| In recent years,with the continuous development of virtual reality technology,panoramic videos have encountered new opportunities for development.Panoramic videos played on VR devices can provide users with immersive sensory and interactive experiences,making it a powerful competitor as the next-generation mainstream content carrier.However,due to the large data volume and high bandwidth requirements of panoramic videos,playing them on the internet is still a challenging task.Although some solutions use tiling technology and Scalable coding technology to reduce bandwidth consumption during transmission,there is still a large amount of redundant data being transmitted due to the accuracy of view prediction.Therefore,playing panoramic videos on the internet remains challenging.To further reduce redundant data,this thesis combines the characteristics of panoramic videos,scalable video coding technology,and edge computing to design a new panoramic video on-demand framework and related cache algorithm strategies,including edge caching algorithm,edge adaptive response algorithms,edge cache replacement policies,and edge proactive pre-caching strategies.The research in this thesis mainly includes the following aspects:(1)This article proposes a collaborative caching model for multi-edge nodes.The model breaks the independence of the shared area,and each edge node shares cache content with the surrounding edge nodes within a fixed distance with itself as the center,thereby achieving true close cooperation and improving cache efficiency.An open collaboration space helps improve the scalability of edge networks.(2)This article proposes a new framework for panoramic video on-demand.Compared to existing solutions,this article innovatively combines the characteristics of edge computing,transferring the cache decision process to edge servers and abandoning the pre-caching of large amounts of video content on the client side.By using the smooth and stable network characteristics between edge servers and clients,an edge buffer algorithm is designed to pre-cache content with greater predicted cache gains to edge storage spaces,compensating for the shortcomings of local caching.Among them,the factors affecting the predicted cache gain mainly depend on the user’s current state and the system’s historical data,and aim to increase the predicted cache gains for recent demand,in-view demand,and basic layer demand,achieving the goal of caching most of the recent demand and a small portion of future demand.Finally,an edge adaptive response algorithm is used to fill the gaps in edge pre-caching,ensuring high-quality and smooth playback on the client side.(3)To reduce redundant data transmission,various approaches have been adopted in this article.Firstly,the response viewport has been reduced in size to eliminate redundant data outside of the visible range.Secondly,scalable coding technology has been employed to encode panoramic videos,reducing redundant data between different picture qualities.In addition,the multi-edge cache collaborative mode has been used to effectively reduce redundant data transmission among edge servers.Furthermore,transferring the cache area from the client to the edge server can further reduce redundant data transmission between clients.Ultimately,these measures provide significant bandwidth advantages for the proposed panorama video on-demand solution.(4)This article addresses the issue of low cache hit rates in edge caching during the early stages of video on demand.It proposes utilizing idle network bandwidth periodically,caching certain video content in edge cache spaces based on the activity of edge nodes and the characteristics of service users.The inclusion of this algorithm has shortened the buffering time for video in some scenarios,further improving the performance of panoramic video on demand solutions. |
