A Framework For the Detection of Malicious Activities on Edge Computing Using Random Forest Classifier and Recurrent Neural Network

Edge computing is a paradigm that involves the transfer of a portion or the entirety of cloud computing tasks to localized edge devices as required. This approach can enhance performance in situations where the network infrastructure poses a constraint on the timely delivery of services. Edge nodes, akin to threat monitors or sensors, are extensively distributed throughout the Internet. While they may not possess the same level of hardware and processing capabilities as data centers, they are still capable of efficiently supporting extensive parallel applications and delivering prompt service for intricate computations. This study presents a robust approach to malware detection in the context of edge computing, leveraging a Recurrent Neural Network (RNN) model trained on feature sets extracted from a Random Forest Classifier. The proposed model demonstrates exceptional performance, achieving an impressive accuracy of 98.99% coupled with an exceedingly low false positive rate of 0.03%. By combining the strengths of both machine learning paradigms, this methodology showcases significant advancements in safeguarding edge computing environments against malicious software, thereby fortifying the security infrastructure of decentralized computing systems.