CSR: Small: Development of Distributed Neural Processing Electronics for Whole-Body Computing and Biomedical Sensor Fusion

CSR：小型：用于全身计算和生物医学传感器融合的分布式神经处理电子设备的开发

• 批准号: 1816870
• 负责人: Jie Gu
• 金额: $ 50万
• 依托单位: Northwestern University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-10-01 至 2023-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1816870&HistoricalAwards=false
• 关键词: CSR; Small; Development; Distributed; Neural

Wearable electronics and health assistive devices have become commonly used for daily activity tracking and medical care. However, the existing devices can no longer satisfy the growing demand on computing power and energy efficiency especially when a large number of sensors are utilized to improve the performance of such devices. This work will develop a new generation of wearable biomedical devices that utilize the emerging computing architecture and state-of-art networking techniques to boost the data processing capability of such devices. The developed techniques will be demonstrated in the state-of-art prostheses for rehabilitation application. Several key intellectual merits are delivered. First of all, this project will develop novel distributed neuromorphic computing architecture featuring a scalable, reconfigurable, and multi-chip neural network system for sensor fusion enabled biomedical devices. Heterogeneous sensor data from physiological signals will be efficiently processed through reconfigurable neural processors. Moreover, to form a highly efficient network and remove routing congestions, body channel communication will be developed and integrated with the neural processors to achieve a high-bandwidth, low-latency, inter-connected network around the human body. The work brings significant advancement and benefits to modern biomedical devices by creating highly efficient sensing, computing and networking solutions. The proposed development promotes technology fusion across fields of computer, electrical, biomedical engineering. The sub-tasks of the project will be delivered to undergraduate and graduate classrooms for hands-on experience of integrated circuits design, biomedical instrumentation, machine learning techniques. New courses on computing techniques for sensor fusion enabled biomedical system will be created to disseminate the learned knowledge to broader audience. All publications, and experimental data, source codes, design files from this project will be retained on institutional servers for 3 years after the completion of the project. All publications along with important data, codes, and supporting design files will be made accessible through PI's data repository, http://nu-vlsi.eecs.northwestern.edu/data_repository.html.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

可穿戴电子产品和健康辅助设备已普遍用于日常活动跟踪和医疗护理。 然而，现有设备已无法满足对计算能力和能源效率不断增长的需求，特别是当利用大量传感器来提高此类设备的性能时。这项工作将开发新一代可穿戴生物医学设备，利用新兴的计算架构和最先进的网络技术来提高此类设备的数据处理能力。所开发的技术将在用于康复应用的最先进的假肢中得到展示。 提供了几个关键的智力优点。 首先，该项目将开发新颖的分布式神经形态计算架构，该架构具有可扩展、可重新配置和多芯片神经网络系统，用于支持传感器融合的生物医学设备。 来自生理信号的异构传感器数据将通过可重构神经处理器进行有效处理。 此外，为了形成高效的网络并消除路由拥塞，将开发身体通道通信并与神经处理器集成，以实现围绕人体的高带宽、低延迟、互连的网络。 这项工作通过创建高效的传感、计算和网络解决方案，为现代生物医学设备带来了重大进步和好处。 拟议的开发促进了计算机、电气、生物医学工程领域的技术融合。该项目的子任务将交付给本科生和研究生课堂，以获取集成电路设计、生物医学仪器、机器学习技术的实践经验。 将创建有关传感器融合生物医学系统计算技术的新课程，以将学到的知识传播给更广泛的受众。该项目的所有出版物、实验数据、源代码、设计文件将在项目完成后在机构服务器上保留3年。 所有出版物以及重要数据、代码和支持设计文件都可以通过 PI 的数据存储库访问，http://nu-vlsi.eecs.northwestern.edu/data_repository.html。该奖项反映了 NSF 的法定使命，并被视为值得通过使用基金会的智力优点和更广泛的影响审查标准进行评估来支持。

项目成果

A Wearable Bio-signal Processing System with Ultra-low-power SoC and Collaborative Neural Network Classifier for Low Dimensional Data Communication

具有超低功耗 SoC 和用于低维数据通信的协作神经网络分类器的可穿戴生物信号处理系统

• DOI: 10.1109/embc44109.2020.9176647
• 发表时间: 2020-07
• 期刊: Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC
• 作者: Wei, Yijie;Cao, Qiankai;Hargrove, Levi;Gu, Jie
• 通讯作者: Gu, Jie

A 65nm Implantable Gesture Classification SoC for Rehabilitation with Enhanced Data Compression and Encoding for Robust Neural Network Operation Under Wireless Power Condition

用于康复的 65 纳米植入式手势分类 SoC，具有增强的数据压缩和编码功能，可在无线供电条件下实现稳健的神经网络运行

• DOI: 10.1109/cicc53496.2022.9772838
• 发表时间: 2022-04
• 期刊: Custom Integrated Circuit Conference
• 作者: Wei, Yijie;Chen, Xi;Gu, Jie
• 通讯作者: Gu, Jie

A Fully-integrated Gesture and Gait Processing SoC for Rehabilitation with ADC-less Mixed-signal Feature Extraction and Deep Neural Network for Classification and Online Training

用于康复的完全集成手势和步态处理 SoC，具有无 ADC 混合信号特征提取和用于分类和在线训练的深度神经网络

• DOI: 10.1109/cicc48029.2020.9075910
• 发表时间: 2020-03
• 期刊: Custom Integrated Circuits Conference
• 作者: Wei, Yijie;Cao, Qiankai;Otseidu, Kofi;Hargrove, Levi;Gu, Jie
• 通讯作者: Gu, Jie

Human Activity Recognition SoC for AR/VR with Integrated Neural Sensing, AI Classifier and Chained Infrared Communication for Multi-chip Collaboration

用于 AR/VR 的人体活动识别 SoC，具有集成神经传感、AI 分类器和用于多芯片协作的链式红外通信

• DOI: 10.23919/vlsitechnologyandcir57934.2023.10185392
• 发表时间: 2023-06
• 期刊: Symposium on VLSI Technology and Circuits
• 作者: Wei, Yijie;Chen, Xi;Gu, Jie
• 通讯作者: Gu, Jie

Exploration of Design Space and Runtime Optimization for Affective Computing in Machine Learning Empowered Ultra-Low Power SoC

机器学习赋能超低功耗SoC中情感计算的设计空间和运行时优化探索

• DOI: 10.1109/dac18072.2020.9218583
• 发表时间: 2020-01
• 期刊: Design Automation Conference
• 作者: Wei, Yijie;Otseidu, Kofi;Gu, Jie
• 通讯作者: Gu, Jie