A low-power non-uniform sampling system for energy-efficient data acquisition utilizing cubic spline interpolation

In this work, we introduce a third-derivative-based sampling (TDS) approach, a non-uniform sampling technique designed to improve the power efficiency of data acquisition systems across a wide range of applications. The proposed TDS method employs well-known cubic spline interpolation for both signal sampling and reconstruction. By monitoring the rate of change of the third derivative of the input signal, TDS identifies regions of high signal activity and selectively triggers the analog-to-digital converter (ADC) to acquire only the most informative samples. The trade-off between reconstruction accuracy and compression factor (CF) can be adjusted by tuning the maximum number of consecutive samples that can be skipped (N) and/or the threshold applied to the rate of change of the signal’s third derivative. Fabricated in a 65-nm CMOS technology and evaluated using a variety of real-world signals, the proposed TDS system draws a maximum current of 9 μA from a 1-V supply and achieves CFs of up to 12 for electrocardiogram (ECG) signals and 14.9 for photoplethysmography (PPG) signals, while maintaining reconstruction quality comparable to state-of-the-art non-uniform sampling techniques. These CFs correspond to reductions of 91.67% and 93.29%, respectively, in the number of ADC conversions compared with conventional Nyquist-rate sampling. Since the TDS circuitry consumes only ~9 μW, the reduction in ADC conversions can substantially reduce the energy associated with data conversion and depending on system implementation, subsequent storage, processing, and wireless transmission. The proposed approach therefore provides a low-power mechanism for reducing sampling activity and data volume in power-constrained sensing systems, particularly wearable and biomedical applications.