Analog-to-Information Conversion with Random Interval Integration

Area of Science:

• Signal Processing
• Information Theory
• Sensor Technology

Background:

• Analog-to-information conversion is crucial for digital signal processing.
• Compressed sensing requires efficient methods for acquiring complex signals like ECG and environmental data.
• Existing methods like random sampling and demodulation have limitations in hardware complexity and performance.

Purpose of the Study:

• To introduce and analyze a novel analog-to-information conversion method: random interval integration.
• To evaluate its applicability for compressed sensing of multi-harmonic signals from various sensors.
• To compare its performance against established analog-to-information conversion techniques.

Main Methods:

• Input signal integration using a randomly resettable integrator prior to analog-to-digital conversion.
• Random sequence generator controlling the integrator's reset.
• Signal reconstruction via a state-of-the-art algorithm minimizing measurement vector distance norm.

Main Results:

• Random interval integration demonstrates superior performance compared to other single-channel architectures.
• The method achieves high performance, comparable to more complex random modulation pre-integrators in specific scenarios.
• Evaluations using ECG, multi-sine, and environmental signals validate its effectiveness.

Conclusions:

• Random interval integration offers a simple yet effective approach for analog-to-information conversion.
• Its straightforward hardware implementation makes it suitable for various sensor applications.
• The method presents a promising alternative for compressed sensing of quasiperiodic and aperiodic signals.