Innovative Segmentation-Based Routing in VANETs: Leveraging the Advantages of TD-ISAC

Reliable, ultra-low-latency communication in Vehicular Ad-hoc Networks (VANETs) is critical for autonomous vehicle safety, yet existing routing protocols struggle with high overhead and frequent link disruptions in dynamic environments. A key limitation is their failure to leverage the rich sensor data already available in modern vehicles for network-level coordination. This paper introduces a novel segmentation-based routing framework that directly addresses these challenges by transforming in-vehicle sensor data into a shared, high-resolution network topology. Our core contribution is a dynamic road occupancy matrix, created by dividing roadways into velocity-adaptive, single-vehicle segments, which provides all nodes with a unified and near real-time view of the traffic environment. This matrix is disseminated efficiently using a Time Division Integrated Sensing and Communication (TD-ISAC) framework inspired by 5G Sidelink. We propose a hybrid Lagrangian Relaxation and Branch-and-Bound algorithm to dynamically optimize the TD-ISAC frame, balancing the trade-off between sensing accuracy and communication throughput. To further enhance topological precision and mitigate vehicle mobility effects, the framework integrates a Kalman Filter and Transformer model for predictive position estimation and employs a reserve relay mechanism to ensure multi-hop link stability. Comprehensive simulations demonstrate that our framework significantly outperforms state-of-the-art protocols. Notably, it reduces end-to-end delay by up to 68% compared to ZRP and 41% against the TD-ISAC-based Starling Flocks, while simultaneously decreasing routing overhead. This dual improvement enables a more efficient and robust network for safety-critical applications.