Computing on the Edge – Redefining Mobility

Multi-access Edge Computing (MEC) is about creating an IT ecosystem for applications near the perimeter of the network. Andy Odgers, CEO of Quortus, explains that the network itself is accessible via multiple wireless technologies, such as 3G, 4G, WiFi and fixed-wireless. Odgers provided an update of the ETSI MEC process at the 2016 MEC Focus Day in conjunction with the 2016 Carrier Network Virtualization conference.

The ETSI MEC standardization process started two years ago and was recently renewed for another two years. As part of that process, the meaning of the “M” in the acronym was changed from mobile to multi-access to reflect the myriad of wireless networks that benefit from having compute power closer to the end-user and/or device. It also reflects the migration to 5G, as having computing power and associated applications at the network edge will be critical to reducing latency and increasing effective throughput through things like edge caching.

In that regard, Odgers explains that the effort is about bringing the cloud closer to the community of interest, whether that community is a corporate campus, a village or a stadium. It is also about being able to customize the cloud to meet the needs of each community of interest. According to the ETSI MEC Technical Requirements document (PDF), the three main categories have been identified as MEC use-cases are:

  1. Consumer-oriented services (e.g. gaming, remote desktop, augmented and assisted reality, cognitive assistance, etc.)
  2. Operator and third-party services (active device location tracking,  big data, security/safety, enterprise services, etc.)
  3. Network performance and OoE Improvements (content/DNS caching, performance optimization, video optimization, etc.)

As part of the specification (page 16, PDF), it is possible for an application to dynamically move from one part of a network to another to “follow” the user as they move about, optimizing the performance for mobile use-cases.

This ability to dynamically change the characteristics of the network via software speaks of the shift of the wireless network from one that moves uses purpose-built to software defined hardware. Odgers indicates that the idea is to leverage the IT world to create apps that run on a virtualized network. As such, ETSI has specified standard and open APIs to appeal to application developers as well as complement Network Functions Virtualization (NFV).

Odgers suggests Software Defined Networking and Dynamic Network Slicing are key elements that make MEC a stepping stone on the path to 5G.  Network slices are a way of carving up the network for different use-cases and providing different characteristics, such as quality of service, latency, bandwidth, applications, security, etc.

Odgers spoke of an example of this in his afternoon talk at the MEC Focus Day, where he looked at the use of Dynamic Network Slicing for a hospital application. That is, physicians might have one network slice, while machine-to-machine (M2M) devices would have another and guests yet another. By having the computing at the edge, it would be possible to dynamically and automatically adjust the slices. For instance,  and as shown in Odger’s presentation, in the event of an emergency, the dynamic slice associated with the guest network might be shut down, so that more resources could be dedicated to the staff’s slice.

As Odgers indicates, the use of standard APIs opens the network an array of use-case specific application developers that will provide greater value to the end-user and will result in a more valuable network.