Urban traffic signal control network automatic partitioning using laplacian eigenvectors

Many existing traffic signal control systems are of a hierarchical structure employing the concept of subsystems or subzone signal control. Therefore, network partitioning becomes an important task of such an urban traffic signal control system. However, urban traffic signal control network partitioning is a multiobjective and multiconstraint problem, which has been shown to be a NP-hard problem. This paper presents a partitioning method using the spectral methodology according to the correlation degree of each link. Two measures are adopted in this problem: modularity and average cut weight. The developed method has been tested on two networks, including a computer generated network and real world road network. The results show that the spectral bisection is a reasonable network partitioning method to support urban traffic signal control.