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

Abstract

Increased capabilities such as recognition and selfadaptability are now required from IoT applications. While IoT node power consumption is a major concern for these applications, cloud-based processing is becoming unsustainable due to continuous sensor or image data transmission over the wireless network. Thus optimized ML capabilities and data transfers should be integrated in the IoT node. Moreover, IoT applications are torn between sporadic data-logging and energy-hungry data processing (e.g. image classification). Thus, the versatility of the node is key in addressing this wide diversity of energy and processing needs. This paper presents SamurAI, a versatile IoT node bridging this gap in processing and in energy by leveraging two on-chip sub-systems: a low power, clock-less, event-driven Always-Responsive (AR) part and an energy-efficient On-Demand (OD) part. AR contains a 1.7MOPS event-driven, asynchronous Wake-up Controller (WuC) with a 207ns wake-up time optimized for sporadic computing, while OD combines a deep-sleep RISC-V CPU and 1.3TOPS/W Machine Learning (ML) for more complex tasks up to 36GOPS. This architecture partitioning achieves best in class versatility metrics such as peak performance to idle power ratio. On an applicative classification scenario, it demonstrates system power gains, up to 3.5x compared to cloud-based processing, and thus extended battery lifetime.

Green and Sustainable Industrial Internet of Things Systems Leveraging Wake-Up Radio to Enable On-Demand IoT Communication

“…So, we do not provide furether detail here. Nonetheless, this field of research is active, producing full AI-ready SoC with WuR capabilities, such as SamurAI [31], which consumes less than 40 nW in an idle mode.…”

Implementation of Edge Computing IoT-based Database Management System for Renewable Energy Management in Indonesia

“…By connecting a smart micro grid with IoT, the issue of voltage violation and grid instability caused by power insufficiency is resolved [25]. Additionally, the versatility of IoT nodes, with optimized machine learning capabilities and data transfers, addresses the diverse energy and processing needs of IoT applications [26]. Mobile Edge Computing (MEC) systems with energy harvesting equipment can provide data offloading services at the network edge, reducing end-toend latency and device energy consumption [27].…”