Recently we got to review Ciscos White paper on traffic predictions for mobile data in the years to come. An interesting discussion is what is going to drive the future generation network development – beyond LTE and its cousins. The “next 50 billion devices” has been the talk of the town – all these machines talking to each other has been one serious candidate. Sure enough the expected growth potential of these applications is huge. What does the Cisco study say in this respect? Have a look at the diagram below.
The traffic volume increase for mobile data is tremendous. The traffic is dominated by mobile video followed by Mobile Web/data. These two “human centric traffic types make up 87% of the total traffic. M2M volumes are dwarfed – less than 5% is M2M, and closer scrutiny of the results in the paper shows that the vast majority of the “M2M” traffic is, in fact, expected to be surveillance videos (i.e. more mobile video) !
So looking at this prediction, the conclusion must be that despite of the billion devices, we can forget M2M as capacity driver in foreseable future. Since capacity is driver of the cost and energy consumption, it seems clear that the design characteristics of future networks have to be geared to handle the human centric domain. M2M may be a reliability and scalability driver -but this will not determine the number of basestations and the bulk of the cost of the operators. In the M2M field we need a number of really smart technical solutions to cope with the massive scale in devices, addresses etc – but are these solutions challenging in terms of network resources (provided that we build the other stuff)? Hardly. (Operators are likely to make some money on the niche application high-QoS M2M, though)
The 25-30 fold increase in capacity until 2015, and a 1000 fold increase until 2020 will not happen by itself. Current architectural solutions are not scalable and just more of the same will simply be at least an order of magnitude to expensive to cut it on the market. 1000 times more capacity and the same cost and same energy is a formidable research challenge to be delt with. Lets focus on this first in order to enable also the other less demanding applications.
Now, should we should undisputedly accept all what Cisco says in their white paper? Not at all. However, this is are, as far I know, the by far best substantiated traffic estimate (based on a well-documented method). Any effort to design future networks should at least relate to this, or find scientific support for an alternative estimate.
Excellent as always, Jens. Thank you. Loutfi Nuaymi
normalizing the above plot by the mean capacity per application, the number of M2M connections (or subscriptions) would very likely be on par with the number of “human centric” subscriptions.
Even though flat rates will soon be a distant memory, operators will still earn by selling more subscriptions (even if they have minor capacity demands).
Either way, i completely agree that providing 1000 times more capacity is a very exciting research challenge which fortunately needs something smarter than increasing the number of antennas or the modulation order.
Time for new revolutionary ideas in 3GPP! :)