Network Intrusion Detection System Using an Evolved Gene Regulatory Network

The extensive inter-connectivity and interoperability of computing systems have become essential for fulfilling our daily activities. However, this increased connectivity has also led to larger attack surfaces and raised numerous security issues beyond human control, allowing malicious actors to exploit vulnerabilities in one system to compromise others more easily. Network intrusion detection is one of the security measures that counter evolving security threats to maintain secure communication. Adaptability and scalability are recognized issues in the development of network intrusion detection systems where many traditional machine learning systems may struggle to analyze large volumes of data in real time. For many years, biological complexity has been a source of inspiration for computer scientists. By modeling the underlying processes that drive this complexity, researchers aim to create systems capable of generating sophisticated behaviors, which can be leveraged for solving challenges in computer science and engineering. This paper introduces a novel bio-inspired model based on Gene Regulatory Networks as an alternative to traditional machine learning models for detecting and classifying network attacks. Our method shows results which are comparable to Machine Learning based methods when tested on UNSW-NB15 dataset, with an accuracy of 89.25% with a false positive rate of 3.81%.