6G SDVN: UAV-Assisted VANETs for AI-Enabled Controller Selection and Heterogeneous Multi-Band Connectivity

초록

Road sparsity is a major cause of routing instability, leading to intermittent connectivity and increased packet loss. To enhance stability in 6G software-defined vehicular networks (SDVNs), we integrate unmanned aerial vehicles (UAVs) into a dynamic vehicular environment. We propose a novel ground-to-air communication protocol that integrates Dedicated Short Range Communication (DSRC), millimeter-wave (mmWave), and terahertz (THz) bands to address path stability, load balancing among controllers, and technology switching to accommodate diverse vehicular demands. Roadside Units (RSUs) and UAVs act as local controllers (LCs), with RSUs communicating via LTE and UAVs via DSRC. The optimal controller (OC) is selected using supervised learning algorithms, Support Vector Machine (SVM), and Random Forest (RF), to ensure balanced load distribution and to manage stable routing across hybrid wireless technologies. The proposed scheme evaluates multiple UAV roles, including LC, relay, and destination, under different communication interfaces. Simulation results in a highway scenario demonstrate that the hybrid approach outperforms the reference DSRC, mmWave, and THz schemes in terms of packet delivery ratio, end-to-end delay, and routing overhead. Specifically, SVM-based OC selection achieved a maximum packet delivery ratio of 97.94%, with an end-to-end delay as low as 0.089 ms and minimal routing overhead, validating its effectiveness in mitigating road sparsity, improving connectivity, and enhancing network performance.

키워드