Streamlining IoT Malware Detection:A Pipeline Based Approach

Internet of Things (IoT) malware detection has become a significant cybersecurity challenge due to the expansion of the threat landscape brought about by the proliferation of IoT devices. IoT malware is growing more complex and dynamic every day, making traditional signature-based detection techniques ineffective against it. By using large-scale data generated by IoT devices to identify patterns and anomalies, machine learning (ML) techniques present a promising way to detect IoT malware. The process of developing effective machine learning models to identify malware for Internet of Things, however, can be difficult and time-consuming owing to the requirementfor meticulous preprocessing,feature engineering, model selection, and evaluation.Here, a pipeline techniquecombined with machine learning to detect IoT malware is presented in a comprehensive way. In order to increase the resilience and accuracy of the detection system, specific methods are used like data cleaning, dimensionality reduction, ensemble learning and cross validation. Standard Dataset UNSW_NB15 is used to evaluate proposed models like XGBoost Classifier, AdaBoost Classifier, Random Forest, ExtraTrees Classifier and Stochastic Gradient Descent(SGD).