JSSC 2023第6期RF & Wireless

SamurAI A Versatile IoT Node With Event-Driven Wake-Up and Embedded ML Accelerat

SamurAI是一款多功能物联网节点，结合低功耗事件驱动和嵌入式ML加速，满足多样化能耗和处理需求。

1.7-MOPS, 207-ns wake-up

物联网节点事件驱动机器学习加速低功耗多功能设计

▸创新点1：低功耗无时钟事件驱动唤醒控制器（WuC）采用异步电路设计，实现207纳秒快速唤醒和1.7 MOPS处理能力，显著降低待机功耗（系统创新）

▸创新点2：高效按需处理子系统（OD）集成ML加速器，通过动态电压频率调节（DVFS）技术实现能效比优化，支持图像分类等复杂任务（电路创新）

▸创新点3：双模式架构创新（AR+OD）通过硬件任务分区满足多样化需求，AR处理事件触发任务（1μW级功耗），OD处理计算密集型任务（方法创新）

▸创新点4：嵌入式ML加速器采用稀疏化计算和权重压缩技术，在IoT节点本地实现高能效比的数据处理，减少云端传输需求（电路创新）

Abstract

Increased capabilities, such as recognition and self- adaptability, are now required from Internet-of-Things (IoT) applications. While IoT node power consumption is a major con- cern for these applications, cloud-based processing is becoming unsustainable due to continuous sensor or image data transmis- sion over the wireless network. Thus, optimized ML capabilities and data transfers should be integrated into the IoT node. More- over, IoT applications are torn between sporadic data-logging and