Research On Optimization Of Video QoE Based On Load Balancing

The continuous innovation of Internet technology promotes the development of mobile video service.With the emergence of various communication and video service software,mobile data occupies the main position of market traffic.For application developers,how to make the user group get better video quality of experience(QOE)is a key step for business development.Traditional network quality of service(QoS)focuses on ensuring the stability of video transmission through technology and improving the quality of user experience.Users' subjective feelings when watching video through mobile clients have a great impact on QoE,but they can not directly use QoS indicators to measure users' real QoE.Therefore,it is of practical significance to study the key indicators of network service quality to predict users' video experience quality.The main research content of this thesis includes the network load problem of the video transmission process and the prediction and optimization of user QoE based on QoS key indicators.The main work is summarized as follows:First of all to the traditional load scheduling algorithm,the principle of ant colony and particle swarm algorithm combined with the specific mathematical theory and model analysis,deficiencies existing in the study algorithm,puts forward improvement strategy,combined with the specific application scenarios,through the cloud computing tasks scheduling simulation experiment platform to improve the algorithm,the experimental results show that the The improved algorithm reduces the task execution time,reduces the task transmission delay and memory occupancy,improves the load balancing degree during network transmission,and improves the QoS of the network layer.Secondly,through the study of QoE related evaluation technology,combined with deep learning algorithms,select QoS indicators that have a greater impact on the user's QoE for modeling and analysis,and respectively propose QoE prediction models based on BP neural network,wavelet neural network and LSTM neural network to compare The user experience quality is analyzed,and the experimental prediction results show that the prediction effect of the LSTM neural network algorithm is better than that of the BP and wavelet network.Finally,this thesis on the basis of the existing scientific research project work,set up based on the mobile client live video system in simulation experiment,and combines the actual conditions of three kinds of neural network algorithm to predict the output optimize the QoE,through load balancing module to reduce the video data transmission delay and packet loss rate,enhance user experience quality video.