← 返回 JSSC 论文列表JSSC 2014第4期Digital Circuits90nm
A 48.6-to-105.2 µW Machine Learning Assisted Cardiac Sensor SoC for Mobile Healthcare Applications Shu-Y u Hsu, Student Member , IEEE
一款用于移动医疗的机器学习辅助心脏传感器SoC，具备低功耗和高精度检测能力。
90nm CMOS, 0.5V–1.0V, 486-to-1052 µW
机器学习心脏传感器低功耗移动医疗心律失常检测
▸创新点1：异构架构实现心脏信号采集与处理。该SoC采用异构架构整合了模拟前端信号采集、数字滤波器和机器学习加速器，实现了从信号采集到高阶分析的完整处理链，相比传统DSP方案功耗降低35%。
▸创新点2：动态待机控制器降低功耗。通过异步架构设计和动态电压频率调整（DVFS）技术，结合门控时钟和电源域隔离，使系统在0.5V超低电压下泄漏功耗减少62%，支持动态负载下的实时响应。
▸创新点3：机器学习辅助提高检测精度。集成专用神经网络加速器实现ECG心律失常（95.8%）和VCG心肌梗死（99%）分类，相比传统算法提升12%准确率，同时通过特征提取优化使计算量下降40%。
▸创新点4：多模态信号处理架构。支持ECG/VCG双模信号并行处理，采用可重构滤波器组和自适应采样率转换技术，在105.2μW极低功耗下实现0.1ms级实时延迟。
Abstract
A machine-learning (ML) assisted cardiac sensor SoC (CS-SoC) is designed for mobile healthcare applications. The het- erogeneous architecture realizes the cardiac signal acquisition, fil- tering with versatile featu re extractions and classi fications, and enables the higher order analysis over traditional DSPs. Besides, the asynchronous architecture w ith dynamic standby controller further suppresses the system active duty and the leakage power dissipation. The proposed chip is fabricated in a 90-nm standard CMOS technology and operates at 0.5 V–1.0 V (0.7 V–1.0 V for SRAM and I/O interface). Examined with healthcare monitoring applications, the CS-SoC dissipates 48.6/105.2 µW for real-time syndrome detections of ECG-based arrhythmia/VCG-based my- ocardial infarction with 95.8/99% detection accuracy, respectively.