Malicious Traffic Detection Framework in Internet of Things Using Optimized Linearly Regressed Deep CNN Classifier
Priyanka Sarath · 2023
Malicious traffic detection (MTD) in the Internet of Things (IoT) network is necessary for its safety to block the unnecessary flow of traffic when occurs in the IoT system since IoT is gradually providing a broad range of applications. In this research, an optimized Linearly Regressed Deep Convolutional Neural Network (DCNN) classifier is employed for detecting malevolent traffic in IoT. The information preprocessing is done using SMOTE to minimize the imbalance issues and assist to overcome the overfitting. The Optimized feature selection is done with the hybrid colonial optimization (HCO), which is used to detect the malevolent traffic and then the optimized logistic regression (LR) based DCNN classifier classifies the traffic as normal or malicious in an effective manner, which assists to predict the probability of occurrence of the malicious traffic. Thus, the supremacy of the research is established with the performance metrics like accuracy, sensitivity, and specificity that attained the values of 94.51%, 95.05%, and 94.19% at 90% of training.