A Hardware-Accelerated Intrusion Prevention System for Attack Mitigation Using DPUs

The increasing frequency of network attacks makes it essential to have an effective mechanism for real-time traffic analysis and attack response. This paper presents a hardware-accelerated framework for detecting and mitigating Distributed Denial-of-Service (DDoS) attacks. Implemented on a Data Processing Unit (DPU), the framework leverages Data Plane Development Kit (DPDK) libraries for high-speed packet processing, hardware-accelerated regular expression (RegEx) for Deep Packet Inspection (DPI), and Suricata for attack prevention. Experimental results demonstrate the framework's ability to mitigate up to 100% of attack traffic with only 50%-60% CPU utilization, significantly outperforming non-hardware-accelerated DPI solutions limited to 20 Gbps throughput with full CPU usage. Our work shows the potential of hardware acceleration in achieving scalable and efficient network security.