This is true of applications which require very low latency, and cannot afford to wait for data and signals to be sent to a distant switch or cloud; and those which involve huge numbers of devices, all transmitting messages and data. These would put excessive strain on the backhaul link if every one were sent to a centralised location.
When these industrial, low latency applications are also highly mobile, it is even more important that communication links are very short and fast. Examples include cars (self-driving or not), high speed trains or city vehicles talking to roadside and trackside infrastructure. That can support many on-board services including safety-critical ones, and even allow city or transport operators to build their own localised networks before 5G has been deployed.
There have been significant contributions from the UK to these kind of projects. UK start-ups like Quortus and Virtuosys have helped define and test how edge compute platforms work alongside 5G and virtualised networks to support new use cases.
For instance, Angel Trains, a railway rolling stock leasing company, recently demonstrated edge computing over a wireless mesh network, using the Edge Application Platform from Virtuosys. Angel owns about one-third of the UK’s railway rolling stock and its pilot is being funded by the Department for Transport (DfT).
Virtuosys’ approach is to install a Linux-based mobile edge server supporting multiple connectivity options as well as management and authentication software that can be run in public or private cloud environments. Network resources are distributed flexibly according to requirements and traffic patterns, across the mesh, and far smaller amounts of traffic need to reach the main mobile network and core, improving performance. Applications will include monitoring and preventative maintenance for trains and other equipment as well as on-board services for passengers, staff and freight customers.
The pilot is running until October 2018 and then Angel Trains will hope to roll it out in earnest.
The combination of wireless and edge compute becomes even more powerful when there is an open programming environment to make it easier to devise new applications for different vertical markets. One candidate is Amazon’s Greengrass platform for edge-oriented IoT developments, which has been moving closer to mobile edge environments.
The most prominent of these among MNOs is the ETSI-defined Multi-access Edge Compute, although some service providers are more interested in alternatives emerging from the IT industry, like OpenFog. MEC has received a boost from joint efforts between Amazon AWS Greengrass and two MEC backers, Nokia and Saguna Networks.
Last autumn, Amazon AWS announced an alliance with Nokia with three key objectives:
- Nokia will support service providers in their AWS implementation strategy with consulting, design, integration, migration and operations services.
- Nokia and AWS will work together to generate new 5G and edge cloud strategies and guidance for customers including reference architectures.
- The partners will work to commercialize IoT use cases based around combinations of AWS Greengrass, Amazon Machine Learning, Nokia MEC and Nokia’s IMPACT IoT platform.