An M/M/s/c Traffic Congestion Queuing Model for Toll Plazas Employed with Artificial Neural Network (ANN)

In recent time traffic congestion in the road system is a very big problem for vehicles. It is seen that many toll plazas are highly loaded with traffic and hence an appropriate analysis under queuing theory is required. At the toll plazas vehicles wait before paying the toll which causes long queues and unnecessary delays for which we develop a mathematical model for the management of traffic intensity. In this study, we analyze the system performance in the context of the parameters: average queue length, average waiting time, and total time spent in the system, and we discuss also how arrival rate, service rate, and number of servers affect queue length and waiting time. Moreover, we evaluate the lane-jockeying probability of vehicles in this study by adopting the M/M/c/c queuing model. The novelty of the study is to verify the lane-jockeying behaviors existing in the traffic system by using the ‘Levenberg-Marquardt Backpropagation’ training algorithm in ANN.