Title: An Efficient Intrusion Detection System for IoT-Based Cybersecurity Using a Deep Temporal Convolutional Network
Authors: N Moorthy
Volume: 10
Issue: 2
Pages: 21-37
Publication Date: 2026/02/28
Abstract:
The fast proliferation of Internet of Things (IoT) devices has caused the network to become highly complicated, and the IoT environment has become highly vulnerable to various cyber threats. Conventional intrusion detection systems (IDSs) and conventional machine learning methods do not have the capability to proficiently analyze IoT network traffic in high dimensions, heterogeneity, and time sensitivity. In order to overcome these difficulties, this study offers an effective intrusion detection model, which relies on a Deep Temporal Convolutional Network (DTCN). The suggested model combines the one-dimensional Convolutional Neural Networks (CNN) to extract spatial features, the Bidirectional Long Short-Term Memory (Bi-LSTM) network to learn the temporal dependencies, and the Deep Neural Network (DNN) layers to be capable of classifying various classes with robustness. Performance evaluation is conducted on the CIC-IDS-2017 dataset and goes through extensive preprocessing in terms of data cleaning, feature normalization, class balancing, and time sequencing. It has been proved experimentally that the proposed DTCN has a high detection accuracy, precision, recall, and F1-score, and a very low false positive and false negative rate. Comparative analysis reveals that the suggested framework is always better than classical machine learning frameworks like SVM, Random Forest, and XGBoost. On balance, the findings indicate that the DTCN model is a successful tool in the complex spatial-temporal intrusion patterns and can be used as a reliable and scalable tool in cybersecurity of IoT).