A Hybrid CNN-LSTM Model for Enhanced Intrusion Detection in Internet of Things Environments

The exponential expansion of the Internet of Things (IoT) has precipitated a concerning increase in advanced cyber-attacks, posing significant problems for the protection of these networked systems. Traditional Intrusion Detection Systems (IDS) frequently encounter difficulties in handling the complexity and high-dimensional characteristics of IoT traffic, especially in situations involving various attack vectors. To mitigate these restrictions, we present a hybrid deep learning model that amalgamates Convolutional Neural Networks (CNN) with Long Short-Term Memory (LSTM) networks. By integrating the advantages of various architectures, the model proficiently identifies spatial patterns using CNNs and temporal dependencies through LSTMs, facilitating precise traffic classification. Utilizing the NF-BoT-IoT dataset, our methodology exhibited remarkable efficacy, attaining an accuracy of 99.52% and an F1-score of 0.9923 in binary classification, with an accuracy of 98.89% in multi-class classification across several attack categories. These findings highlight the model's resilience and its capability as a real-time intrusion detection system, facilitating improved security in IoT networks across diverse settings.