Toward securing the integrity of memory storage devices: classification of solid-state drives using power-based side-channel analysis

Toward securing the integrity of memory storage devices: classification of solid-state drives using power-based side-channel analysis

연구 분야: Analysis

학회: The Journal of Supercomputing

Solid-state drive (SSD) technology continues to evolve, with advancements in established technologies like 3D NAND competing with innovations such as phase change memory, featured in Intel’s Optane-branded memory from 2017 to 2022. As with traditional NAND flash SSDs, the firmware on these newer drives can make it difficult for users to verify whether operations are functioning as intended and uncover unanticipated behavior. This is an increasingly critical issue as instances of firmware tampering and counterfeit drives become more prevalent in the SSD marketplace. This work proposes a classification-based method that can be used to detect these security compromises. We leverage power-based side-channel analysis and machine learning techniques to classify operations and characteristics of three SSD models from different manufacturers, employing phase change or 3D NAND memory technologies. We present sample waveforms, develop associated classifiers, and achieve classification rates of 94.3% by drive operation, 98.9% by drive model, and 99.8% by technology employed. Additionally, we demonstrate that power-based side-channel analysis can be used to identify and investigate anomalies impacting performance or reliability.