Lstm-based attack clasification in Iot networks

Abstract

The proliferation of Internet of Things (IoT) devices has heightened the need for robust security measures to protect against a growing number of cyberattacks. In this project, a Long ShortTerm Memory (LSTM) neural network approach is presented for classifying attacks in IoT networks, leveraging the comprehensive IoT-23 dataset. The objective of the research is to develop an effective Artificial Intelligence (AI) model that accurately identifies various types of network intrusions and anomalies indicating botnet infiltration, thereby enhancing the security posture of IoT environments. The IoT-23 dataset is preprocessed to extract relevant features, and the LSTM model is trained using these inputs. Superior performance is demonstrated by this approach, achieving an accuracy of 98.8% in classifying attacks, thereby improving upon traditional machine learning models. These results underscore the potential of LSTM networks in IoT security applications, offering a scalable and adaptive solution to detect emerging threats. Future work will explore the integration of this model into real-time security systems and its applicability to diverse IoT architectures.