Enhancing Cybersecurity in Smart Cities: IoT Applications with a Hybrid Deep Neural Network Model

Continuous and pervasive data flow is made possible by the IoT, a network of networked computing devices. The expansion of smart cities, which employ IoT apps extensively to boost operational efficiency, service quality, and people's well-being, is largely attributable to this trend. Malicious assaults can penetrate IoT devices through sensors linked to big cloud servers, which has increased the risk of cybersecurity risks, which has coincided with the rise of smart city networks. This study proposes a cybersecurity strategy for IoT applications in smart cities that makes use of sophisticated machine learning methods to tackle these issues. The dataset was refined using preprocessing procedures, which included scalar and normalisation functions. A pair of entropy-based techniques, IG and GR, were employed for feature extraction and selection. The data was processed using a CBiLSTM hybrid deep learning model, where the CNN layer was responsible for capturing spatial information and the BiLSTM layer for extracting temporal relationships. When compared to traditional models, the suggested model's 99.14% classification accuracy for identifying cybersecurity risks was significantly higher. This research lays the groundwork for better digital protection by demonstrating the value of hybrid DL approaches in bolstering IoT security in smart cities.