Server-Side Distinction of User Mobility Using Machine Learning on Incoming Data Traffic

During two decades, there have been a revolution in the field of digital communication and internet access. Today, it became possible for users to access the internet while on the move through an infrastructure of high-speed mobile broadband networks. Technologies such as LTE and 5G became essential. Mobile broadband networks allow mobility; connection reliability drops during movement. Thus, some failure intolerant processes, such as system updates, necessitates the utilization of a reliable connection. This paper introduces a model that predicts whether the user is mobile or stationery. This is done based on the traffic patterns at the server-side. Distinct network technologies entails distinct nature of traffic patterns. In this paper, machine learning is utilized at the server-side to allow differentiating between data transmitted by a stationary user and data transmitted by a mobile user at the server-side. Supervised training is utilized to train the model. Then, the model was tested and prediction accuracy of this model was 92.6 percent. Finally, the proposed system is a novel work and the first of its kind since it is the first to attempt to predict mobile network user's mobility at the server-side by utilizing packets' arrival patterns. The proposed system can be applied at mobile apps and allow them to collect data about the apps users mobility while using this service without needing to access the GPS. Also, it can be used network management and public safety.