Stacked Classification Model with Cryptographic Process in IoT Data to Prevent and Detect Attacks
Authors:-N. Shashikala, T.N Anitha, Priti Mishra, Renuka Patil Herakal, Jayasudha Kolur
Abstract-The paper introduces a novel approach termed Gradient Optimization Features Integrated Stacked Classifier (GOFI-SC) for enhancing intrusion detection and attack prevention in Internet of Things (IoT) environments. GOFI-SC integrates feature optimization, cryptographic processes, and stacked classification models to effectively identify and classify various types of attacks. The feature importance analysis underscores the significance of attributes such as source IP address, destination IP address, and protocol in detecting malicious activities. Optimization of features with GOFI-SC demonstrates progressive reduction in loss values over iterations, indicating improved model convergence. the cryptographic process ensures secure communication and data integrity, with encryption and decryption speeds of approximately 150 MB/s and 175 MB/s, respectively. Classification results exhibit high accuracy, precision, recall, and F1-scores across different epochs and attack types, with accuracy reaching up to 0.990 and F1-scores exceeding 0.990.
