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 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 into 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 article presents SamurAI, a versatile IoT node bridging this gap in processing and 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 1.7-MOPS event-driven, asynchronous Wake-up Controller (WuC) with a 207-ns wake-up time optimized for sporadic computing, while OD combines a deep-sleep RISC-V CPU and 1.3-TOPS/W machine learning (ML) for more complex tasks up to 36 GOPS. This architecture partitioning achieves the 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 $3.5\times $ , compared to cloud-based processing, and, thus, extended battery lifetime.
In the context of highly data-centric applications, close reconciliation of computation and storage should significantly reduce the energy-consuming process of data movement. This paper proposes a Computational SRAM (CSRAM) combining In-and Near-Memory Computing (IMC/NMC) approaches to be used by a scalar processor as an energy-efficient vector processing unit. Parallel computing is thus performed on vectorized integer data on large words using usual logic and arithmetic operators. Furthermore, multiple rows can be advantageously activated simultaneously to increase this parallelism. The proposed C-SRAM is designed with a two-port pushed-rule foundry bitcell, available in most existing design platforms, and an adjustable form factor to facilitate physical implementation in a SoC. The 4kB C-SRAM testchip of 128-bit words manufactured in 22nm FD-SOI process technology displays a sub-array efficiency of 72% as well as an additional computing area of less than 5%. The measurements averaged on 10 dies at 0.85V and 1GHz demonstrate an energy efficiency per unit area of 35.6 and 1.48TOPS/W/mm2 for 8-bit additions and multiplications with 3ns and 24ns computing latency, respectively. Compared to a 128-bit SIMD processor architecture, up to 2x energy reduction and 1.8x speed-up gains are achievable for a representative set of highly data-centric application kernels.
18 February 2019 in IEEE Solid-State Circuits Letters
This paper presents a single-rail two-port static random-access memory (SRAM) designed in 28 nm FD-SOI technology specifically for a synchronous/asynchronous IoT node. This SRAM supports an asynchronous interface communication and a fast transition sleep/active mode. It enables simultaneous synchronous and asynchronous accesses on its two independent ports, as well as a selective virtual ground (SVGND) to support ultra-low voltage (ULV) in read operations. Measurements on a 64 kbits memory macro prove the functionality for a supply voltage ranging from 0.25 V to 1.25 V. The sleep mode enables 158x reduction of the SVGND static-power consumption. The SVGND read assist achieves a gain of 50 mV on the read port operations lowest voltage. The memory achieves, at 0.25 V and 27∘C, an average energy cost per access of 1.45 fJ/bit at 451 kHz and a leakage of 25 pW/bit in sleep mode.
Due to low activity in Internet of Things (IoT) applications, systems tend to leverage low power modes in order to reduce their power consumption. Normally-off computing thus arose, consisting in having turned off most part of a system’s power supply, while dynamically turning on components as the application needs it. As wake up sources may be diverse, simple controllers are integrated to handle smart wake up schemes. Therefore, to prevent overconsumption while transitioning to running mode, fast wake up sequences are required. An asynchronous 16-bit Reduced Instruction Set Computer (RISC) Wake-up Controller (WuC) is proposed demonstrating 50.5 ns@9.2 Million Instructions Per Second (MIPS)@0.6 V wake-up latency, drastically reducing the overall wake-up energy of IoT systems. A clockless implementation of the controller saves the booting time and the power consumption of a clock generator, while providing high robustness to environmental variations such as supply voltage level. The WuC is also able to run simple tasks with a reduced Instruction Set Architecture (ISA) and achieves as low as 11.2 pJ/inst @0.5 V in Fully Depleted Silicon On Insulator (FDSOI) 28 nm.
Reinforcement learning (RL) has received much attention in recent years due to its adaptability to unpredictable events such as harvested energy and workload, especially in the context of edge computing for Internet-of-Things (IoT) nodes. Due to limited resources in IoT nodes, it is difficult to achieve self-adaptability. This paper studies online reactivity issues of fixed learning rate in the linear actor-critic (LAC) algorithm for transmission duty-cycle control. We propose the LAC-AB algorithm that introduces into the LAC algorithm an adaptive learning rate called Adam for actor update to achieve better adaptability. We introduce a definition of “convergence” when quantitative analysis of convergence is performed. Simulation results using real-life one-year solar irradiance data indicate that, unlike the conventional setups of two decay rate of Adam, smaller such as 0.2–0.4 are suitable for power-failure-sensitive applications and 0.5–0.7 for latency-sensitive applications with . LAC-AB improves the time of reactivity by 68.5–88.1% in our application; it also fine-tunes the initial learning rate for the initial state and improves the time of fine-tuning by 78.2–84.3%, compared to the LAC. Besides, the number of power failures is drastically reduced to zero or a few occurrences over 300 simulations.