Prediction of Traffic Flow by Sequencing Spatial–Temporal Traffic Dependency on Highways

Abstract

Traffic flow on highways is a dynamic process in which characteristics of road segments vary in time and space. Traffic congestion is the adverse effect caused dby an increase in travel time which is unprecedented by time and space. Temporal and spatial information of traffic flow is an integral component in the assessment of highway traffic flow. The spatial–temporal traffic flow dependency on highways can be well assessed when temporal traffic information in preceding time instances is sequenced. Thus, the SCAE-LSTM network is proposed considering time and space. This study investigates the estimation of traffic flow on highways based on spatial and temporal traffic sequences. Sequencing highway traffic information has motivated the authors to propose the method. The performance of the method is experimented on State Highways SH 49 and SH 49-A of Chennai Metropolitan City, Tamil Nadu, India. The computational complexity of the method is analyzed empirically. The significant outcome of the proposed method is reported in the experimental study. The traffic flow estimated using the proposed method has shown reduced complexity compared to other baseline methods. Finally, research directions to work in future are presented towards the end.