Hybrid Metaheuristic-Driven Intrusion Detection System Using Opto-Romar Swarm Bee Genesis Optimization on IoT Network Data
Swetha A, Ramesh Sekaran, Annamalai Selvarajan · International Journal of Electronics and Communication Engineering · 2025
The Internet of Things (IoT) devices have rapidly grown in numbers, posing critical threats in the process of securing networks against emerging cyber-attacks. Traditional Intrusion Detection Systems (IDS) suffer from low accuracy, are not flexible and are inefficient in handling massive multidimensional IoT data. To mitigate these shortcomings, the study presents a new hybrid metaheuristic-based IDS system that combines Cat-Scale normalization technique, the feature selection algorithm: Mutualk-Best, the optimization technique Opto-Romar Swarm Bee Genesis and the hyperparameter optimization algorithm: Evalmax Hyper Net. The framework is developed to provide a balanced representation of features, minimize redundancy, enhance convergence, and learn dynamic parameters to provide robust intrusion detection. Experimental results on the IoT-IDS dataset show the effectiveness of the proposed work. The framework recorded 95% accuracy, 97% precision, 98.6% recall and 98.4% F1-score with an AUC of 0.9999, utilizing better results compared to other techniques of IDS like BESO-HDL and Modified Isolation Forest. These findings support that the combination of swarm intelligence and genetic algorithms, along with adaptive tuning, can present a better detection performance with the ability to scale in highly complex IoT environments. The results indicate the framework as a promising future-proof IDS solution. Future work will consider the deployment of real-world IoT to resource-limited devices, robustness against adversarial attack, and variants to edge and mobile computing tasks.