Mobile data demand is growing fast – that much is clear. But the rate at which it will grow in the future is the subject of much debate, especially if the long term is considered. For some, the “long term” means the next five years. But imagine being asked to consider data demand out to 2030. That’s exactly what Ofcom challenged us to do in the study we announced back in November.
Some folks claim that spectrum is running out. Others say that, given the advent of small cells, data offload, LTE-Advanced and other capacity-enhancing measures, more spectrum is not needed. In reality of course, there’s a trade-off: less spectrum means you need to spend more on other capacity enhancing techniques, while in some cases using more spectrum could be more expensive than upgrading your existing cells to newer technology.
Trading off all these factors is tough – but increasingly necessary if sensible decisions are to be made over the long-term need for spectrum
In taking on this challenge, we combined several previous models we had created and enhanced them for the purpose. The resulting modelling tool – which we call CAPisce (“cap-eesh”) – takes a (potentially large) example area, maps current and future demand in detail over the area, and then determines how spectrum, technology and cells available at the the time can be combined to meet the demand. In doing so it takes into account the capability of the existing network and the relative costs of each potential action.
By rerunning CAPisce with different assumptions – partly concering the available spectrum in this project – we can compare the scale of the network and the associated costs necessary to meet the demand, without making arbitrary assumptions about which capacity-enhancing techniques provide the most efficient approach. We included all of the techniques shown in the table below, based on detailed research as to the opportunity for these to bring gains over the long term. For example, we had to forecast the capabilities of of a hypothetical Release 18 of 3GPP, while Release 11 is still being worked on!
All the techniques we studied are found to play a potentially signficant role in enhancing capacity, but the relative importance and savings vary significantly by time and place. Putting it another way, you never actually run out of spectrum, it just gets more (sometimes much more) expensive to apply the other options if you don’t have enough of it.
The full report is available here (7MB), and the wider consultation documents (including extensive annexes with the details of our modelling) are here. The following video shows the model in action for a particular case:
(We recommend setting to full screen and HD to see the details)
Simulation conditions: Central London, Medium spectrum efficiency growth, medium demand growth and medium data offload growth. Medium spectrum growth, with 700 MHz from 2026.
This takes a bit of explanation: first, you see the demand for 2012 over the simulation area. CAPisce builds a variety of sites (macros with 6 and 3 sectors, some small cells) to serve the whole of the demand as a starting point. We calibrated the size of that initial network against the size of real networks to ensure everything was realistic. Then, in 2013, you’ll see the extra demand which has built up over that year which is not served by the initial network, despite the growth in capacity in the existing network arising from better device performance and increases in the quantity of data offloaded to indoor Wi-Fi and femtocells. CAPisce then evaluates a huge range of possible upgrades to existing sites, factoring in the better available technology in that year and options to build new sites, both macrocells and small cells. It compares them all and determines which of them yields the lowest cost for a given incremental demand. The black squares show locations where a cost-effective capacity upgrade has been identified. This process is repeated until the network serves all of that year’s existing demand, and tells us how much has to be spent to meet that demand. The whole process is repeated for each year in the study period (i.e. up to 2030).
The graph at top right shows the available spectrum bands (width indicates the available bandwidth) and the extent to which the capacity associated with that band is used. For example, in the early years only the 2100 MHz band and a small part of 900MHz is available for mobile broadband. As time goes on more bands become available and lots more small cells are built , but nevertheless once 700 MHz becomes available in 2026 it becomes rapidly used to reduce the overall cost of meeting additional demand.
The graph in the middle on the right shows the number of each cell type deployed. This masks some detail: each individual site also has a variety of technology upgrades and spectrum bands according to its needs.
And at the bottom right we show the cumulative network costs associated with the extra capacity , which include all incremental network costs, including capex (initial and replacement) and opex in excess of that incurred in creating the network seen in 2012. The numbers represent the present value of all these costs in 2012 calculated on social discount rates. The numbers shown do not include the costs of offload devices or of user devices, though we have included the offload costs in our report.
This is a single run out of the many cases we ran in our published study. By comparing runs under different conditions (e.g. with and without certain spectrum bands, with different assumptions about the rate at which LTE-A and beyond will increase spectrum efficiency etc.) we can determine how the costs change and hence deduce the relative value of spectrum, advanced technology etc.
Do let us know if you’d like to use CAPisce to help answer your own questions about future mobile capacity. Download our data sheet for more details.