Research On MEC Resource Management Algorithm Based On User Mobility Prediction

With the emergence of artificial intelligence equipment and the rapid development of wireless communication technology,how to improve resource utilization and service quality has become the key.As a new computing paradigm,Mobile Edge Computing(MEC)emphasizes that edge servers with computing resources and mobile end users can achieve close distance,aiming to reduce the delay of mobile end users ’ request task processing and reduce network congestion caused by unreasonable use of resources.However,the computing resources and energy of the mobile terminal itself are limited,and its self-interest leads to resource grabbing,which greatly reduces the utilization rate of resources and causes waste of resources.In addition,the uncertainty of user movement makes it impossible to guarantee the activity within the coverage of the edge server.If the mobile user does not receive the processing result after updating the location,it will send the computing task request again,which will also cause waste of resources.Therefore,the mobile edge computing resource management algorithm based on user mobility prediction is worthy of our in-depth study.Using the convenience of mobile communication to obtain data,this paper proposes a mobility prediction service MPS-MEC(Mobility Prediction Service provided by MEC)framework,which provides mobility prediction services for third parties and can improve the mobility management function in mobile networks.Based on the prediction service framework,a trajectory prediction model based on Long Short-Term Memory(LSTM)is proposed,which uses its long-term memory analysis function to predict the historical trajectory of mobile users.The simulation results of the public dataset show that the prediction model has high accuracy.This paper further studies the problem of resource waste and unreasonable resource management caused by mobile users ’ inability to guarantee the activities of edge servers in the coverage area.In the mobile model,in order to more conveniently study the relationship between distance and connectivity between edge servers and mobile users,this paper uses a single mobile user system to study the problem of connectivity.A computational resource management algorithm based on long short-term memory and unscented Kalman filter algorithm(LSTM-UKF)prediction is proposed.According to the mobile user ’s own computing task,the local calculation or offloading calculation is selected,and the delay and energy consumption model is established.UKF is mainly integrated into the LSTM prediction model,and the update parameters are constantly revised under the limited number of iterations.The best position of mobile users in nonlinear motion is predicted and the nearest edge server position is provided,and the edge server is reasonably allocated to improve resource utilization.The simulation results show that the connection success rate of LSTM-UKF algorithm is significantly higher than that of Kalman filter and extended Kalman filter.