Artificial intelligence will improve the quality of MTS mobile network coverage

2024: Geoplatform launch to create and update digital 3D maps of terrain

MTS, a digital ecosystem, has developed and launched its own geoplatform for creating and updating digital 3D maps of the area by processing satellite images with generative artificial intelligence based on a neural network. The MTS solution will allow you to more accurately and quickly select the locations of base stations, which will improve the quality of mobile communication services and increase the efficiency of investments in infrastructure development. MTS announced this on February 20, 2024.

During operation and modernization of the mobile network, MTS specialists create a digital model of radio coverage using accurate and current digital maps of the area, which should take into account all factors affecting the signal propagation and the number of subscribers in locations. Such factors include the emergence of new residential areas and high-rise buildings, social, commercial and sports facilities, industrial zones, as well as the development of transport infrastructure and a change in the natural landscape.

Previously, MTS purchased updated digital maps from external suppliers, the company's specialists visually assessed topographic changes in the area and buildings based on satellite data, then manually made changes to digital models of the area. It was a lengthy, time-consuming and costly process.

With the launch of the geoplatform, current high-resolution satellite images are automatically submitted to the input of a neural network trained for mapping analysis, which performs a large amount of work on visual analysis and three-dimensional identification of objects in seconds. As a result, one of the most labor-intensive processes in the processing of cartographic information - vectorization of depicted objects - is accelerated almost tenfold. If a specialist in cartographic software spends about 33 seconds on processing the image of one building, then in neural networks this process takes 3-4 seconds. In this case, neural calculation algorithms automatically evaluate the height of buildings and their configuration from shadows or visible parts of walls to build a 3D model.

The planning of a new network or the modernization of the existing one begins with a thorough calculation of radio coverage based on digital maps of the area, which take into account the terrain, construction features and places of concentration of subscribers. Even where the network has been operating for a long time, these parameters should be taken into account when analyzing the current and future need for network coverage and capacity. To increase the efficiency of radio planning, at the initial stage we chose more than 460 cities with a population of over 50 thousand inhabitants, where geodata require constant updating due to changes in the urban landscape, transport flows and population migration, - commented Alexey Maslyankin, Director of the MTS Radio Access Network Department.

Maslyankin stressed that the impetus for the creation of the geoplatform was a successful pilot project carried out in 2021-2022. Now the high-tech and continuous process of updating digital maps will increase the accuracy of planning the MTS network throughout Russia, as well as improve related business processes that require cartographic calculations. And this means not only improving the quality of our services and expanding the ecosystem, but also increasing the return on investment in the development of digital infrastructure.