As the penetration rate of connected and autonomous vehicles (CAVs) increases on highways, their role as network nodes for communication tasks raises significant challenges. Spatial heterogeneity in node distribution creates density discrepancies that substantially impact network lifetime and stability, thereby constraining optimal deployment strategies for road side units (RSUs). This study introduces an enhanced low-energy adaptive clustering hierarchy (LEACH) clustering algorithm tailored for vehicular networks. By identifying dense and sparse regions through dynamic clustering, the algorithm categorizes node functions according to regional characteristics to balance energy consumption and improve network connectivity. MATLAB simulations validate the algorithm’s performance under non-uniform vehicle distributions.The research further analyzes how vehicle node distribution patterns and CAV penetration rates affect optimal RSU deployment intervals. Key findings reveal that with consistent RSU-vehicle communication ranges: At a CAV traffic density of 0.01 and relative spatial density of 0.5 (uniform distribution), RSUs should be deployed at 641 m intervals. At a relative density of 0.9 (concentrated distribution), deployment intervals can expand to 1,887 m while maintaining high network connectivity. This adaptive strategy reduces communication blind spots by 32%, lowers deployment costs by 18%, and enhances vehicle-road coordination efficiency and traffic safety. The results provide critical technical support for intelligent vehicle-road collaboration systems.