Realtime Predictive Patrolling and Routing with Mobility and Emergency Calls Data

Abstract

A well-planned patrol route plays a crucial role in increasing public security. Most of the existing studies designed the patrol route in a static manner. Situations when rerouting of patrol path are required due to the emergencies, e.g., an accident or ongoing homicide, are not considered. In this paper, we formulate the crime patrol routing problem jointly with dynamic crime event prediction, utilising crowdsourced check-in and real-time emergency call data. The extensive experiment on real-world datasets verifies the effectiveness of the proposed dynamic crime patrol route using different evaluation metrics.