Anomaly Identification in Real-Time for Predictive Analytics in IoT Sensor Networks using Deep

Abstract

The proliferation of IoT devices has exposed networks to an increased risk of cyberattacks, as their number is increasing at an exponential rate. This has resulted in more sophisticated methods for identifying outliers. This research suggests a DL architecture for the Internet of Things (IoT) sensor network based on the Long Short-Term Memory (LSTM) model for anomaly detection. The ToN-IoT data on Kaggle was utilized and it was undergone through numerous preprocessing processes like missing values, label encoding, normalization and class balancing through the use of SMOTE. The number of IoT devices is growing at an exponential rate and this has augmented the susceptibility of networks to cyber threats. This has led to advanced techniques of detecting abnormalities. The results prove that the model is accurate and valid for identifying typical and non-standard network behavior. The proposed framework is an intelligent framework of detecting anomalies in an intelligent way to enhance the security of the IoT network on a scalable, high-quality, and real-time framework.