Methodological Framework for Developing an Adaptive Intrusion Detection System (IDS) Incorporating Sustainability Factors

Abstract

Cybersecurity has emerged as a global concern, amplified by the rapid expansion of IoT devices and the growing digitization of systems. In this context, traditional security solutions such as firewalls and static signature-based IDS prove increasingly ineffective in detecting evolving and sophisticated cyber threats. This issue is particularly critical in Africa, where limited resources, technological dependency, and outdated infrastructures exacerbate vulnerabilities. Next-generation firewalls (NGFWs), though powerful, are often financially and operationally inaccessible in these regions. To address these limitations, this paper advocates for the development of low-cost, adaptive security solutions that integrate core principles from software engineering specifically, Lehman’s laws of software evolution. We propose a novel methodological framework for designing an intelligent Intrusion Detection System (IDS), grounded in three pillars: reinforcement learning for dynamic threat response, the application of Lehman’s laws to ensure long-term adaptability, and software product line engineering to allow context-specific customization. The resulting system was implemented within a core network environment and achieved a detection accuracy of 99.99%, validating the effectiveness of this approach. Beyond this implementation, future extensions include deploying the system in IoT and critical infrastructure environments, and incorporating advanced AI methods such as federated learning and generative models. The findings of this study highlight the potential of combining adaptive AI with sustainable design principles to overcome the shortcomings of conventional cybersecurity models. The proposed IDS offers a scalable, robust, and locally applicable alternative particularly for regions facing structural and technological constraints.