Self-Powered Long-Life Microsystem for Vibration Sensing and Target Recognition.

Area of Science:

• Microsystems Engineering
• Internet of Things (IoT)
• Embedded Systems

Background:

• Microsystems are crucial for unattended Internet of Things (IoT) applications requiring small, lightweight, and long-lasting devices for covert, large-scale, and long-term deployment.
• Existing microsystems often face limitations in power autonomy and continuous operation for prolonged target detection and recognition tasks.

Purpose of the Study:

• To develop a novel low-power, long-life microsystem for unattended, long-term target perception and recognition.
• To integrate self-power supply, event wake-up, continuous vibration sensing, and target recognition functionalities into a compact microsystem.

Main Methods:

• Designed a composite energy source combining solar energy and battery for self-powering.
• Optimized sensing, circuit, signal processing, and transceiver modules for reduced size and power consumption.
• Developed and implemented a low-computational support vector machine (SVM) recognition algorithm within the microsystem.

Main Results:

• The 15 cm³, 35 g microsystem achieved target recognition (pedestrian, wheeled/tracked vehicles) with >84% (indoor) and >65% (outdoor) accuracy.
• Achieved self-powering up to 22.7 mW under midday sunlight.
• Demonstrated that 11 minutes of self-powering can sustain 24 hours of microsystem operation in sleep mode.

Conclusions:

• The developed microsystem offers a viable solution for unattended, long-term target perception and recognition in IoT applications.
• The integration of self-powering, event-driven operation, and efficient recognition algorithms enables extended autonomous deployment.
• The system's small form factor and low power consumption make it suitable for covert and large-scale distribution.