An Ultra-Low Energy Human Activity Recognition Accelerator for Wearable Health Applications

Human activity recognition (HAR) has recently received significant attention due to its wide range of applications in health and activity monitoring. The nature of these applications requires mobile or wearable devices with limited battery capacity. User surveys show that charging requirement is one of the leading reasons for abandoning these devices. Hence, practical solutions must offer ultra-low power capabilities that enable operation on harvested energy. To address this need, we present the first fully integrated custom hardware accelerator (HAR engine) that consumes 22.4 μJ per operation using a commercial 65 nm technology. We present a complete solution that integrates all steps of HAR, i.e., reading the raw sensor data, generating features, and activity classification using a deep neural network (DNN). It achieves 95% accuracy in recognizing 8 common human activities while providing three orders of magnitude higher energy efficiency compared to existing solutions.

[1]  Kenneth Meijer,et al.  Activity identification using body-mounted sensors—a review of classification techniques , 2009, Physiological measurement.

[2]  Martín Abadi,et al.  TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems , 2016, ArXiv.

[3]  Susmita Sur-Kolay,et al.  Wearable Medical Sensor-Based System Design: A Survey , 2017, IEEE Transactions on Multi-Scale Computing Systems.

[4]  Faicel Chamroukhi,et al.  Physical Human Activity Recognition Using Wearable Sensors , 2015, Sensors.

[5]  Paul J. M. Havinga,et al.  A Survey of Online Activity Recognition Using Mobile Phones , 2015, Sensors.

[6]  Paul Lukowicz,et al.  Collecting complex activity datasets in highly rich networked sensor environments , 2010, 2010 Seventh International Conference on Networked Sensing Systems (INSS).

[7]  Ümit Y. Ogras,et al.  Online Human Activity Recognition using Low-Power Wearable Devices , 2018, 2018 IEEE/ACM International Conference on Computer-Aided Design (ICCAD).

[8]  Yanqing Zhang,et al.  21.3 A 6.45μW self-powered IoT SoC with integrated energy-harvesting power management and ULP asymmetric radios , 2015, 2015 IEEE International Solid-State Circuits Conference - (ISSCC) Digest of Technical Papers.

[9]  Jochen Klucken,et al.  A clinical view on the development of technology‐based tools in managing Parkinson's disease , 2016, Movement disorders : official journal of the Movement Disorder Society.

[10]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[11]  Hyung Gyu Lee,et al.  REAP: Runtime Energy-Accuracy Optimization for Energy Harvesting IoT Devices , 2019, 2019 56th ACM/IEEE Design Automation Conference (DAC).

[12]  Reza Lotfian,et al.  An ultra-low power hardware accelerator architecture for wearable computers using dynamic time warping , 2013, 2013 Design, Automation & Test in Europe Conference & Exhibition (DATE).

[13]  Ian H. Witten,et al.  The WEKA data mining software: an update , 2009, SKDD.

[14]  P. Gasnier,et al.  Human-motion energy harvester for autonomous body area sensors , 2017 .

[15]  Iain A. Anderson,et al.  Stretch sensors for human body motion , 2014, Smart Structures.

[16]  Xin Liu,et al.  An Ultra-Low Power ECG Acquisition and Monitoring ASIC System for WBAN Applications , 2012, IEEE Journal on Emerging and Selected Topics in Circuits and Systems.

[17]  C. Van Hoof,et al.  A 1 cm3Modular Autonomous Sensor Node for Physical Activity Monitoring , 2006, 2006 Ph.D. Research in Microelectronics and Electronics.

[18]  Michelle A. Burack,et al.  Telehealth Management of Parkinson’s Disease Using Wearable Sensors: An Exploratory Study , 2017, Digital Biomarkers.

[19]  Raghuraman Krishnamoorthi,et al.  Quantizing deep convolutional networks for efficient inference: A whitepaper , 2018, ArXiv.

[20]  Jay Stewart,et al.  The American Time Use Survey , 2005 .

[21]  Ahmed Kattan,et al.  Better Physical Activity Classification using Smartphone Acceleration Sensor , 2014, Journal of Medical Systems.

[22]  Miguel A. Labrador,et al.  A Survey on Human Activity Recognition using Wearable Sensors , 2013, IEEE Communications Surveys & Tutorials.

[23]  Jaehyun Park,et al.  Near-optimal energy allocation for self-powered wearable systems , 2017, 2017 IEEE/ACM International Conference on Computer-Aided Design (ICCAD).

[24]  Baihua Li,et al.  Perception of Human Periodic Motion in Moving Light Displays-a Motion-based Frequency Domain Approach , 2003 .

[25]  Norbert Gyorbíró,et al.  An Activity Recognition System For Mobile Phones , 2009, Mob. Networks Appl..

[26]  K. Malmgren,et al.  Wearables in epilepsy and Parkinson's disease—A focus group study , 2018, Acta neurologica Scandinavica.

[27]  Hitesh Joshi,et al.  Flexible PV-cell Modeling for Energy Harvesting in Wearable IoT Applications , 2017, ACM Trans. Embed. Comput. Syst..

[28]  Chun-Huat Heng,et al.  A 93μW 11Mbps wireless vital signs monitoring SoC with 3-lead ECG, bio-impedance, and body temperature , 2017, 2017 IEEE Asian Solid-State Circuits Conference (A-SSCC).

[29]  Umit Y. Ogras,et al.  OpenHealth: Open-Source Platform for Wearable Health Monitoring , 2019, IEEE Design & Test.

[30]  Tatsuo Nakajima,et al.  Feature Selection and Activity Recognition from Wearable Sensors , 2006, UCS.

[31]  Thomas Plötz,et al.  Deep, Convolutional, and Recurrent Models for Human Activity Recognition Using Wearables , 2016, IJCAI.

[32]  Daniel Roggen,et al.  Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition , 2016, Sensors.

[33]  Luis González Abril,et al.  Low Energy Physical Activity Recognition System on Smartphones , 2015, Sensors.

[34]  Yufei Chen,et al.  Performance Analysis of Smartphone-Sensor Behavior for Human Activity Recognition , 2017, IEEE Access.

[35]  Max A. Little,et al.  Feasibility of large-scale deployment of multiple wearable sensors in Parkinson's disease , 2017, PloS one.

[36]  M. Haghi,et al.  Wearable Devices in Medical Internet of Things: Scientific Research and Commercially Available Devices , 2017, Healthcare informatics research.

[37]  Paul J. M. Havinga,et al.  Activity Recognition Using Inertial Sensing for Healthcare, Wellbeing and Sports Applications: A Survey , 2010, ARCS Workshops.

[38]  Refet Firat Yazicioglu,et al.  18.3 A multi-parameter signal-acquisition SoC for connected personal health applications , 2014, 2014 IEEE International Solid-State Circuits Conference Digest of Technical Papers (ISSCC).

[39]  Morwenna Kirwan,et al.  Using Smartphone Technology to Monitor Physical Activity in the 10,000 Steps Program: A Matched Case–Control Trial , 2012, Journal of medical Internet research.

[40]  Nicholas D. Gilson,et al.  Measuring and Influencing Physical Activity with Smartphone Technology: A Systematic Review , 2014, Sports Medicine.

[41]  Gary M. Weiss,et al.  Activity recognition using cell phone accelerometers , 2011, SKDD.