Distributed Classification of Urban Congestion Using VANET

Abstract

Vehicular ad hoc networks (VANETs) can efficiently detect traffic congestion, but detection is not enough, because congestion can be further classified as recurrent and nonrecurrent congestion (NRC). In particular, NRC in an urban network is mainly caused by incidents, work zones, special events, and adverse weather. We propose a framework for the real-time distributed classification of congestion into its components on a heterogeneous urban road network using VANET. We present models built on an understanding of the spatial and temporal causality measures and trained on synthetic data extended from a real case study of Cologne. Our performance evaluation shows a predictive accuracy of 87.63% for the deterministic classification tree, 88.83% for the naïve Bayesian classifier, 89.51% for random forest, and 89.17% for the boosting technique. This framework can assist transportation agencies in reducing urban congestion by developing effective congestion mitigation strategies knowing the root causes of congestion.

Cooperative Evaluation of the Cause of Urban Traffic Congestion via Connected Vehicles

“…Each vehicle is equipped with a method to detect excessive congestion and a classification algorithm able to attribute a possible cause to it, as in [5]. We then propose that each vehicle represent its uncertainty about the cause of congestion in a vector of probabilities associated to each of the possible causes of:…”

“…The VP is an improvement of the Back-Propagation algorithm presented in [5]. In BP vehicles wait a certain duration and transfer messages only if they have total knowledge about the cause.…”

“…We consider a realistic map configuration of the city of Cologne in the evaluation of our methods. Compared to the Back-Propagation algorithm (BP) proposed in the literature [5], the proposed techniques enhance the estimation accuracy by 48% for the VP, 58% for the BF, 71% for the DAT, and 70% for the adapted data association technique (β-DAT). The methods also enhance the detection time by 10.3% for VP, 9.40% for BF, 9.45 for DAT and 7.09% for β-DAT.…”

Prediction of Traffic Flow via Connected Vehicles

“…To do so, we implement on board of each vehicle the algorithm for the detection of congestion via connected vehicles presented in [25]. Also, we implement the algorithm in [26] that permits vehicles classify the cause of the detected congestion. The results of the local real-time monitoring done by the CV is disseminated reactively to others on the segment.…”

“…Procedures of lines 9-12 represent the analysis phase. We compute variables as per [26]. If the observed travel time is above a threshold, the vehicle estimates the cause of congestion by creating the feature vector and inferring with the classifier the cause.…”