A M2M network-based realistic mobile user movement prediction in emergencies

In the last few decades, the number of emergencies has increased worldwide. Often, in these emergencies, both wireless and wired communication infrastructures could be damaged partially or completely. Hence, mobile devices, being ubiquitously present and closely attached, usually become a dominant way of communication. In general, a prior knowledge of user movements in mobile environments can improve both network and application-level performances. However, a realistic movement prediction for mobile users in emergencies is especially crucial since the emergency-affected network is often resource constrained. Although an extensive research has already been conducted on user movements, till now, influence of an emergency on user behavior as well as on networks, e.g., geographical constrains, and thus user movements has not been investigated. As a result, this work proposes a realistic movement prediction framework for mobile users in emergencies which considers both user´s behavioral changes and the geographical constraints by implementing a machine-to-machine (M2M) network of the mobile devices. Simulation results exhibit that the prediction accuracy reaches up to 96% and thus the predicted movements closely follow the real movements.