Research On Energy Efficiency Optimization Of Mobile Web Application Based On User Experience

With the popularity of embedded devices and the development of Web technology,mobile Web applications are also rapidly developing.In order to improve the user's browsing experience,the performance of mobile Web applications is also constantly improving.The default resource scheduling strategy of the mobile system aims at the lowest response delay and allocates a large amount of computing resources to the Web application,which brings serious energy consumption problems.The study found that reasonable task scheduling can greatly reduce the energy consumption of heterogeneous multi-core embedded devices,but it has not yet been effectively applied in the interactive process of mobile Web.To this end,this paper first studies the working mechanism of mobile Web interaction,and then builds a resource scheduling model,combined with the minimum acceptable rendering frame rate of users in different scenarios to dynamically schedule the Web rendering process,so as to ensure the user experience Effectively reduce energy consumption.The specific research work is as follows:(1)A dynamic resource scheduling model for mobile Web interactive energy consumption optimization based on neural network is proposed.Build a neural network model for typical mobile Web interactive operations and learn what can be achieved when running a computationally intensive Web rendering process with different CPU configurations(clock frequency,processor core)under different usage scenarios(webpage content,event rate)Rendering frame rate(FPS).Then,in the process of interactive operation,the best processor configuration is quickly searched according to the lowest FPS value acceptable to the user,and the CPU resources are scheduled.Experimental results show that the resource scheduling model proposed in this paper reduces system energy consumption by more than 36% on both platforms.(2)This paper proposes a rapid migration strategy for mobile web resource scheduling model.By deploying a pre-built general model,a small amount of new data(user's personalized data)is used to fine-tune the model on this basis,so as to quickly build a resource scheduling model that conforms to the current user's usage habits.Experimental results show that,compared with the general resource scheduling model constructed in this paper,the model migration strategy reduces the average user experience(Qo S)loss rate by more than 5%,and the energy consumption optimization effect is equally good,and the average energy consumption is reduced by 34.6%.At the same time,compared with the latest related methods,the model migration strategy can quickly and accurately build user personalized models by collecting less user data.