in edge computing environment:user viewing history(popularity),delay and view distance.However,these caching strategies can only reflect the impact of video popularity and viewpoint popularity on the caching strategy,and the consideration of viewpoint distance is just a simple consideration.After caching one viewpoint,other viewpoints can be synthesized to increase the cache hit rate.Therefore,to solve the above two problems,this paper designs four innovations of multi-view video caching strategy in edge computing environment based on popularity and view distance algorithm1)Popularity algorithm and viewpoint distance algorithm.Based on the idea of improving cache hit rate and reducing latency,a popularity algorithm is designed.This paper mainly explores the relationship between video popularity,video segment popularity,view popularity,view proximity and video block popularity,which can reflect user preferences,and the impact of the four popularity on the caching strategy.At the same time,based on the characteristics of multi-view video,a view distance algorithm is designed.For multi-view videos with different popularity and different number of views,an optimal view distance is flexibly set for caching,so as to get more views that users can choose,improve the cache hit rate,and improve the quality of user experience.2)Cache policy.In the edge computing environment,according to the characteristics of multi-view video,a caching strategy is designed based on the reasonable layout of multi-view video cache location and cache content,so that the edge computing collaborative cache domain can provide services more efficiently and improve the quality of user experience.In this cache strategy,we design a pre-cache strategy for new video,and a cache replacement strategy when the cache space is insufficient.3)Cost calculation algorithm and cache revenue algorithm.In order to provide users with the least cost video segment from the cache system,this paper designs a cost calculation algorithm considering the synthesis cost,transmission cost and cache cost.The cache revenue algorithm is designed to select the video data to be replaced in the cache replacement strategy,taking into account the number of synthesized views,the quality improvement,the reduction of calculation consumption and the reduction of view acquisition delay.4)Using Python language to design the simulation experiment under the layered MEC network environment,the multi-view video caching strategy in the edge computing environment is fully implemented.Then,by adjusting the popularity calculation and viewpoint distance calculation,the overall comparison of the pre caching strategy in this paper is made.The proposed strategy is compared with the best performance algorithms in FIFO,LIFO and LRU.It shows that the cache strategy adopted in this paper is superior to other strategies in terms of multi-view video cache hit rate,user experience quality(quality and delay)and cache replacement rate in edge computing environment.