With the rapid adoption of autonomous vehicles, research on Connected and Autonomous Vehicles (CAVs) has intensified, exposing significant security challenges—particularly in the realm of anomaly detection, where even a single malicious attack (e.g., location spoofing) can have catastrophic consequences. While energy efficiency is paramount in these applications, most recent studies have focused on performance improvements through increasingly complex model designs. Although Spiking Neural Networks (SNNs) have attracted attention for their energy-efficient properties, research on their application to anomaly detection in vehicular networks is almost non-existent. In this work, we address this gap by proposing an SNN-based anomaly detection framework designed to overcome energy constraints in vehicular networks while maintaining high detection performance. We introduce a novel RSNN model that enhances the capabilities of conventional SNNs and incorporate advanced feature engineering techniques to further optimize system performance. Our experimental evaluation demonstrates robust performance across a variety of attack scenarios, achieving high precision, accuracy, recall, and F1 scores, with notable improvements observed under location spoofing and random offset attack conditions.