Efficient traffic control system now a day's getting importance from different perspectives like number of the vehicle density, time management and less proportionate road lane increase. Different factors engross under various researchers, industries, improving the road transportation system. Major obstacle for vehicle on the road is the junction and traffic clearance at mixed scenarios at the junctions. In this paper, we have proposed a Vehicular Ad-hoc Network (VANET) architecture for real time traffic information exchange from one traffic signal to another traffic signal via vehicle. A probabilistic function also involved to capture future prediction of traffic count at one step ahead of the current traffic signal. As per the traffic information received time to time a modified Round Robin algorithm is introduced with different and segmented quantum time. In our approach, neither complex computations are required, like image processing nor costly infrastructure is required, like GPS systems and centralized management. In this way we can create a smart, decentralized vehicle transportation system for a mega city. © 2017 IEEE.