Accurate traffic flow prediction plays a crucial role in designing Ad hoc vehicular mobile networks in modern Internet of things (IoT) based intelligent systems. Several deep learning techniques have been deployed to predict traffic conditions to make vehicular communication more reliable. However, not all these approaches deal with complex road networks and spatial temporal dependencies of traffic data. In this paper, we analyze this problem using long short-term memory (LSTM), gated recurrent unit (GRU) and hybrid CNN-LSTM models. We trained our models using actual traffic flow data provided by the California Department of Transportation (Caltrans) over a 6 month duration and showed that our deep learning models outperform the traditional linear regression method. Moreover, an architectural study of deep learning models is carried out for the traffic flow prediction problem. The performance of these models is evaluated using MSE and MAE metrics. It is observed that the GRU model is the best to handle the complex vehicular traffic mechanisms. Also, that a complex hybrid model like CNN-LSTM does not always outperform the much simpler architectures such as LSTM and GRU.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Deep Learning Based Traffic Flow Prediction for Autonomous Vehicular Mobile Networks


    Contributors:


    Publication date :

    2021-09-01


    Size :

    559932 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Deep Learning Based Autonomous Driving in Vehicular Networks

    Su, Zhou / Hui, Yilong / Luan, Tom H. et al. | Springer Verlag | 2020


    Traffic prediction with deep learning for vehicular communication network

    Wang, Jin / Liu, Peixin | British Library Conference Proceedings | 2022


    Deep Spatio-Temporal Residual Networks for Connected Urban Vehicular Traffic Prediction

    Wu, Hanlin / Zhou, Haibo / Zhao, Jiwei et al. | IEEE | 2020


    Deep Reinforcement Learning for Traffic Light Control in Vehicular Networks

    Liang, Xiaoyuan / Du, Xunsheng / Wang, Guiling et al. | ArXiv | 2018

    Free access