A casual reflection on the last few years in the evolution of the wireless network provides us all the insights necessary to reason that there is at least one final frontier coming down the road. Who can deny that the last few years have been owned by the cloud, virtualization and softwarization (if that is even a word!). Edge Computing too, which is really nothing more than the pushing of all of these concepts deep into places in the network where they have never been before. Fog computing is another term (created by Cisco) for something similar but driven in its genesis more bottom up from the many Internet of Things use cases. The bigger trend is obvious; network intelligence is distributing but where will it, can it go, beyond this?
I believe the next big step in the future will be for this trend to consume our end-user devices. More than this, when technology support catches up to this vision there will be a profound reshaping of what we all call devices. This vision is not as far-fetched as it seems. This area of technology research has a name, device virtualization, and some are referring to the new devices of this new technology area as Service Endpoint Agents, or SEA, devices.
What is a Service Endpoint Agent (SEA)?
A SEA device is a complete rethink of how your smartphone works. The phone in your pocket today is basically a fully integrated device consisting of many blocks of hardware and software all dedicated for you and your own purposes alone. With the exception of application software (e.g. gaming, music, personal assistants) that run in some part in the cloud, most everything you do on your device relies in some way or another on a local execution. A SEA device will not necessarily work like this. In the device virtualization paradigm, the same principles that allow virtualization across data centers or the abstractions of EPC elements in the cloud are applied to enable the dynamic decomposition of functions in a device into executable tasks. These functions may then be assigned for execution wherever this may be optimal, e.g., in another end user device or in a nearby Edge node. In this way, a personal virtual device experience can be enabled via any user interface that may be immediately available to an end user. This interface may be simple or complex. The key, however, is that the interface in whatever form it takes provides the role of a SEA to the end user.
Why would we go to all this trouble?
The key motivation driving this nascent technology trend is flexibility. This is the exact same flexibility motivation that is behind the very observable transformation going on in the wider network today. The advantages are similar; the terminal functions, in this case, can be executed at more immediately capable nodes, providing better performance including battery savings. Another advantage is better management of end user services possibly over a centralized management entity along with increased service scalability.
The profound benefit of distributed operation offered by virtualization in networks via service chaining has similar implications in end user device virtualization. The SEA nodes have the ability to assemble multiple distinct (virtual) functions/services, which are physically delivered by multiple devices but can be instantiated in perception on a single device. Such outcome has particularly significant impact and benefit considering the anticipated proliferation of new user Interfaces with challenging form factors (e.g. AR, paper thin displays, etc.) that inherently bring with them a new level of challenges in terms of power and processing requirements.
The road map to disintegration
Device disintegration has already begun. If you have ever been frustrated on an airplane trying to listen to music that you can just about remember at one time actually lived on your device, then you have some early experience of it. Local storage of music and photos is one obvious area that has been surrendered by our devices. However, the end game for device virtualization is foreseen to go well beyond these cloud assisted application execution models.
The full realization of this distributed model will certainly take some time. It will demand continued innovation directly in this space but will be equally dependent on external enablers. SEA systems will be complex systems, and one of the key challenges will be to continuously and dynamically match functional support needs with physical nodes able to support them. Once matching is performed successfully, service/sub task requests along with their executed versions will need to be routed at high speeds at extremely low latencies to achieve a functioning and stable SEA system.
Today’s wireless systems and internet technologies fall a little short on enabling such large scale, low latency, dynamic service clouds at the network edges. Luckily, 5G is in the works and one of its primary goals is actually to deliver a ubiquitous, ultra-reliable, low latency and high data rate system. 5G will very likely be a key enabler for device virtualization and as such should ensure that meteorological metaphors will continue to dominate the wireless industry’s vernacular for a long time to come.