ENHANCING CYBERSECURITY VIA ANOMALY RECOGNITION USING THERMAL EXCHANGE FRACTALS OPTIMIZATION WITH DEEP LEARNING ON IOT NETWORKS

The Internet of Things (IoT) refers to the interconnected network of objects and devices that seamlessly communicate and share information. The need for robust cybersecurity measures becomes paramount with the increase of IoT devices, ranging from smart home devices to industrial sensors. The inherent vulnerability of the IoT ecosystem to cyber threats necessitates cutting-edge security protocols to ensure the integrity of connected systems and safeguard sensitive information. IoT security is crucial to protect against potential manipulation of connected devices, unauthorized access, and data breaches. An essential facet of IoT cybersecurity, Anomaly detection, includes the detection of unusual behaviors or patterns in device activity or network traffic in many complex systems that may indicate security breaches. Deep learning (DL), with its ability to analyze complex and vast datasets, has improved anomaly detection in IoT environments. By leveraging DL techniques, IoT systems can better adapt to evolving cyber threats, which offer a proactive defense system against complex cyber threats in various complex systems. In essence, incorporating anomaly detection and DL within the IoT cybersecurity framework is crucial to ensure the entire IoT ecosystem's trustworthiness and fortify interconnected devices' resilience. This study presents an anomaly recognition using fractals thermal exchange optimization with deep learning (ARA-TEODL) technique for cybersecurity on IoT Networks. The ARA-TEODL technique focuses on identifying anomalous behavior in the IoT network to achieve cybersecurity. In the ARA-TEODL technique, Z-score normalization is primarily used to scale the input networking data. Besides, the selection of features takes place utilizing the chimp fractals optimization algorithm (ChOA). Moreover, a modified Mogrifier long short-term memory (MM-LSTM) model is used to identify anomalies in the network. Finally, the hyperparameter tuning process takes place using the TEO algorithm. The experimental evaluation of the ARA-TEODL technique takes place using a benchmark dataset. The experimental results stated that the ARA-TEODL technique reaches optimal cybersecurity in the IoT networks.