The cloudification of spectrum

As a person, I tend to seek and look out for elegant, sustainable and scalable ways to do things. Unsurprisingly, I am an unabashed advocate of the #cloud - especially along three dimensions that are well illustrated in the eloquently written paper by Rolf Harms and Michael Yamartino - the Economics of the cloud.

They are

• Supply side savings
• Demand aggregation
• Multi-tenancy efficiency

I find corollaries of this everywhere - from simple public transportation and #uber -share rides, to shared office spaces (#wework) and multi-purpose venues. Rather than reserve and under-utilize an asset, especially one that has a fixed cost to maintain and operate, why not share its capabilities among as many users as possible as long as each user's needs are satisfied.

This brings me to the topic of #spectrum. Spectrum has traditionally been split between either shared (primarily contention based - think WiFi) and reserved (think licensed spectrum of the operators). There have been some cases where spectrum sharing has been attempted - such as TV Whitespaces ( Victor Bahl and Ranveer Chandra have done pioneering work here), but it has never reached mass adoption. #darpa did also attempt this via its Spectrum Challenge; this was a technical success but did not spark the kind of revolution compared to other programs such as the Grand Challenge that kick started the autonomous car industry.

Spectrum has another interesting aspect; while available frequencies may be infinite the “good” spectrum has traditionally been around the sub 3/3.5 GHz bands for mobile. This is the sweet spot between being able to cover a large area while maintaining a positive business case for the operator. There is more bandwidth available as you go up in frequency adding more capacity - but we are limited by coverage. The laws of physics definitely do apply here. If we look at contention based mechanisms like #wifi, they do work to some extent, but the total amount of spectrum allocated is low and well - it can get quite crowded!

The question begging to be asked - can we design and implement a cloud based approach to spectrum itself - and if so, how could it look like.

The idea in itself is relatively simple - while execution complex. Rather than operators having reserved access to spectrum for which they have paid billions for - the entire spectrum is available for all, but not all the time. When an application is opened on an end device, based upon the service needs a sliver of spectrum and related SLA’s are “reserved” for the period of the session. This is not much different than the serverless computing paradigm. That way, whether you desire 4K streaming, a simple chat with your friends or a batch upload of data - you get the bandwidth you need, when you need it - no more, and no less. And when you are done with playing your game or watching your video - this spectrum is released for the next user.

If that is possible - then the next question is the mechanism to do so. A starting idea would be similar to that of a #googleadsense platform, where allocation is done based upon the localized demand, the availability of “spare capacity” and the willingness to pay. If such a trading platform was available at the #operator #edge - it could make split second decisions on how to allocate resources to maximize a desired outcome. There would of course have to be considerations for certain users - such as public safety; but this could be factored in as part of the rule book. The end devices themselves would essentially be software radios, and we have certainly progressed in leaps and bounds beyond my initial introduction to this concept by Jeff Reed , James Neel and Thomas Rondeau over 2 decades ago.

Where would it leave the end user - I daresay, better off. By “cloudifying” spectrum we can now simultaneously incorporate and include a broad cross section of apps and services. What about the operator? Spectrum monopoly may end, but they could very well play a role as a broker of bandwidth, and offer their end users increased value without worrying if they have the spectrum to support it. And the governments all across the world, many of whom salivated at the ability to auction off access to “ether” for billions of dollars; they could potentially earn much more as a tax on each spectrum transaction. Operating highly efficient systems ensures that spectrum is priced to incentivize high utilization.

Of course, none of this is easy - there are issues such as managing this over a handoff, or indoor coverage etc. Not to mention the politics - operators continue to maintain significant lobbying power around the world, and spectrum is an asset that they hold dear to themselves. But just for a moment, imagine if the entire telecom value chain - from core to RAN and spectrum was offered as a service…. What would such a world look like!