演讲摘要：With the continuous rise in the number of vehicles in urban areas, Vehicular Ad-hoc Networks (VANETs) have emerged as a pivotal component in the development of future Intelligent Transportation Systems (ITS). In addition to safety-related applications and Internet access, VANETs also provide a diverse range of user applications to serve the needs of both drivers and passengers. One key aspect of VANETs is the adoption of clustering, which facilitates efficient data exchange among geographically proximate vehicle nodes, with Cluster Heads (CHs) facilitating inter-cluster data communication. Despite the advancements in clustering-based VANET communication algorithms, their complexity and performance have been impeded by the rapid mobility of vehicles and the variability of parameters, including location, speed, and acceleration.
This dissertation proposes a novel Hybrid Clustering Algorithm based on an adaptive self-learning method, comprising three distinct steps. Firstly, the algorithm transforms the vehicle nodes into a linear distribution, utilizing Roadside Units (RSUs) as static cluster heads. Next, the entire network area is partitioned into multiple regions, and a clustering algorithm based on k-means is employed to predict the effective centroid/weight value of each vehicle. Lastly, the cluster head selection is determined by utilizing the results obtained from the neural fuzzy training sample, assigning cluster head percentages to each region to effectively alleviate end-to-end delay and congestion.
The effectiveness and practical significance of the proposed algorithm are evaluated through extensive simulations. The results demonstrate that our algorithm exhibits superior efficiency and applicability compared to existing methods. This research contributes to the enhancement of VANET clustering techniques and fosters the advancement of Intelligent Transportation Systems to accommodate the growing demands of urban vehicular networks.
讲者简介：王伯元，男，天津工业大学2021级工业工程本科生。主要研究方向为边缘计算，强化学习。
首届CCF中国智能汽车学术年会（CIVS 2023）
本期嘉宾
京公网安备11010802032778号