Real-Time Power Analysis of Smart Sensors Using Advanced Debugging Methods

Area of Science:

• Electrical Engineering
• Computer Science
• Embedded Systems

Background:

• Accurate power consumption estimation is critical for embedded systems, especially those with limited power supplies like batteries or energy harvesting.
• Current methods include formal methods, power debugging tools on real hardware, and simulation-based estimations, each with limitations.
• Developing efficient power management strategies requires precise, reproducible power consumption data.

Purpose of the Study:

• To propose a novel method for estimating the power consumption of embedded systems and their firmware.
• To combine the benefits of existing power estimation techniques into a single, efficient approach.
• To enable reproducible, run-time power analysis without requiring dedicated power debugging hardware.

Main Methods:

• Utilizing a sensor-in-the-loop architecture enhanced with power estimation functionality.
• Developing a method that leverages recorded data for run-time analysis.
• Validating the proposed method against traditional power debugging hardware.

Main Results:

• The proposed method achieves an error rate of less than 1% compared to power debugging hardware.
• The technique allows for reproducible power consumption analysis using recorded data.
• The method eliminates the need for specialized power debugging hardware for estimation.

Conclusions:

• The novel sensor-in-the-loop based method offers a reliable and fast approach to estimate embedded system power consumption.
• This technique provides a valuable alternative for systems with constrained power budgets.
• The findings contribute to more efficient design and development of low-power embedded systems.