Robust Wireless Distributed Learning Empowered by Thz Communications Data for Internet of Unmanned Vehicles Agents: Efficient Cluster Driving Decision-Making

Li, Zihong, Wu, Jun ORCID: https://orcid.org/0000-0003-2483-6980, Bashir, Ali Kashif ORCID: https://orcid.org/0000-0003-2601-9327 and Li, Xingwang ORCID: https://orcid.org/0000-0002-0907-6517 (2025) Robust Wireless Distributed Learning Empowered by Thz Communications Data for Internet of Unmanned Vehicles Agents: Efficient Cluster Driving Decision-Making. IEEE Internet of Things Journal. ISSN 2372-2541

Preview

Accepted Version Available under License Creative Commons Attribution. Download (542kB) | Preview

Abstract

With the rapid development of the Internet of Unmanned Vehicles Agents (IUVA), efficient and secure communication has become a key requirement. However, unstable wireless channel conditions pose several challenges to existing Wireless Distributed Learning (WDL) in the IUVA environment. First, the parameter transmission of WDL in the IUVA will be interfered by dynamic changes in vehicle position, which will affect the training accuracy and the loss of the learning model. Second, increased communication overhead due to the large amount of data generated by vehicles sensors, and third is the complexity of making real-time driving decisions with diverse vehicle data. This paper presents an innovative WDL framework based on Terahertz (Thz) communication technology, addressing communication and data processing challenges in the IUVA environment. Our framework designs a Thz communication encoding method, treating each vehicle as a local model node participating in the WDL process. First, we established a IUVA cluster based on Thz communication, addressing the current issues of high latency and low efficiency in IUVA communications. Second, we designed a WDL framework where vehicles within the IUVA act as distributed learning participants, reducing communication overhead in IUVA wireless communication. Finally, our proposed wireless distributed driving decision-Making model leverages the physical parameters of participating vehicles to derive collective driving decisions for the IUVA cluster, enhancing the accuracy of IUVA driving decisions. Overall, the framework proposed in this paper provides a new approach for achieving efficient and secure IUVA communication and contributes significantly to intelligent unmanned decision-making in IUVA.