A systematic literature review on deep learning-driven IDS for software-defined WSN

Abstract

Wireless Sensor Networks (WSNs) are self-organizing, multi-hop systems composed of stationary or mobile sensors that monitor and transmit environmental data. Despite their widespread applications, WSNs face significant security challenges due to their inherent limitations in power, storage, bandwidth, and processing capabilities. To address these constraints, Software-Defined Networking (SDN) has been integrated into WSNs, forming Software-Defined Wireless Sensor Networks (SDWSNs). This integration enhances flexibility, scalability, and overall network performance. However, SDWSNs remain vulnerable to sophisticated cyber threats and suffer from poor network management and inefficient routing, which compromise both security and Quality of Service (QoS). Intrusion Detection Systems (IDS) are essential for safeguarding SDWSNs, especially under resource-constrained conditions. Traditional IDS approaches are increasingly ineffective against evolving, intelligent attacks. This systematic literature review presents recent advancements in IDS frameworks powered by Machine Learning (ML) and Deep Learning (DL) techniques. It explores foundational concepts, architectural models, and the application of various DL algorithms in intrusion detection. The review also identifies current research gaps and outlines future directions for developing intelligent, adaptive, and resource-efficient IDS solutions tailored to SDWSNs.