LKSVC: A novel VANET caching method by integrating location-based K-means clustering into spiking neural network

With the increasing demand for on-road communication, Mobile Ad-hoc Networks (MANET) have evolved into Vehicular Ad-hoc Networks (VANET) to support data retrieval during road trips. However, the highly dynamic network topologies pose significant challenges to conventional VANET communication methods. Furthermore, the deep learning model is too large to be deployed in vehicles, leading to Quality of Service (QoS) degradation and high link loads. To address these drawbacks, this paper proposes a novel VANET caching method by integrating Location-based K-means clustering into spiking neural networks (LKSVC). The proposed LKSVC was the first time applying Spiking Neural Networks in the VANET caching and incorporated with the Kmeans clustering. Empirical simulations show that LKSVC significantly reduces Link Load by least 23%, improves Local Satisfaction Ratio by least 10%, and One-hop Hit Ratio by least 50% while reducing Data Retrieval Time compared to conventional methods.