Cyber Security Threat Detection in Internet of Things Using Optimized Deep Learning Technique

Cybersecurity threat detection in the Internet of Things (IoT) refers to the process of identifying and addressing potential security risks and attacks within IoT systems and devices. It involves monitoring IoT networks and devices for abnormal behavior, detecting intrusions or unauthorized access attempts, ensuring the integrity of data, managing vulnerabilities, analyzing behavioral patterns, and staying informed about emerging threats. This ongoing process aims to maintain the security and integrity of IoT infrastructure by proactively identifying and mitigating potential security threats. Recent studies have demonstrated that deep learning algorithms are highly efficient and offer numerous advantages compared to earlier methods when it comes to conducting security assessments of IoT devices. Dual Channel Convolution Neural Network with Spider Monkey Optimization (DCCNN-SMO), a deep learning technique which visualized colored images for finding harmful intrusions in IoT network. Here, DCCNN used two channels of CNN and combined in fully connected layer for detecting intrusions in IoT network. The hyper parameters of DCCNN were fine tune by SMO which enhance the performance of DCCNN. In this paper, the convergence speed and effectively fine tuning the hyper parameters of DCCNN an Adapted SMO (ASMO) is introduced for better detection of intrusions in IoT network. The position update equation of the basic SMO algorithm is modified by incorporating a dual seek scheme. This approach, employed in the local leader (LL) phase, global leader (GL) phase, and LL decision phase, facilitates the acquisition of fresh outcomes, representing best hyper parameters of DCCNN for better detection of intrusion in IoT network. The performance of the proposed DCCNN-ASMO is demonstrated to be superior in terms of accuracy, precision, and recall, as evidenced by the experimental results.