Abstract:
Using ML in malware protection systems for fog computing platforms is a groundbreaking way to enhance cybersecurity in the age of the Internet of Things. Because
fog computing is located close to end devices and has a decentralized design, it offers
significant advantages in terms of reduced latency and bandwidth use. However, there
is a greater chance of being targeted by sophisticated malware with these benefits. Due
to the dynamic and sophisticated nature of new cyber threats, traditional security
solutions often fall short of offering robust protection. Key machine learning methods, such as Random Forest, Logistic Regression, and
Neural Networks are assessed for their efficacy in detecting and thwarting malware in
fog computing environments. Using real-world datasets, the methodology entails
thorough data collection and preprocessing, model training, and validation. The
purpose of deploying these models on fog nodes is to provide low latency and real-time
threat detection, which are essential for IoT applications to operate efficiently. ML
models in fog computing improves overall system reliability by reducing false positives
and increasing malware detection rates. The ethical issues covered in this paper include
algorithmic bias, data privacy, and the requirement for openness in ML decisionmaking procedures. It also outlines the software & hardware prerequisites needed for
these models to be seamlessly integrated into the frameworks for fog computing that
are currently in use. Notwithstanding the encouraging progress, there are still a number of unresolved
problems, such as the requirement for high-quality datasets, guaranteeing the resilience
of the model against hostile attacks, and getting beyond the computational limitations
posed by fog nodes. The consideration of the wider social and environmental effects of
ML-enhanced security measures that closes the report highlights the possibility of
enhancing operational effectiveness and fostering more confidence in digital
technology
