Revealing Embedded System Behaviors: A Comparative Analysis of Power Consumption and Hardware Performance Counters

As computing evolves, analyzing system behavior has become as critical as optimizing performance and energy efficiency. Traditional methods of behavioral analysis often rely on software-level profiling or predefined signatures, which limit their applicability in dynamic and resource-constrained environments. This study takes a novel approach by identifying and characterizing system workloads based on distinct behavioral patterns, such as memory usage, arithmetic operations, and control flow. By leveraging hardware performance counters (HPCs) and power consumption metrics, this work demonstrates how hardware-level insights can reveal unique operational signatures. These findings highlight the potential of HPCs and power consumption to enhance the understanding of system behaviors, offering practical solutions for behavior identification that could be instrumental in detecting vulnerabilities or anomalies. The results underscore both the opportunities and challenges of using these metrics for behavioral analysis in embedded systems.