Enhancing AI-Based Intrusion Detection with GPU Acceleration for Next-Generation Networks
Within the SMARTY project, a Graph Neural Network (GNN)-based solution for real-time malicious traffic detection has been enhanced to provide faster and more efficient AI-driven cybersecurity capabilities.
The activity addresses a key limitation of the previous approach: the graph construction phase, which represented the main bottleneck in the overall prediction process. By redesigning the graph generation mechanism and moving this process from the CPU to the GPU, the solution enables parallel processing and significantly reduces detection latency.
The upgraded GPU-based system achieves the same level of intrusion detection accuracy while reducing prediction time compared to the previous CPU-based approach. Additional optimisations, including more efficient graph representations, further improve memory usage and scalability.
This advancement strengthens the capability of AI-based security solutions to operate in low-latency environments, supporting the deployment of more responsive and secure programmable infrastructures for future 5G and 6G networks.
Follow SMARTY for updates:
LinkedIn: https://www.linkedin.com/company/smartyeu/?viewAsMember=true
Website: https://www.smarty-project.eu