Yesterday we had the kick off of the EIT Digital federated SDN Test Bed, FSTB, and one of the topics discussed were "case studies", i.e. area where it makes sense to use the FSTB. One of them was IoT.
The Internet of Things has become the talk of the town, as much as Smart Cities...., and there is both concreteness and hyperbole. For sure the number of "things" already connected is huge and all points to a rapidly increasing number with most observers indicating in 50 billions the number of connected objects by 2020. Although it is not exactly clear how to monetise this (and who will monetise it) the general feeling is that there is big bucks in IoT.
IoT connectivity is reshaping the overall traffic on networks. When networks were first designed they were serving voice calls. And the traffic pattern for voice call saw a 3 minutes average length of each call. If you plot the length of each call on the x axes and the number of calls on the y axes you'll get a bell curve with the peak at 3 minutes (the average call duration). With the digitalisation of transmission you can transform the length into number of bytes, and the peak for the voice call distribution is around 1.5MB (do your math: 64kbps multiplied by 180 seconds...).
In the last fifteen years Internet users have started to overwhelm the voice traffic and a good portion of this traffic was downloading of clips, music and movies. The result is that if we look at the traffic distribution the bell curve has morphed into an S curve since the majority of transaction is now well over the 1.5MB. Now comes the IoT. In most cases things talks by sending just a few bytes of data. And there are billions of transactions (this kind of traffic is also sustained by short messaging that have already changed the traffic pattern). The result is that the S curve changes into a "reversed bell curve" with the lowest point represented by voice traffic.
It should be obvious that the networks designed to support a bell curve shaped traffic may not be ideal to support a reversed bell curve shaped traffic. Also notice that the traffic today, and more so tomorrow, is a mixture of different traffic flow with different characteristics. Hence the need for a much more flexible use of resources, and for SDN.
IoT is a nice wording to denote whatever "thing" is connected to the Internet (there are different views, and therefore estimates, on what it is meant by "connected": directly connected, as a cellphone, or indirectly connected as a smartwatch or a gyroscope inside a smart phone, with corresponding estimates ranging from 50B to 1,000B by 2020). However, as you dig into it you realise that the IoT is a very generic name encompassing "things" that are not just very different from one another but also that have very different requirements in terms of connectivity.
And here lies the reason why SDN is considered to be a catalyst for IoT.
The possibility to reconfigure network resources make it possible to customise the connectivity to the specific need of a specific breed of IoT. The Network Operator can offer a standardised set of network resources and the Service Provider (it may often be one and the same) can customise them dynamically to fit the needs of that IoT.
The flexibility translate into better service, lower cost and sometimes into better use of the "things" battery, a crucial aspect in many IoT application.
At EIT Digital we have just launched a High Impact Initiative on IoT where the focus is on being able to create flexible connectivity platforms to sustain different IoT requirements. There is clearly a good potential for synergy with the FSTB!