Information-centric networking (ICN) is receiving increasing interest in the context of future wireless networks. So, what is ICN? Why is it important and will it have a role in next generation 5G networks? I think it will and for some very good reasons.
ICN is a research area that dates back to about 2006 and has been cooking for over a decade. It defines a new way of inter-networking or, as I see it, a very natural next step now for the internet—if it is done right.
The principles and essential architecture of today’s internet are rooted in the system that was created for simple file transfers some 40-plus years ago. The internet has, of course, evolved within this restrictive construct, and the media consumption experience you most likely enjoy today is as much a function of carefully choreographed Content Delivery Networks (CDN) as the old internet client-server model. CDNs and the careful management of content (aka information) are essential to the smooth functioning that we perceive today.
Behind the curtain, however, the internet is becoming an ever-increasing mess of redirections and inefficient operation. Even with Moore’s law providing its perennial Band-Aid, this strategy is not sustainable in the long term and certainly not in all the promises that are being made for 5G.
The primary use case of the Internet is large-scale media or information distribution. This is a fact that has been well understood by the ICN Research community for many years. Information Centric Networking is precisely that: an internet architecture that puts information at the center where it needs to be.
Replacing client-server with publish-subscribe
ICN replaces the client-server model with a new publish-subscribe model. What does that mean? Picture this: you are in an open-plan office, and you need a stapler. Following a classic internet model, you would need to call someone who knows where all of the staplers are located. This purveyor of stapler information would then most likely redirect your request to someone else who would deliver the stapler to you.
In publish-subscribe, if you find yourself in this same desperate need, you would simply shout this out and the person closest to you in possession of a coveted stapling instrument would then bring it to you. If that person cannot meet that request, they will ask the person next to them if they can help you out, and so on. At 10,000 feet, this is the basic principle of operation of the ICN. As you can see, it is intuitively a lot simpler and there is really no mystery to it.
At a slightly deeper level, ICN introduces an architecture where the named information need itself (e.g., our proverbial “stapler,” a website or a video clip) is used to signal an interest or a subscription to the receipt of information. This is, of course, different from the classic internet where a pointer is used (i.e. an IP address) for the same purpose. This interest is then transported through the network only as far as is necessary to where the named information is found. This could be one network hop away or several. The improved efficiencies of this model are intuitively obvious.
The devil, of course, is in the details. But one common question is how does all this information magically get distributed? The simple answer is that this is already happening. Aforementioned CDN systems are already deployed quite deeply in operational networks. Edge Computing and Fog Computing embody other trends that, in part, are about moving information closer to all of us.
The SDN/NFV vision of the future
There is one other well-known networking trend that is likely to have the most profound impact on enabling greater information distribution. This is, of course, software-defined networking/network functions virtualization (SDN/NFV). In the complete SDN/NFV vision of the future, there is theoretically and practically no reason why any deployed COTS infrastructure could not be “spun up” as a dynamic web server when and where there is a need for it to perform in this virtualized function. ICN is perfectly aligned and complementary with this larger vision.
So, will ICN play a role in 5G? I believe it must because it ticks many of the requirements boxes being targeted, and all the tools are finally falling into place to make it possible.
To be clear, so-called native ICN or the wholesale replacement of the internet with a new model is a long way out. Let’s call that 6G! There is, however, an interim step that will allow many of the benefits of ICN without ripping up any concrete. This can be achieved by leveraging the primary slicing and dicing capabilities of SDN that will allow an ICN network to run comfortably in parallel with legacy IP systems. This is precisely the rich application and transformative purpose for which SDN was conceived.
Add to that simple IP/ICN interworking or gateway points that would be a lot less complex than the common SS7/IP gateway solutions of old, and this interim step can be achieved without requiring any changes to the installed IP application base (i.e. backward compatibility can be guaranteed).
Finally, there is one other intuitive reason why we must embrace ICN in our next-generation vision. As explained, the ICN model is publish-subscribe. There is another mega trend emerging that operates or will operate in the exact same model. It is called the Internet of Things (IoT). Evolving the network to an ICN will not only deliver much-needed efficiencies and performance improvements, but it will also align our networking architecture with perhaps the largest engine of information generation that the world has ever seen. Just seems to make sense, doesn’t it?