Robust Intrusion Detection Systems with Explainable Artificial Intelligence: A Review

With the recent growth of interconnected networks, Internet of Things (IoT), cyber-physical systems, and next-generation mobile communication networks, the complexities and dynamics of cybersecurity threats are rising at an unprecedented level. Intrusion Detection Systems (IDS) are important defense systems that detect harmful activities from the network traffic. Traditional approaches to IDS are faced with severe limitations regarding their high rates of false positives, low malleability to zero-day attacks, and low resistance to adversarial attacks. Recently, the use of advancements from Machine Learning (ML), Deep Learning (DL), and Explainable Artificial Intelligence (XAI) concepts gained significance to boost the resistance, interpretability, and trustworthiness aspects of IDS systems. Explainable AI allows security experts to better comprehend the rationale behind the IDS system’s predictions, observe anomalies, and optimize their response strategies accordingly. A thorough literature review is introduced, depicting the overall analysis of recent IDS systems integrated with the concept of Explainable AI, along with their algorithms, datasets, strengths, and weaknesses. A literature review analysis is introduced, followed by an embryonic approach on how IDS systems can become Explainable and Resistant.