A Secure Trust-based Approach for Detecting and Preventing Network Attacks in 5G Environments

Abstract

The rapid roll-out of 5G communication networks, 5G Users can experience high-speed connections, super low latency and wide-open connectivity with Internet of Things. The 5G infrastructures are based on the distributed architecture, which makes them vulnerable to advanced cyber-attacks such as Distributed Denial-of-Service, Sybil attack, Spoofing attack, Insider attack and Data manipulation attack. The prevailing security paradigm in recent years has been focusing on either intrusion detection or privacy protection, but little effort has been made to achieve a combined and integrated security strategy to construct a proper trust and base the decisions that are taken. In order to overcome the above challenge, the present work introduces the TrustChain-5G system, a safe and secure trust based system, that combines the advantages of Federated Learning, Adaptive Trust Evaluation, Blockchain and Machine Learning for detecting attacks. The trusted behavior economy model constantly monitors node communication patterns, past interactions, security compliance index to identify suspicious nodes in a system. Federated learning can be used to create a collaborative model, keeping privacy intact, while blockchain can be used to store security events (tamper-proof), and trust records. The simulated 5G environment for the evaluation of the framework consisted of 100 network nodes in a mixture of CICIDS2017, CICDDoS2019 and UNSW-NB15 datasets. This is experimentally demonstrated to provide 96.2% accuracy, 95.4% precision, 94.9% recall and a minimum detection delay in various attack scenarios. The results validate the success of the proposed architecture in increasing the efficiency of attacks detection, trust management system, and ensuring security and reliability in future 5G communication network.