A Bi-level Programming Model for Network Traffic Surveillance of Optimal Camera Location

Cameras enable synthesis of disaggregated vehicle information from multiple locations. This paper addresses the problem of how to determine the optimal locations of cameras for best traffic surveillance in urban road networks. All users are divided into multiple classes according to their own acceptance degree of camera monitoring (ADOCM). A bi-level programming model for camera location problem (CLP) is introduced. The objective of upper level problem is to maximize the total observed traffic flow from the viewpoint of network planners. The lower level problem is to depict the logit-based stochastic route choice behavior of users. The method of successive averages (MSA) and Frank Wolfe algorithm are employed to solve the problem. Numerical examples show the managerial insight that the behavior against camera monitoring has brought considerable loss to network traffic surveillance.