Memory Strategies International

Semiconductor Memory Services

Purchase Report

IoT Applications for Embedded NVM, May 2019

The Internet of Things (IoT) extends from “the Cloud” where high end servers store and process large amounts of data acquired from the worldwide IoT network. Below the Cloud is the “Fog” where Ethernet systems, mobile communication systems and other communications networks connect large numbers of users to the cloud. Below the Fog are small gateways where Wired LAN and WiFi exist. Below that are the large numbers of embedded edge devices using Systems on Chips (SoC) and other processors that connect to sensors that obtain data from the physical world. These edge device systems tend have ultralow power requirements since they are powered by either batteries or energy harvesting systems and operate primarily in standby mode with brief rapid wake-ups.

Sensors include devices for: motion detection, strain sensors, light sensors, noise sensors, and many other sensor types. Sensor nodes are frequently wireless requiring receive and transmit capability for long periods of time with minimal energy expenditure. Emerging NVM have been used in such applications. In many cases, as ultra low power unified memories. Repeated memory reads can be achieved at very low power using single-line-cache architectures that reduce the read activity factor. Near-memory-computing and In-memory computing are both used in IoT edge systems to couple high bandwidth with ultralow power. Specifications for low end IoT devices are considered for the RAM and Flash used to support real time operating systems. The use of nonvolatile logic which saves the systems state in flipflops coupled with emerging NVM upon shutdown was considered for systems powered by harvested energy. An HfO2 FeFET capacitor in an nvSRAM was found capable of storing and recalling the previous memory state before power shutdown and therefore suitable for intermittent operation of IoT devices with deep sleep mode.

STT-MRAM, Spin hall effect devices, Ferroelectric memories, ReRAMs and Atom Switches are considered for use in IoT edge systems along with negative capacitance SRAMs. nvSRAMs made of SRAMs integrated with NVM are useful in normally off and ultra low power Iot applications. For Logic-in-Memory (LIM), or compute-in-memory (CIM) arithmetic and logic operations can be processed in the memory efficiently using minimal power. FPGA platforms can be used to evaluate ultra low power SoC for IoT by modeling a complex SoC including processors, memories and peripherals where every signal is reachable. A 2T gain cell embedded DRAM was considered as an alternative to eDRAM in an IoT chip. Since the transistor gain cell is fully logic compatible it has smaller area and lower retention power than SRAM without requiring refresh. Security is an issue for IoT sensor networks resulting in the use of neural net systems for lightweight crypto engines.

IoT Applications for Embedded NVM, May 2019

Table of Contents

1.0 IoT Applications for Embedded Memory

• 1.1 The IoT System Hierarchy from Edge to Cloud
• 1.2 A Non-Volatile Asynchronous Logic-in-Memory for Ultra-Low Power Applications
• 1.3 16MHz FRAM MCU in Sub-1uA Strain Sensor for ULP Motion Detection
• 1.4 A 3.6 Mb Embedded NV ReRAM Macro in 22 nm FinFET Technology
• 1.5 A Double Pumped Single-Line-Cache SRAM Architecture for Low Energy IoT
• 1.6 Using Unified NVM SoC for IoT-Wireless Sensor Node Applications
• 1.7 Computing-in-Memory with 3D Negative Capacitance-FinFET for Intelligent IoT
• 1.8 A 14 ns write 128Mb eSTT-MRAM for low power MCU IoT Chips
• 1.9 Survey of Low to High End IoT Devices
• 1.10 Energy Efficient and Variation Tolerant NV Logic Design for Harvested Systems
• 1.11 Data Aggregation Algorithm in Flash for IoT Based Power Grid Storage System
• 1.12 Energy Efficient Edge Computing with Embedded HfO2 Ferroelectric Devices.
• 1.13 FPGA-Based Platform to Evaluate Ultra Low Power SoC for IoT
• 1.14 2T Gain Cell eDRAM for Ultra-Low Power IoT Applications in 28 nm DF-SOI.
• 1.15 Factors in Dynamic Allocation of Heterogeneous Memories in IoT
• 1.16 eSTT-MRAM in 28 nm FDSOI Logic Process for Industrial Temperature
• 1.17 nvSRAM Using Ferroelectric HfO2 capacitor for Ultralow Power IoT
• 1.18 Energy Efficient Adiabatic FRAM for IoT Applications
• 1.19 Embedded STT-MRAM Macro to Replace eFlash for IoT Chips
• 1.20 Ultra-Low Power Atom Switch (RRAM) for Use in NV SoC
• 1.21 Low Power SONOS eNVM in 40 nm CMOS Logic for Low Power IoT Applications
• 1.22 Power Efficiency of Edge Sensors for Health Monitoring IoT Applications
• 1.23 Self Powered Intelligent Secure IoT Edge Mote with Ultra Low Power SoC
• 1.24 Overview of pMTJ STT-MRAM for Use in Secure IoT Systems
• 1.25 2 Mb eRRAM Macro in 65 nm CMOS Logic process
• 1.26 Ferroelectric HfO2 MIM Capacitor NV SRAM for Normally-Off Devices
• 1.27 Low Power Cu Atom Switch Programmable Logic Made in 40 nm CMOS

2.0 Firmware for Running an IoT System

• 2.1 WebletScript - A JavaScript Engine for IoT
• 2.2 Blockchain Verification Requiring only KBytes of RAM
  

Bibliography

Back to Top

To Order "IoT Applications for Embedded NVM, May 2019"

Select:

For more information, please Contact Memory Strategies