uMEC (aka MicroMEC, or Micro-MEC) is a project under the Akraino Edge umbrella. SUSE joined in 2019 and together with partners we managed to create the 1st prototype infrastructure.
The concept has been successfully used on a few hackathons last year:
A few weeks ago we sat down with Nokia and discussed how could we reshape the existing code and documentation. We agreed to relocate the code to github and properly document the architecture and the installation. This work will already highlight the architecture changes we are planning to accommodate different edge computing units (like Nvidia Jetson Nano) and more sensors in general. The uMEC APIs will also be redesigned and the results of the work will hopefully be presented on a couple of conferences and hackathons this year.
I would like to focus on the installation and architecture documentation. The setup we have been using throughout 2019 included quite a few pieces:
The prototype edge units were based on Raspberry Pi 3B+ and 4 computers. The RPis ran openSUSE TW (64 bit) and Raspbian.
The uMEC APIs were deployed in containers and orchestrated by k3s. Along the uMEC APIs we also had a hosted Tensorflow service where we deployed models for simple object detection.
On the "server side" we had k8s and OpenFaaS Cloud; OFC. With OFC we offered an easy to use development environment for quick prototyping using the uMEC APIs.
This year we will setup a proper CI / CD pipeline using github. The deployment will take place on hardware that is hosted at Metropolia University where the students are also working on uMEC use cases.
Our goal is to enable the setup and creation of uMEC environments by anybody. As the hardware and software is all available, we just need to document these properly. I hope that by the end of the hack week uMEC will be in a much better shape.
Please join, if you are interested in 5G Edge computing and would like to learn more about industrial use cases in the area.
This project is part of:
Hack Week 19