Computer Sciences and Information Technology
Title : | Privacy-aware Federated Learning based Security Solutions for Beyond 5G Networks |
Area of research : | Computer Sciences and Information Technology |
Focus area : | Cybersecurity, Machine Learning |
Principal Investigator : | Dr. Vignesh Sivaraman, Indian Institute Of Technology (BHU), Uttar Pradesh |
Timeline Start Year : | 2023 |
Timeline End Year : | 2025 |
Contact info : | svignesh16.rr@gmail.com |
Equipments : | Network router
Workstation |
Details
Executive Summary : | Beyond 5G networks promise high performance, reliability, and mass connectivity while consuming minimal latency and energy. This will enable applications like autonomous driving, real-time remote surgery, industry 4.0, and smart-grid 2.0. However, the increasing number of connected things in B5G networks poses new security risks. To address these challenges, designing and developing defensive countermeasures is crucial. Artificial Intelligence (AI) can be leveraged to create self-learning and self-correcting network security systems that can detect hidden patterns from a large volume of network traffic data. This knowledge can help detect anomalies and mitigate security threats. B5G requires a massive amount of user data, which poses threats to user privacy. Federated learning is a potential machine learning technique that provides distributed user-privacy-aware learning models. It trains a machine learning model on multiple local nodes holding local data without data exchange, and the parameters of the trained models are shared between local nodes to generate a global model. This proposal focuses on the specific case of Internet Service Providers (ISPs) and aims to develop intelligent network security systems using federated learning among various ISPs. The primary objective is to deploy federated learning models among ISPs' own nodes to generate an intra-ISP model, and exchange this model with other ISPs to generate a global one. |
Total Budget (INR): | 22,47,340 |
Organizations involved