The increasing number of vehicles has brought great pressure to urban traffic. The development of intelligent transportation provides new ideas for alleviating traffic problems. If drivers can get the road condition of certain road sections, it will greatly improve the driving efficiency and optimize the whole traffic network. Road condition recognition often needs a large amount of road condition data collected by multiple vehicles and aggregated by crowdsensing. However, the data collected by vehicles may carry privacy information, privacy leakage problem will affect the incentive of car owners to complete the crowdsensing tasks. In order to protect the privacy and security of the collected data, the aggregation of raw data could be replaced by the aggregation of road condition recognition models. On this basis, we propose a new secure crowdsensing algorithm based on federated learning and blockchain. Specifically, the task publisher deploys the crowdsensing tasks on the blockchain, and the leader of each team in the crowds decides whether to accept the task. When a team leader accepts the task, the team members train the road state recognition model, and the team leader aggregates the models provided by the team members through federated learning. In addition, each member adds differential privacy noise to its own model to further protect privacy. The team rewards each member based on the accuracy of the traffic recognition model provided by the member. Experimental results show that the proposed strategy performs well in road condition recognition accuracy for intelligent transportation.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Privacy-Preserved Mobile Crowdsensing for Intelligent Transportation Systems


    Weitere Titelangaben:

    Internet of Things: Tech., Communicat., Computing


    Beteiligte:
    Garg, Sahil (Herausgeber:in) / Aujla, Gagangeet Singh (Herausgeber:in) / Kaur, Kuljeet (Herausgeber:in) / Hassan Ahmed Shah, Syed (Herausgeber:in) / Miao, Qinyang (Autor:in) / Lin, Hui (Autor:in) / Hu, Jia (Autor:in) / Wang, Xiaoding (Autor:in)


    Erscheinungsdatum :

    2021-12-15


    Format / Umfang :

    14 pages





    Medientyp :

    Aufsatz/Kapitel (Buch)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch




    SPEED: A Deep Learning Assisted Privacy-Preserved Framework for Intelligent Transportation Systems

    Usman, Muhammad / Jan, Mian Ahmad / Jolfaei, Alireza | IEEE | 2021


    Mobile crowdsensing with mobile agents

    Leppänen, T. (Teemu) / Lacasia, J. Á. (José Álvarez) / Tobe, Y. (Yoshito) et al. | BASE | 2017

    Freier Zugriff