The advancement in IoT has prompted its application in areas such as smart homes, smart cities, etc., and this has aided its exponential growth. However, alongside this development, IoT networks are experiencing a rise in security challenges such as botnet attacks, which often appear as network anomalies. Similarly, providing security solutions has been challenging due to the low resources that characterize the devices in IoT networks. To overcome these challenges, the fog computing paradigm has provided an enabling environment that offers additional resources for deploying security solutions such as anomaly mitigation schemes. In this paper, we propose a hybrid anomaly mitigation framework for IoT using fog computing to ensure faster and accurate anomaly detection. The framework employs signature- and anomaly-based detection methodologies for its two modules, respectively. The signature-based module utilizes a database of attack sources (blacklisted IP addresses) to ensure faster detection when attacks are executed from the blacklisted IP address, while the anomaly-based module uses an extreme gradient boosting algorithm for accurate classification of network traffic flow into normal or abnormal. We evaluated the performance of both modules using an IoT-based dataset in terms response time for the signature-based module and accuracy in binary and multiclass classification for the anomaly-based module. The results show that the signature-based module achieves a fast attack detection of at least six times faster than the anomaly-based module in each number of instances evaluated. The anomaly-based module using the XGBoost classifier detects attacks with an accuracy of 99% and at least 97% for average recall, average precision, and average F1 score for binary and multiclass classification. Additionally, it recorded 0.05 in terms of false-positive rates.
T4SE-XGB: Interpretable Sequence-Based Prediction of Type IV Secreted Effectors Using eXtreme Gradient Boosting Algorithm
