An Effective Cyber Security Threat Detection in Smart Cities Using Dueling Deep Q Networks

In recent days, smart city technologies are developing rapidly with the incorporation of IoT devices and AI-driven systems has modernized and thus facilitated for enhancing productivity, sustainability, and connectivity. The motive is to mitigate the rising cyber threats which are affecting the interrelated organizations of smart cities. Despite advancements in threat detection technologies, challenges persist, including real-time anomaly detection in vast data streams, managing resource constraints on IoT devices, and ensuring adaptability against evolving attack vectors. Therefore, this research suggests a novel approach by using Dueling Deep Q Networks (DDQN) method. Initially, the data is collected from the dataset and then Inter Quartile Range (IQR) is imposed at the preprocessing stage for detection of outliers. Then, extraction of optimal features is done by using Term Frequency-Inverse Document Frequency (TF-IDF) and Recursive Feature Elimination (RFE) which transforms the data into statistical vectors and eliminates the least significance feature respectively. Finally, the obtained optimal features are detected using DDQN method as it has the ability to detect effectively with respect to the state of action. The proposed DDQN has resulted with accuracy of 99.32% when compared with Improved Bacterial Foraging Optimization using optimum deep learning for Anomaly detection (IBFO-ODLAD).