Project Introduction
This proejct aims to propel the next generation of intelligent networked systems by combining privacy-preserving AIoT data collection with advanced AI-driven network optimization in O-RAN architectures. The initiative integrates local differential privacy (LDP) techniques to secure and anonymize IoT data at its source, addressing critical data privacy and security challenges that arise from the massive proliferation of AIoT devices across smart cities, healthcare, and industrial ecosystems. Simultaneously, it leverages foundation models and federated learning approaches at the RAN Intelligent Controller (RIC) to improve network decision-making efficiency, especially under complex, non-independent and identically distributed (non-iid) data scenarios characteristic of distributed RAN environments.
Project Targets and Goals
Develop Privacy-Preserving AIoT Data Collection:
Create mechanisms based on Local Differential Privacy that enable IoT device-level data anonymization and secure encryption prior to cloud or network transmission.
Design Privacy-Aware AI Modeling Techniques:
Build AIoT analysis and intelligence tools that leverage privacy-protected data to provide accurate decision support without compromising individual or commercial confidentiality.
Advance RAN Intelligent Controller Capabilities:
Integrate federated learning and foundation models within the O-RAN RIC to address the challenges of non-iid data distribution and enhance real-time network optimization and resource allocation.
Establish Industry-Standard Privacy and AIoT Technology:
Define practical standards and frameworks suitable for industrial adoption, enabling secure, privacy-compliant AIoT services at scale.
Drive AIoT 6G Network Innovation:
Support forward-looking 6G network technologies by embedding privacy and AI intelligence into the network fabric, enhancing user trust, industrial competitiveness, and societal benefits.