Abstract: This paper presents the Quantum-inspired Fuzzy Destructive Interference (QFDI) model, a novel hybrid intrusion detection system that treats network flows as wave phenomena rather than static feature vectors. Normal traffic is encoded as a multi-dimensional reference wave; any test flow is subjected to destructive interference with this reference. Perfect cancellation indicates benign traffic, while residual energy—graded by a lightweight Fuzzy Inference System (FIS) comprising eight data-driven If-Then rules—indicates an attack. Experiments on the UNSW-NB15 benchmark dataset demonstrate that QFDI achieves a False Positive Rate (FPR) of 0.079, representing a 77.5% reduction compared to Random Forest (0.350) and a 76.5% reduction compared to XGBoost (0.336), while attaining a Precision of 0.903. QFDI deliberately trades higher recall (0.600) for significantly lower false positives, making it ideal as a high-precision second-stage filter in layered IDS architectures. The complete model requires only 39 numerical parameters and 0.30 KB of memory, enabling deployment on embedded and IoT devices where state-of-the-art deep learning models are computationally infeasible. QFDI is fully interpretable through eight linguistic rules, addresses alert fatigue in security operations centres, and requires no offline training phase.

Keywords: Intrusion Detection System, Quantum-Inspired Computing, Fuzzy Logic, Destructive Interference, Network Security, Embedded Systems, False Positive Reduction, UNSW-NB15

Pages: 29-35

WSEAS Transactions on