Police Patrol Optimization With Security Level Functions

Public security is a key concern around the world. Efficient patrol strategy can help to increase the effectiveness of police patrolling and improve public security. In this paper, we propose a new security measure characterized by the security level function. Furthermore, we formulate the patrol process as a Markov decision process and propose a cross-entropy-based ε-optimal patrol strategy to deal with the curse of dimensionality. We also design a randomized strategy for adding uncertainty into patrolling. Numerical studies demonstrate that the proposed patrol strategy achieves up to 70% and 95% performance improvement over the previously proposed Hamilton algorithm and Q-learning algorithm, respectively.