The UK’s National Infrastructure Commission is undertaking its first ever National Infrastructure Assessment, of which telecommunications is a key component. The aim of this task is to ensure efficient and effective digital infrastructure delivery over the long term, the results of which will be used to direct both industry and government over coming decades. However, taking a strategic long-term approach to the assessment of telecommunication infrastructure is a challenging endeavour due to rapid technological innovation in both the supply of, and demand for, digital services.

In this paper, the uncertainty associated with the National Infrastructure Assessment of telecommunication infrastructure is explored, focusing specifically on issues pertaining to uncertainty in future demand. We not only quantify the uncertainty of different future demand scenarios, but also measures the performance of different infrastructure interventions. This is vital information for policy makers to understand the long-term relationship between available capacity and future demand.

The methodology applies the Cambridge Communications Assessment Model, which has been developed exclusively for the evaluation of national digital infrastructure strategies, over 2017-2030. We focus not only on future data demand growth, but also on using long-term demographic forecasts that capture changes in fertility, mortality and migration. We test the impact of different infrastructure intervention strategies by developing novel infrastructure deployment algorithms which represent operator behaviour. Consequently, four key intervention options are explored for different demand scenarios including (i) minimum intervention, (ii) spectrum integration on brownfield macrocell sites, (iii) greenfield small-cell deployment or (iv) a hybrid approach using both spectrum and small cells.

We find that a minimum intervention approach performs poorly against future demand growth, while spectrum-based strategies only meet future needs in a low demand growth scenario (which is unlikely). The best performing strategies across a range of scenarios include deploying either a small cell layer (operating at 3700 MHz), or a combination of a small cell layer and the simultaneous integration of available spectrum resources on brownfield macrocell sites (700, 800, 2600, 3500 MHz). No capacity deficit occurred for any of these interventions across the demand scenarios. However, this additional capacity comes with a much larger cost (at a time when revenue is static or decreasing in the mobile industry), leading to a trade-off between available capacity and efficiency of capital deployed.

The results conclude that while national infrastructure assessment is a valid method for thinking more strategically about our long-term infrastructure needs, we must recognise the inherent uncertainty associated with this particular sector. Rapid technological innovation affects our ability to accurately forecast long-term roll-out, making it essential that rigorous examination of this uncertainty is both quantified and visualised to support policy decision-making. To aid in this endeavour, the Cambridge Communications Assessment Model has been developed in accordance with open-source principles, with the model code being freely distributed.