Mobility-Aware Proactive QoS Monitoring for Mobile Edge Computing

This article presents a novel probabilistic QoS (Quality of Service) monitoring approach called LSTM-BSPM (DonLSTM-Den based BayeSian Runtime Proactive Monitoring), which is based on the DouLSTM-Den model and Gaussian Hidden Bayesian Classifier for mobile edge environments. A DouLSTM-Den model is designed to predict a user’s trajectory in mobile edge environments. The predicted trajectory is leveraged to obtain the mobility-aware QoS and capture its spatio-temporal dependency. Next, a parent attribute is constructed for each QoS attribute to reduce the influence of dependence between QoS attributes on monitoring accuracy. A Gaussian hidden Bayes classifier is trained for each edge server to proactively monitor the user’s mobility-aware QoS. We conduct a set of experiments respectively upon a public data set and a real-world data set demonstrate the feasibility and effectiveness of the proposed approach.