Intrusion Detection System for loT Using CICloT2023 Dataset

Development in electronics manufacturing and 5G networks have made The Internet of Things (loT) an expanding technology that allows electronic devices to interact and share data with no human involvement. Nevertheless, this expansion escalates the susceptibility to security breaches. DDoS is a severe attack that commonly affects loT networks. Intrusion detection systems (IDSs) are utilized to secure loT devices by continuously monitoring network activities and instantly identifying and responding to any threats and attacks. Nevertheless, the implementation of Intrusion Detection Systems (IDS) in Internet of Things (loT) networks poses various obstacles, such as the limitations in processing resources, battery life and memory that loT devices face. This study employed lightweight machine learning methods, including decision tree, logistic regression and random forest, to construct intrusion detection models for the identification of network attacks. The most recent dataset, CICloT2023, was used, which contains many attacks that were not present in previous loT datasets. We examined the model's performance by measuring accuracy, Fl-score, recall, precision, and the time taken for training and testing. The results indicate that the random forest and decision tree models achieved superior performance compared to other detection models, achieving a high accuracy rate. In terms of time efficiency, the decision tree model demonstrated superior performance compared to other models.