Many metropolises seek to relieve traffic congestions and reduce vehicle accidents by implementing Intelligent Traffic Information Systems. These systems manage continuous communication between vehicles, various roadside IoT devices and a central server in real time for traffic control and vehicle navigations. Short response time is critical to the success of these time-sensitive systems. For a small area, a system with centralized server architecture may just work fine. For a larger area with more IoT devices and traffic, however, the system may experience excessive response time as a result of increased network distance and constrained server processing capacity. We propose a decentralized server system to properly manage and reduce service response time. We have also developed a binary nonlinear constrained programming model with Genetic Algorithm for a heuristic solution.