Estimation of Traffic Congestion States using GPS Data

Abstract

Road traffic congestion is major problem in urban area of both developing and developed countries. In order to reduce this problem, traffic congestion states of road networks are estimated so that congested road can be avoided. In this system, we estimate the current traffic congestion states of user’s desired source and destination and present the estimated results in Google Map. To get the traffic data we use GPS data from mobile phones on vehicles but this GPS points can have ambiguity. The decision support topological based map matching algorithm that can solve the ambiguity of GPS data is used to identify which vehicles are on which road by matching these GPS data with road networks. The historical traffic condition data of each road network on each time using the pre-collected data are utilized in this research. We use Hidden Markov model (HMM) for estimating the traffic condition states of these road network using historical traffic data. These estimated traffic probabilities states are presented by coloring (traffic jam for red, traffic heavy for blue and traffic normal for green) users’ desired source and destination road segments on Google Map. We evaluate our estimating system using dataset generated by collect data from mobile phone-equipped vehicles over a period of 4 months in Yangon.