Prediction and avoidance of real-time traffic congestion system for Indian metropolitan cities

Abstract

The expansion of sensors and data is the catalyst for traffic simulators to help monitor and manage transportation systems in real-time to improve key performance metrics and overall efficiency. The number of accidents has rapidly increased in Indian perspective over the past few decades. Owing the accidents, a large number of losses occur in form of infrastructure, resources and people. In this paper, we have proposed an architectural framework for Indian urban transport system to gather, store, data mining and getting information to avoid the traffic accidents, congestions and predicting the alternative routes in real time. A combination of machine learning and data mining techniques are used to implement the proposed system. The implemented system will help in reduction of traffic accidents and suggesting alternative routes to avoid the traffic congestion for an Indian urban transport network operation in real-time. Real-time data is fed into a traffic simulation, which generates future states of the road network and alternative routes to avoid congestion. Because the results of the system need to be both faster than real-time and accurate, the acceleration of the simulation execution and the accuracy of the prediction models are critical.