| With the popularity of mobile terminals and the advent of the 5G era,the mobile streaming media industry has ushered in leaps and bounds,and has been widely used in video live broadcast,remote education,and security monitoring.In order to meet the needs of mobile users for high-quality video content,the video bit rates released by content providers are getting higher and higher,but this requires high-quality network conditions and high-performance encoders as support.At the same time,in the process of enjoying high-quality streaming media services,massive mobile traffic and intensive computing tasks have also placed a heavy burden on mobile device batteries.In this regard,this article is oriented to mobile heterogeneous platforms.According to the current network status of mobile devices,the complexity of video content,the amount of cached data,and the remaining power,the most reasonable CPU configuration and video bit rate are selected to improve the quality of user experience while reducing energy consumption.The research work is as follows:1.This paper is oriented to mobile heterogeneous platforms,and proposes an energy optimization model based on Q-learning for mobile streaming media applications.This model comprehensively considers the current network status,video buffer status,and current power of the device.Through the reinforcement learning model,the mobile streaming media calculation-intensive data loading process is dynamically scheduled.Experimental results show that,compared with HMP,the default CPU scheduling strategy of mobile operating systems,this method reduces energy consumption by 14% on average.2.Aiming at the problem that the test standard of the QoE evaluation model of the traditional streaming media service cannot meet the application of adaptive streaming media with bit rate,an adaptive streaming media user experience quality evaluation model based on BP neural network is proposed.First,by testing the user 's real subjective influence factors in adaptive streaming media video,the key influencing factors affecting adaptive streaming media are selected.Secondly,BP neural network is used for subjective test sample learning,to determine the weight ofkey factors affecting the user experience of adaptive streaming media services,and to establish an adaptive streaming media user experience quality prediction model.Finally,through comparative experiments,it is proved that the model has an average fitting degree of 0.92,which can effectively predict the subjective perceived MOS value of adaptive streaming media services.3.Aiming at the problems of frequent interruption of mobile adaptive streaming media technology,frequent bit rate fluctuations,and poor adaptability of complex networks,a video bit rate adaptive adjustment algorithm based on user experience is proposed.Subjective perception test experiments have found that the impact of video complexity on user's subjective experience at different bitrate levels is very different.Therefore,the algorithm reduces the number of video interruptions and bit rate switching frequency by re-dividing the level of the buffer area and introducing the judgment conditions of the complexity of the video picture.The experiment is verified in two kinds of bandwidth fluctuation environments,and the subjective MOS value of the DASH streaming service is increased by about 12% without affecting the average video bit rate. |
