PROPEX: The Innovative Tool for Anticipating DTT Disruptions
What is exceptional propagation?
It’s a natural phenomenon that occurs under specific weather conditions (intense heat, clear skies, pressure changes, etc.), allowing radio waves to travel far beyond their usual range.
Ce phénomène est invisible mais son impact est bel et bien concret : une réception de la TNT perturbée et une augmentation significative des réclamations des téléspectateurs. En octobre 2023, un épisode de trois jours a ainsi multiplié par dix le nombre de signalements quotidiens.
Although invisible, its effects are very real: disrupted DTT reception and a sharp increase in viewer complaints.
For example, in October 2023, a three-day episode led to a tenfold increase in the number of daily incident reports.
The ANFR has developed its own forecasting tool based on a dual complementary approach: one physical, the other artificial intelligence-based.
Physical modelling
The propagation of an electromagnetic wave above the Earth depends on the vertical gradient of the refractive index, which is mathematically linked to meteorological parameters such as temperature (T), atmospheric pressure (P), and humidity (H).
When this gradient decreases, it can lead to the formation of a tropospheric waveguide, which causes exceptional propagation phenomena.
By using these parameters, the tool generates forecast maps based on daily downloads of weather data.
The first tests, carried out using past episodes such as the one in 2023, have proven successful.
AI-based approach
Using measurement data and reports collected by the ANFR’s call centre, instances of exceptional propagation were identified.
The corresponding weather data (temperature, pressure, and humidity) were labelled and used to train AI models — specifically machine learning models, a branch of artificial intelligence that enables machines to “learn” from data through mathematical models.
Weather forecasts are then regularly fed into these models, which in turn estimate the probability of an exceptional propagation event.
As more cases are added over time, the accuracy of these models will continue to improve.
Toward better public information and broader applications
The PROPEX project demonstrates the ANFR’s commitment to using research and innovation to enhance its understanding of the radio spectrum and to improve forecasting and public service.
PROPEX is the result of a collaborative effort involving engineers from various ANFR departments and students from diverse backgrounds — including physics, computer science, data science, and artificial intelligence.
In the short and medium term, teams are continuing to refine the forecasting models and are working on setting up alert bulletins and public information messages, which could be distributed through standard communication channels.
In the long term, the tool could be extended to other frequency bands beyond those used for DTT, enabling broader applications.