| In recent years,with the popularity of live video,various live platforms such as Douyu,Tik Tok,and Kuaishou have emerged,attracting more attention to live video services,and the demand for live video from users is increasing.Everyone can become a producer and consumer of video content,resulting in a huge production and consumption of live video content.However,the difference between live video and on-demand video is that live video content is generated and transmitted in real-time,which poses great challenges to the transmission of live video.Mobile Edge Computing(MEC)servers,with their proximity to users and faster response to user requests,and the ability for edge servers to collaborate to provide computing and storage resources,have become the preferred choice for solving the above problems.Pre-caching on MEC is a good solution to improve user viewing experience.However,MEC server resources are limited,which means that it is almost impossible to pre-cache all video content,and existing solutions have not effectively considered user preferences and actual situations for pre-caching decisions.Therefore,making pre-caching decisions based on user preferences can better meet the demand for improving user experience quality.In addition,the network may experience jitter during user viewing,which affects the user’s viewing experience.Adaptive bitrate decision-making is a good solution to improve user viewing experience.To address the two issues mentioned above,this thesis proposes a user preference-based pre-caching strategy(PCS)based on edge computing and video transmission optimization theory.This solution ensures a high cache hit rate while significantly improving the user’s initial startup delay.Based on the differences in user demand for watching live broadcasts,an adaptive bitrate decision-making solution based on user preference(BARUP)is proposed,which demonstrates the ability to improve user Quality of Experience(QoE)when watching live video services under different network conditions.The research content of this thesis mainly includes the following aspects.(1)In terms of pre-caching strategy,this thesis proposes a user preference-based precaching scheme for live video under the edge-cluster collaborative network architecture.In addition to considering the popularity of the live video itself,the scheme also analyzes the impact of user preference factors on watching live streams from the user’s perspective.In short,this scheme integrates the influence of both live video content itself and user factors,which in turn affects the decision of live pre-caching.This solution not only improves the resource allocation efficiency of MEC,but also effectively reduces the initial start-up delay of users watching live streaming,which brings great viewing experience to users watching live streaming.In addition,the performance of this scheme in terms of cache hit rate of pre-caching is greatly improved compared to other schemes.(2)In response to the phenomenon of network fluctuations that may cause video stuttering,delays,and other problems while users are watching live video.This thesis proposes a user demand-based adaptive bitrate(ABR)solution.During the process of making adaptive decisions,this thesis discovers and analyzes the impact of user personalized demand on ABR decisions,and further adjusts the adaptive decision-making solution from this perspective to meet the adaptive decision-making needs of different users.This solution exhibits more intelligent performance in terms of personalized adaptive decision-making for users compared to other solutions.Experimental results show that under different network conditions,this solution can effectively improve user QoE and satisfy their personalized demand compared to other adaptive decision-making solutions. |
