Intelligent traffic management is an important issue for smart cities. City councils try to implement the newest techniques and performant technologies in order to avoid traffic congestion, to optimize the use of traffic lights, to efficiently use car parking, etc. To find the best solution to this problem, Birmingham City Council decided to allow open-source predictive traffic forecasting by making the real-time datasets available. This paper proposes a multi-agent system (MAS) approach for intelligent urban traffic management in Birmingham using forecasting and classification techniques. The designed agents have the following tasks: forecast the occupancy rates for traffic flow, road junctions and car parking; classify the faults; control and monitor the entire process. The experimental results show that k-nearest neighbor forecasts with high accuracy rates for the traffic data and decision trees build the most accurate model for classifying the faults for their detection and repair in the shortest possible time. The whole learning process is coordinated by a monitoring agent in order to automate Birmingham city’s traffic management.
Multi-Agent System for Intelligent Urban Traffic Management Using Wireless Sensor Networks Data
