Apps World Series

Navigation Area

Lowering Costs and Driving Service Innovation by Migrating Capabilities to the Network Edge

Mobile Edge Congress

Speaker Spotlight – Mahadev Satyanarayanan Article

Close up of hand playing Pokemon GO on smartphone

In 2009, computer science professor Mahadev Satyanarayanan wrote an article entitled ‘The Case for VM-based Cloudlets in Mobile Computing’. His research prompted technical efforts worldwide to intersect mobile and cloud computing, and it was this piece of work that led to the emergence of ‘edge computing’.

We met with him to discuss recent developments in edge computing, and to hear his thoughts on what the future looks like for this pioneering new technology.

What have been the biggest developments in mobile edge computing over the past 12 months?

This year has been about developing the standards that will allow Mobile Edge Computing to succeed.

There’s been collaboration between vendors, service providers, our own research group and the ETSI MEC standardisation group to develop a fully specified set of requirements, due to be finalised later this year. It’s vital that the requirements and critical interfaces identified by ETSI MEC meet with the findings of our research on cloudlets, and it’s been a productive relationship.

How has your role in edge computing changed since your initial research in 2009?

Initial interest in edge computing came from Vodafone in 2013. Two years later, a group of companies including Huawei, Intel and Nokia came together to form the Open Edge Computing Initiative. Now we’re now starting to see the momentum build, with groups working together to help imagine a future where edge computing can succeed. Individuals have been talking to my research group for insights into cloudlets to make sure the standards being developed are good.

We’re currently in the process of setting up a ‘living edge lab’ in Pittsburgh to encourage edge computing Proof of Concepts, and in October of this year I’ll chair the First IEEE/ACM Symposium on Edge Computing, following academic research and interest around the topic.

There is lots of debate about how Mobile Edge Computing will power 5G. How do you think the two will work together?

To me it’s simple – 5G will be useless without edge computing. 5G promises to offer better bandwidth and ultra-low latency, as low as 1 millisecond. 5G and the ultra-low latency it offers will be a worthless investment unless edge computing is present to transform this 1 millisecond first-hop latency to 1 millisecond response time end to end. The deployment of edge computing is critical and a precursor to the deployment of 5G.

Describe what the benefits of edge computing look like to a user?

The phenomenon of Pokémon Go is a perfect example. The game has topped Twitter’s daily users and has more engagement than Facebook. It’s extremely popular, but the kind of augmented reality it uses is very simple. Pokémon are rendered by the smart phone, and placed on top of an image that is obtained by the camera. Imagine the next generation of Pokémon Go leveraging edge computing. It might integrate facial recognition, or transform the room you’re standing in depending on what’s happening in the game.

How would this work? The image is streamed to the edge node, which then applies sophisticated transforms and sends the video back to the smart phone.

Futuristic augmented reality is possible by high end compute very close to the user. The image obtained by the smartphone could be transformed using a high end GPU on a cloudlet located close by. This is just a small example of how edge computing could change a single experience for the user.

What steps should be taken for edge computing to be deployed smoothly?

A piloting phase for operators and application developers is needed in the next year or two. Operators have no experience managing edge computing platforms and need to learn the pitfalls and challenges upfront.

There are potential issues when flash crowds gather unexpectedly and the collective edge computing demand overloads the infrastructure. You could load share, or migrate some users to another node completely transparently, but this would need to be deployed with experience – which telcos don’t yet have. Hands on experience managing edge computing through a piloting scheme on a small scale is key.

What use cases are yet to be explored?

We’ve barely touched the surface in terms of what edge computing can do. How can its full potential be realised? Collaboration across multiple industries.

The transformation of the spectator experience is particularly exciting. Imagine you could collect video from hundreds of smartphones, and fast. You suddenly make available vast amounts of video footage, putting total control into the hands of the spectator, whether they’re live or watching the experience from home. At the moment, a user submits the video to YouTube which then produces the mp4 and makes it available to the user. If Google were to deploy parts of YouTube collection to edge nodes, this process would become more effective.

What will help encourage the development of edge computing applications?

The telco industry needs to actively reach out to developer communities. Start-ups, the IT industry and gaming companies should engage with them to develop Proofs of Concepts. Historically, operators have taken a passive approach to new technologies, but they’ll need to engage with potential users of edge computing upfront to educate them about the potential value, opportunities and challenges.

What are the potential revenue opportunities for edge computing?

There are a few business models possible. It could be similar to the current relationship between operator and customer where you agree to a plan upfront. Alternatively, the user could pay for edge computing services indirectly. Take Pokémon Go for instance, you might pay for an upgraded version of the app with premium services, Pokémon Go then pays the telco accordingly.

There are variants of these ideas, and they’ll need thorough exploration and trialling. It depends on the market, what they’re used to and the specific requirements of the application.

What message are you hoping to bring to MECC?

It will be two-fold. Firstly, that there’s an immediate need for hands on experimentation of edge computing through piloting schemes. Secondly, that operators should extend well beyond their comfort zone in order to make edge computing a success.


  Hear more from Mahadev Satyanarayanan at Mobile Edge Computing Congress 2016, 20 – 22 September,where he will be discussing the changing role of telcos as mobile edge computing technologies evolve.


Share this post:


Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>