Comparing the Performance of Artificial Intelligence Techniques for Internet of Things Security

Abstract

The use of AI into IoT security has become an important step forward, greatly improving the capacity to identify, stop, and react to cyber-attacks in a digital environment that is very interdependent. Cyberattacks on IoT devices have become more common as their number has grown, highlighting the necessity for strong security protocols. There needs to be strong detection and mitigation strategies developed since the proliferation of Internet of Things (IoT) devices has brought huge security risks. This research looks at how well four different AI methods—Support Vector Machine (SVM), Decision Tree, Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM)—handle security risks associated with the Internet of Things (IoT). These models were trained and evaluated using publicly accessible datasets, such as CICIDS2017, NSL-KDD, and UNSW-NB15. Important measures including F1-score, recall, accuracy, and precision were used to evaluate the efficacy of each AI method.