Research On Multi-dimensional Service Quality Prediction Method In Mobile Edge Computing Environment

Accompanied by technology of mobile edge computing maturity,numerous services in different fields have appeared on the network.Meanwhile,with the advent of the Internet era,the demand for Internet services is increasing,and people ’s life and work depend on Internet to complete.A problem appears: when users need reliable services,it is difficult to find a specific one to meet their needs in the environment of mobile edge computing,and that seriously affects the user satisfaction and the utilization of service resources.Quality of Service(QoS)has become an important indicator to measure the pros and cons of services when faced with similar types of services.Therefore,how to forecast the trend of QoS to give priority in the MEC environment and recommend suitable services to users has become one of the important challenges in the field of service computing.In response to this problem,this article proposes a multi-dimensional QoS prediction method which based on long short memory network and multi task deep neural network.Firstly,construct a multi-dimensional scenario model according to the change of scenario information in MEC environment.Secondly,build a prediction model based on long short memory network(LSTM)to predict the scenario information that has great changes over time,such as service load,task queuing and network transmission rate.Finally,according to the combined results of predictied scenario information and QoS attributes,and use the prediction model based on multi task deep neural network to predict multiple QoS attributes in the future.The research is intended to deepen the combination of artificial intelligence and service computing which has theoretical and practical value.The main contents of this paper are as follows:1.A scenario information prediction method based on LSTM is proposed.Firstly,a multi-dimensional scenario model is constructed according to the characteristics of diverse and fast changing scenario information in MEC environment.Then,by collecting the data of historical service load,queuing volume and network transmission rate,the prediction model of service load,queuing volume and network transmission rate after a period of time is constructed to predict the future service load and other information.2.A QoS prediction method based on multi task deep neural network is proposed to predict the multi QoS attributes of services.According to the characteristics of more scenario information and QoS attributes,the multi task learning method suitable for multi input multi output problem is adopted,and the multi task deep neural network is used to predict the multi QoS attributes in multi-dimensional scenarios.