The MapReduce-based approach to improve vehicle controls on big traffic events

Recently, with the integration of Hadoop framework in the resolution of many related transportation problems, it has been show that the concept of parallel and distributed computing plays an important role in the management of large-scale traffic data. The advantages of Hadoop components allow intensive computing with huge volume of streaming data which cannot be processed by traditional Intelligent Transportation System (ITS). In this paper, we present a MapReduce-based approach to analyze big stream data in order to detect abnormal traffic events. Our approach consists in partitioning the big log file of traffic events into a set of sub-events then analyzing traffic events on each event block in parallel way. Experimental tests reveal that the proposed approach is very inspiring and achieves significant gain in term of calculation time.