RT-HMD: A Novel Statistical Approach for Robust Real-Time Hardware-based Malware Detection under Weak Supervision Formulation

In the rapidly evolving landscape of cybersecurity threats, the need for efficient and real-time malware detection mechanisms, especially at the hardware level, has never been more critical. Traditional software-based malware detection solutions, while useful, often fall short in the face of sophisticated attacks that target the hardware layer. Moreover, the reliance on heavily supervised learning models requires extensive labeled data, which is not always feasible for emerging malware threats.

Addressing these significant challenges, our work, RT-HMD, presents a novel statistical approach designed specifically for hardware-based malware detection that operates effectively under weak supervision. By harnessing the power of weakly supervised learning, RT-HMD is able to detect and mitigate malware threats in real-time, even in scenarios where only limited or imprecise training data is available. This method not only enhances the detection capabilities against advanced malware attacks but also significantly reduces the dependency on extensive labeled datasets, making it a highly practical solution for protecting critical hardware systems.

Our methodology integrates seamlessly with existing hardware infrastructures, providing a robust layer of security without compromising on performance or efficiency. As we move towards an era where hardware vulnerabilities are increasingly exploited, the development of RT-HMD represents a crucial step forward in fortifying cybersecurity defenses from the ground up.

By submitting our work to ISLPED 2024, we aim to contribute to the broader discourse on low-power electronics design and cybersecurity, highlighting the urgent need for innovative solutions like RT-HMD that ensure the security and integrity of hardware systems in an energy-efficient manner.

Recommended citation: Harshit Kumar, Sudarshan Sharma, Biswadeep Chakraborty, Saibal Mukhopadhyay. (2024). "RT-HMD: A Novel Statistical Approach for Robust Real-Time Hardware-based Malware Detection under Weak Supervision Formulation." Submitted in ISLPED 2024.