Abstract
Today, Smart City projects and their initiatives are continuously developing with the vast deployment of the Internet of Things (IoT) devices. Smart cities efficiently manage assets, services, and resources with better IoT technologies and new smart devices. However, their deployment elevates the potential security threats making IoT smart city projects vulnerable to disparate attacks. Hence, Intrusion Detection Systems (IDS) must be developed for IoT-enabled smart cities to alleviate IoT-associated security attacks. Achieving high-level detection accuracy and lower training time are significant challenges in IDS. However, the existing traditional IDS are inefficient in handling them. This paper proposes a hybrid optimization and deep-learning-centric IDS to face these challenges in IoT-enabled smart cities. The dataset initially undergoes pre-processing to acquire an effective and accurate IDS. Then, feature selection and clustering are performed utilizing the Hybrid Chicken Swarm Genetic Algorithm (HCSGA) and MinK-means Algorithm. Lastly, the transformed data is loaded to the Deep Learning-based Hybrid Neural Network (DLHNN) classifier, classifying the normal and attack data. The proposed model is validated with the NSL-KDD dataset and the experimental outcome shows the efficiency of the proposed IDS compared with similar state-of-the-art techniques.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call