An Early-Exit Hybrid Intrusion Detection Architecture for Resource-Constrained Iot Edge Devices

Abstract

As the Internet of Things (IoT) equipment spreads very fast; embedded systems have become the best targets of edge computing environments to complex network attacks. Conventional Network Intrusion Detection Systems (NIDS), on the contrary, is resource-prohibitively expensive when the edge nodes are constrained, requiring far CPU cycles and memory capable of surpassing that of typical embedded Linux hardware. This paper suggests and installs a Lightweight NIDS particularly in resource-limited platforms like Raspberry Pi and IoT gateways. The system takes a new hybrid architecture which embraces a new system, fuses a deterministic, rule-based fast-path detector layer with a lightweight and highly quantized Machine Learning (ML) inference engine. Stateless flow tracking and minimized memory footprint is achieved by his aggressive dimensionality reduction, which gave strong detection of Port. Scans, Distributed Denial-of-Service (DDoS) floods and anomalous traffic patterns in keeping CPU utilization less than 15% and RAM consumption under 200 MB. The architecture incorporates the Packet Capture, temporal feature extraction, multi-class, hybrid threat detection classification, and automatic response of firewalls into one coherent pipeline. The outcomes of the experimental work with synthetic network traffic prove that the proposed system can attain competitive detection accuracy and meeting strict embedded performance constraints, providing an effective and edge computing security solution, which is scalable.