Original title: A global mobile network coverage raster product at 1km resolution, 1999--2030
Till Koebe, Theophilus Aidoo, Ali El Chami, Ali Kanso +4 · September 2026
Having a mobile phone signal shapes how people work, learn, bank, and get healthcare. Yet, the world lacks a consistent, local record showing where mobile signals actually reach. To solve this, researchers built annual maps of mobile signal probability at a 1km scale across 214 countries and territories from 1999 to 2030. They mapped 2G, 3G, and 4G networks using three independent tools: a calibrated machine-learning model, a techno-economic network build-out simulator, and a spatial deep-learning model. Because cell networks track human development, these models rely on data like population spread, electricity, and physical infrastructure. The team trained them on 2,409 operator-reported coverage maps up to 2020, predicted 2021 to 2024 using recent spatial data, and projected 2025 to 2030 using future trends. They combined the three estimates into one best estimate with uncertainty bands to help track the digital divide and plan infrastructure.
You are trying to figure out which parts of a large city have good cell service each year, going back to 1999 and ahead to 2030. You only have a few reports from phone companies, mostly from before 2020. The service changes as new towers are built, and you do not know exactly where. In this paper, the researchers use three different methods to estimate coverage based on things like population and roads, then combine them into a single map with uncertainty bands.
Knowing where mobile signals reach affects billions who rely on phones for banking, education, and emergency help. Until now, no consistent global record existed. This dataset lets researchers track the digital divide year by year and plan infrastructure where it is needed most.
The abstract notes that predictions for 2021–2030 are modeled, not observed, and uncertainty increases for extrapolated years. Validation is reported only for the ML model on held-out countries; the combined product's external validation is described elsewhere. The maps rely on operator-reported data up to 2020, which may have biases. The 1km resolution is coarse for local planning, and 5G is not covered.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
A global mobile network coverage raster product at 1km resolution, 1999--2030 — Till Koebe, Theophilus Aidoo, Ali El Chami et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.