A Deep Learning Approach for BGP Security Improvement

Border Gateway Protocol (BGP) is a critical protocol for inter-domain routing in the Internet. However, it suffers from security vulnerabilities due to the lack of authentication and the trust-based nature of the protocol. As a result, BGP attacks are becoming increasingly common, causing significant damage to network availability and security. In this paper, we propose a deep learning-based approach to improve BGP security. Our approach uses a neural network to analyze BGP messages and detect. We trained and evaluated the performance of our model using RIPE and BCNET datasets. The results from our experiment demonstrate that our method has the capability to effectively identify BGP attacks with a high level of accuracy, achieving a 98% detection rate (accuracy). Additionally, our approach exhibits a precision of 0.98 and a recall of 0.96%, outperforming state-of-the-art BGP security solutions. Our proposed approach has the potential to significantly improve BGP security and protect the Internet’s routing infrastructure from malicious attacks.