Class Scatter Ratio Based Mahalanobis Distance Approach for Detection of Internet of Things Traffic Anomalies

Hackers are drawn to attack Internet of Things (IoT) devices as a result of technological improvements. By spotting irregularities in IoT connections, smart devices must be safeguarded against these threats. Statistical methods are frequently used to identify intruders in IoT connections. The curse of dimensionality makes it difficult for current statistical anomaly detection systems to detect anomalies with high accuracy when the statistical presumptions are changed. A method known as Class Scatter Ratio based Mahalanobis Distance (CSRMD) is suggested as a solution to these problems. The CSRMD determines anomalies based on how closely a group of attributes correlates with one another after culling attributes that are crucial to anomalies. CSR culls the attributes by calculating the weights and ranking of the attributes. Using the Mahalanobis Distance Graph (MDG), the interrelationship between the selected attributes is calculated. By using the generated IoT normal traffic description and the MDG of the new traffic, the anomaly is discovered. Studies using the publicly available benchmark dataset demonstrate that the suggested CSRMD technique is yielding fruitful results. It has been demonstrated that the proposed approach outperforms state-of-the-art anomaly detection approaches with regard to accuracy.