Demand Characterization and Forecasting for Cost-Effective Mobile Network Planning

We consider mobile network demand characterization and forecasting from the perspective of enabling cost-effective mobile network planning. This is however, challenging because of market and service usage dynamics and unanticipated events like COVID-19. To this end, in collaboration with a prominent national-scale mobile network operator, we conduct, for the first time, a demand characterization and forecasting study that is based on multi-year and nationwide region-level mobile network traffic data. Specifically, we analyze the demand characterization of mobile and fixed wireless access (FWA) services at national and regional scales. Crucially, we introduce a clustering-driven methodology for national and regional mobile network demand forecasting at a yearly timescale. Our results highlight the significant diversity in region-level traffic characterization as well as differences between mobile and FWA services. Differently from prior work studying the effect of COVID-19 on mobile network traffic, we do not find any observable effect due to COVID-19 on region-scale traffic characterization. Our proposed clustering-driven methodology, besides being scalable, is shown to yield significant gains in accuracy around an order of magnitude relative to the baseline approach for both mobile and FWA service demand forecasting.