A low-power bio-processor for wearable healthcare platform

Abstract

A low power ExG (ECG, EEG, EOG, and EMG) signal processor is proposed for compact the health monitoring system. To effectively manage the 4 kinds of vital signs from at least 16 sensor nodes attached on a human body, 3 low power schemes are adopted in the proposed processor: the Adaptive Buffer-triggered Clock Controller, a CSD-coefficient FIR filter, and the tag-shared QLV compressor. The adaptive Buffer-triggered Clock Controller reduces power consumption by buffer-triggered clock gating which enables the controller and the data path if and only if they are requested by each ExG signal. In addition, the two accelerators, a CSD-coefficient FIR filter and a tag-shared QLV compressor, are integrated to decrease energy consumption by only 1/80 compared to the CPU with the simplified hardware and the reduced number of complex operations. Especially in the proposed compressor, the 13\% less memory capacity is required than the previous work since it shares the duplicated QLV tags so that data transaction overhead through the external interface can be further released. By using these proposed 3 schemes, the proposed 1.7x1.8 mm$^{2}$ ExG signal processor chip implemented in a 0.18$\mu$m CMOS process consumes only 30$\mu$W with 1.5V supply in the compact health monitoring system.