Information Theory Based Traffic Pattern Detection in Mobile Ad Hoc Network of Vehicles

Prior knowledge of road traffic conditions on the freeways is of prime importance for motorists. With recent developments in technology it is possible for the vehicles to be equipped with communication and GPS systems. The equipped vehicles on the road can act as nodes to form an ad hoc network. These nodes can collect information regarding traffic conditions such as position, speed and direction from other participating nodes. Depending upon the number of participating nodes this collected information can provide useful information of driving conditions to the node collecting this information. With proper analysis this information can be used in detecting and or predicting traffic jam conditions on the freeways. In this paper the traffic information gathered by a node in an ad hoc network is viewed as a snapshot in time of the current traffic condition on the road segment. This snapshot is considered as a pattern in time of the current traffic conditions. The pattern is analyzed using pattern recognition techniques. A weight of evidence based classification algorithm is presented to identify different road traffic conditions. The developed algorithm is tested using data generated by microscopic modelling of traffic flow for simulation of vehicle or node mobility in ad hoc networks. Test results are presented under assumption of different levels of vehicles equipped with communication capability.