Review on Hybrid AI-Based Intrusion Detection System with Quantum-Inspired Optimization

One after another, the half-dozen studies pointed out weaknesses in classic intrusion detection when dealing with modern IoT needs. Faced with these shortfalls, researchers began exploring blends of artificial intelligence approaches. With quantum-inspired techniques entering the mix, collective intelligence models added extra strength. Detection rates climbed sharply - at the same time, false alerts grew far less frequent. What stayed clear was that older methods simply could not keep pace. Now deep learning stepped in, helping spot which bits of data were worth keeping. Only when everything connected did efficiency take a whole new form. Older setups falter, yet fresh designs handle hidden cyber risks better. Rather than copying outdated methods, certain programs blend quantum concepts with clever grouping strategies. Wolves hunting in packs inspire one method - sifting through tangled information early on. Drifting plankton guide another, adjusting focus where it counts. Patterns from nature let computers see routine actions versus real trouble with sharper eyes. Out of nowhere, small vision transformers - made lean by design - spot slight changes in how factory networks move data. Not their bulk that matters, rather the way they take in surroundings, much like someone watching quietly from the edge. Still, after gains made so far, certain hurdles refuse to fade. What catches attention most? A lot of quantum-based safety tech operates inside simulated settings instead of live quantum hardware. That gap means glitches found outside labs - like fading signals, noise from surroundings, or limits built into real gear - tend to slip through cracks. When those factors stay ignored, rolling out working versions feels uncertain, at minimum. Apart from that, quite a few of the reviewed studies stick to one dataset - NSL-KDD shows up most often - with only a handful using even two sources When models appear to pick up just certain patterns rather than catch breaches consistently, doubts start forming. Since performance can shift between setups, checking them on various data sets through uniform approaches helps reveal what they actually handle. Most heavy-duty number crunching trips up hybrid quantum-classical systems. Quick threat detection in connected gadgets? That needs pace - yet low-energy units lag behind. Still, plenty of today’s approaches ignore the sheer hassle of fitting them into actual environments. Consider QBCMVT: solid performance shows, but only if you accept its hunger for processing muscle - which blocks smooth operation on edge machines until models shrink down. Future research should focus on the following areas:   •      Development of lightweight hybrid IDS architectures for IoT edge devices   •      Validation of quantum-enhanced IDS models on real quantum hardware   •      Standardized cross-dataset benchmarking methodologies   •      Energy consumption and latency analysis for real- time deployment   •      Adaptive online intrusion detection systems capable of handling evolving attack patterns Overall, the reviewed studies establish a strong foundation for future research in hybrid AI-based intrusion detection systems. Although practical deployment challenges remain significant, the combination of quantum computing, swarm intelligence, and advanced deep learning architectures represents a promising direction for nextgeneration IoT and IoT cybersecurity solutions.

-[1]  R. Kumar and M. Swarnkar, “QuIDS: A Quantum Support Vector Machine-Based Intrusion Detection System for IoT Networks,” Journal of Network and Computer Applications, vol. 234, Art. no. 104072, Feb. 2025, doi:10.1016/j.jnca.2024.104072. [2]  I. B. Bharathi, V. Sonai, and S. Sridevi, “QuantumDriven Enhanced Machine Learning Algorithm for Intrusion Detection in Internet of Things Environment,” EPJ Quantum Technology, vol. 13, no. 1, Art. no. 20, Jan. 2026, doi:10.1140/epjqt/s40507-026-00463-5. [3]  W. Chen et al., “Large-scale IoT Attack Detection Scheme Based on LightGBM and Improved Salp Swarm Algorithm,” Scientific Reports, vol. 14, 2024. [4]  E. I. Elsedimy et al., “A Novel Intrusion Detection System Based on Hybrid Quantum Support Vector Machine and Improved Grey Wolf Optimizer,” Cluster Computing, vol. 27, 2024. [5]  “QBCMVT: An Effective Quantum-Based CoatiMobileViT Model for Intrusion Detection in IIoT,” Computers and Electrical Engineering, vol. 126, 2025.