Prediction System for Traffic Congestion using GPS Data on Hadoop Cloud Storage

Abstract

The high values of vehicles, the inadequate infrastructure cause traffic congestion. Congested roads can be avoided by determining the travel-time for a particular road ahead of time. Traffic prediction and travel time estimation has traditionally relied on expensive measuring methods such as loop detectors, vehicle identification devices. In this paper, we use mobile GPS equipments on vehicles to gather data for cheaper and real time travel-time estimation. We use this data to develop the prediction system for traffic congestion in order to improve the quality and safety of vehicle movement and for minimization the time and costs when vehicles are moved at the specified routes. We collect the GPS data and classify them with K-Means algorithm. Moreover, framework based on Markov model is used to predict traffic and Hadoop is used as cloud storage and platform, to accelerate the processing computing speed and allow handling of large-scale data.