DPU-Enhanced Network Security Situation Awareness Model for New Power Systems

As the development of new power systems progresses, the edge side exhibits emerging challenges such as diversified new network services, rapid extensions of access and local networks, and a dramatic increase in the number of sensing terminals. These conditions pose extremely severe challenges to security protection. Traditional network security situation awareness models are predominantly software-based, neglecting the intensive I/O overhead in big data network environments. Such approaches fail to keep pace with the increasing network traffic in new power systems and do not fully meet the demands for high concurrency and low latency. Additionally, conventional methods lack effective security isolation measures, rendering systems vulnerable to malicious attacks and security breaches, potentially triggering a series of stability issues within the system. In response, this paper proposes an integrated hardware-software co-design architecture for traffic extraction based on Data Processing Unit (DPU). By offloading some regex matching tasks to the hardware, this architecture significantly enhances the system's overall processing capability and throughput. Furthermore, this study introduces a multi-source fusion-based network security situation awareness model tailored for new power system environments, markedly improving the accurate identification of attack behaviors and providing robust support for network security.