November 8, 2018
Editor’s Note: This content is contributed by Wes Skeffington, Principal Architect - Industrial & Medical
Xilinx and AWS have been collaborating to bring AWS Greengrass to Zynq® UltraScale+™ MPSoC devices and Amazon FreeRTOS (a:FreeRTOS) to Zynq-7000 devices, helping developers accelerate Industrial IoT (IIoT) solutions. The combination of Xilinx’s scalable, secure, and adaptable hardware platform technologies combined with AWS technologies for secure connectivity, cloud-based system management, and a rich portfolio of AWS IoT Cloud Services provides a foundation for IIoT developers to build upon.
IIoT is not about simply sending all data and intelligence to the cloud. Realistic IIoT solutions must assume a distributed intelligence model in which the edge and cloud collaborate over intermittent connectivity. The AWS and Xilinx integration provides an ability for realizing this distributed intelligence through an intelligent edge that can implement mixed criticality applications—isolating high-speed deterministic industrial control functions in FPGA fabric and real-time processors while IIoT functions are deployed in application processors. Further flexibility can be added to application processors by partitioning control and IoT functions using hypervisor technology such as Xen or Jailhouse. More on the topic of hypervisors in IIoT solutions will be discussed in a later blog post – the initial release discussed here is without the hypervisor technology included.
To help developers get started with the combination of Xilinx and AWS technologies, the following hardware/software platforms have been made available:
Xilinx has fully qualified Amazon FreeRTOS (a:FreeRTOS) on the Zynq-7000 Avnet MicroZed platform. a:FreeRTOS offers a superset of standard FreeRTOS features by adding secure connectivity, device synchronization, and over-the-air (OTA) update via AWS Cloud Services. The Zynq-7000 SoC platform represents many small footprint controller and intelligent I/O modules in the industrial space, many of which run the standard FreeRTOS operating system today. The addition of a:FreeRTOS libraries for incrementally bringing these nodes into the context of an IIoT solution is a logical evolution for these edge nodes. Get started with Zynq-7000 and a:FreeRTOS by following the “Getting Started” guide at https://aws.amazon.com/freertos/getting-started/.
Xilinx has developed and tested two Amazon Greengrass compatible devices, the Avnet Ultra96 platform and the Xilinx ZCU104 development board. Both are based on the Zynq UltraScale+ MPSoC, which has Arm® Cortex™-A53 and Cortex-R5 processors, flexible programmable logic, and a rich set of platform security functions for supporting edge applications. The Zynq UltraScale+ MPSoC family represents many mid-range industrial controller and IoT gateway solutions today. The ability with Greengrass to deploy higher level language (Python and Java) IIoT applications as long-lived AWS Lambda functions with secure cloud connectivity and tools for deploying and maintaining them allows IIoT application developers to focus on their applications and points of differentiation.
Greengrass compatibility for Zynq UltraScale+ MPSoCs is built on the Xilinx PYNQ framework. Linux images are available on the pynq.io website. Get started with AWS Greengrass by first following the instructions for setting up a Zynq UltraScale+ MPSoC PYNQ platform at these links:
Once the base PYNQ Linux image is installed, the user should follow the AWS Greengrass Getting Started Guide.
AWS and Xilinx will be hosting a hands-on workshop showing how these technologies come together in a distributed industrial control application with integrated edge machine learning inference at AWS re:Invent on Wednesday, 28 November. Look for session “GPSWS409 - Integrate the AWS Cloud with Responsive Xilinx Machine Learning at the Edge” for exact time and location. Hope to see you at the workshop where you will learn how Greengrass on Ultra96 and a:FreeRTOS on MicroZed enable an adaptable and intelligent edge for real-life and real-time industrial applications!