Using Floating-Gate MOS as a Non-Volatile Analog Memory for Energy-Efficient Adaptive Thresholding in ECG Sensors

We present a floating-gate metal-oxide-semiconductor (FGMOS) device in a 65 nm standard CMOS technology as a non-volatile analog memory element for energy-efficient adaptive thresholding in an electrocardiogram (ECG) sensor. We present the write and erase characteristics of the FGMOS obtained through measurements on fabricated pMOS-type FG elements. An adaptive thresholding circuitry that generates, stores, and dynamically updates a threshold value on a FGMOS element to identify the R-waves in an ECG recording is demonstrated. The proposed circuitry uses the outputs from a Pan-Tompkins-algorithm-based ECG processor that accentuates the R-waves of a 50 s ECG recording from the MIT-BIH arrhythmia database. Circuit simulations of the proposed adaptive thresholding circuitry and the measured characteristics of the analog memory element demonstrate the similarity between the adaptive thresholds from the proposed circuitry and the digital implementation of the original Pan-Tompkins algorithm. The proposed circuitry consumes 19 nW.