Enterprises

This project focuses on developing an Endogenous Security Defense System tailored for heterogeneous network environments involving 5G and Low Earth Orbit (LEO) satellites. By leveraging the privacy-preserving features of Federated Learning (FL), we aim to overcome the challenges of collaborative defense across heterogeneous nodes, such as base stations, ground stations, and satellite terminals, where direct data sharing is restricted.
Research and Development Tasks
•    Deployment of Strategy-Driven Federated Learning (FL) Architectures: Design asynchronous parameter aggregation strategies to enhance convergence efficiency, specifically addressing the constraints of satellite link bandwidth and node heterogeneity.
•    Cybersecurity Threat Detection: Utilize FL to train cross-node Intrusion Detection Models (IDS) capable of identifying Distributed Denial of Service (DDoS) or model poisoning attacks without exchanging raw data.
•    Security Validation and Implementation: Implement lightweight encryption technologies on edge computing devices to align with international cybersecurity standards and narrow the gap between R&D and commercial deployment.