Trustworthy IAP: An Intelligent Applications Profiler to Investigate Vulnerabilities of Consumer Electronic Devices

As a typical representative of the Internet of Energy (IoE) intelligent era, consumer electronic (CE) devices continue to evolve at a remarkable pace. Computers, as typical and essential CE devices, have been instrumental in enhancing efficiency, communication, entertainment, and information access. As part of this evolution, a significant trend in computer design focuses on achieving low power consumption while maintaining high performance. For instance, a computer’s central processing unit (CPU) dynamically modulates its output power in response to the varying workload demands of running applications. However, these power efficiency mechanisms may inadvertently introduce implicit patterns into the operational states of CE devices. Particularly, the power consumption of a CE device executing various tasks can manifest distinguishable temporal patterns, thereby exposing potential vulnerabilities. Thus, this work aims to investigate the vulnerabilities of CE devices on power consumption mechanisms. We focus on exploring the possibility of using alternating current (AC) power consumption to infer the running applications on a consumer computer. To achieve that, we construct a physical attack system that employs data acquisition, processing, classification, and inference stages to establish a “profiler” for application profiling. The extensive experiment results on the self-collected power consumption dataset (36 applications) demonstrate the effectiveness of the attacking system.