AI-Driven Collaborative Security Protection for Cloud-Edge Computing Ecosystems: Architecture Design and Performance Evaluation

Abstract

With the rapid expansion of cloud-edge computing ecosystems, traditional passive security defense mechanisms have become inadequate in coping with the increasingly complex and dynamic threat landscape, such as adaptive malware, targeted ransomware, and distributed denial-of-service (DDoS) attacks evolving with edge intelligence. Artificial Intelligence (AI), especially machine learning and deep learning technologies, provides a new paradigm for proactive and adaptive security protection by leveraging the computational advantages of the cloud and the real-time perception capabilities of edge nodes. This study proposes an AI-driven collaborative security protection architecture (AICSPA) for cloud-edge ecosystems, which realizes seamless collaboration between cloud-side global threat decision-making and edge-side real-time threat detection. The architecture consists of four core modules: edge-side lightweight AI detection engine, cloud-side intelligent threat analysis center, secure collaborative communication channel, and dynamic policy optimization module. Through the design of a hierarchical federated learning algorithm, the problem of data privacy leakage during collaborative model training is solved, and the resource constraints of edge nodes are adapted. Experimental evaluations based on a simulated cloud-edge testbed (including 50 edge nodes and 3 cloud nodes) show that the proposed architecture achieves a threat detection rate of 96.3% for unknown attacks, which is 18.7% and 23.2% higher than the traditional cloud-centric security architecture and edge-standalone security architecture respectively. Meanwhile, the average detection latency is reduced to 12.5ms, meeting the real-time requirement of edge applications. The research results demonstrate that the AI-driven collaborative security architecture can effectively improve the security resilience of cloud-edge ecosystems, providing a feasible technical solution for the security protection of emerging cloud-edge integrated applications.

Keywords

Cloud-edge computing; AI-driven security; Collaborative protection; Federated learning; Lightweight detection; Security architecture

pdf