Wireless Network Coverage in the Wild: A Multi-City Multi-Operator Data Set

Citation Author(s):
Orlando Eduardo
Martinez-Durive
IMDEA Networks Institute
Marco
Fiore
IMDEA Networks Institute
Submitted by:
Orlando Martine...
Last updated:
Wed, 11/09/2022 - 18:29
DOI:
10.21227/kf04-yr28
Data Format:
Link to Paper:
License:
93 Views
Categories:
Keywords:
0
0 ratings - Please login to submit your rating.

Abstract 

It is a large-scale data set of wireless network coverage for over 22,000 4G base stations in France. This data set is generated by applying VoronoiBoost and official sources of base station deployment. The data covers ten main metropolitan areas in the country, encompassing a variety of dense urban, suburban and rural areas. The coverage information in each area is reported separately for the four major mobile network operators active in France.Coverage is represented as a set of polygonal shapefiles for each base station in the data set, each associated with a given probability that end terminals associated with the base station are located within the polygon (see image). The data set substantially improves current practices for cellular coverage representation in large-scale studies, which primarily rely on plain Voronoi tessellations of the geographical space. As such, the data set can support data-driven networking research where coverage overlap or interference are vital factors and multidisciplinary investigations based on network metadata that needs to be mapped on the territory.

Instructions: 

The dataset is a collection of 40 (zip compress) pickle files encompassing four operators :

  • Bouygues
  • Free 
  • Orange
  • SFR

and 10 French cities:

  • Bordeaux 
  • Lille 
  • Lyon 
  • Le Mans
  • Nantes 
  • Nice 
  • Orleans 
  • Paris
  • Rennes 
  • Toulouse 

The pandas dataframe structure is the following: 

  • lon: Base station longitude in WGS84;
  • lat: Base station latitude in WG84;
  • voronoi_polygon: shapely polygon of the legacy tessellation;
  • voronoi_boost_polygons: list of optimal scaled Voronoi polygon for different probabilities of association;
  • voronoi_grid: 60km x 60km discrete representation of the probability of association in a 2D matrix with 100m x 100m spatial resolution;
  • voronoi_boost_grid: 60km x 60km discrete representation of the probability of association in a 2D matrix with 100m x 100m spatial resolution;
  •  x: x component of the BS coordinates in the Lambert93 metric system;
  • y: y component of the BS coordinate in the Lambert93 metric system;
  • xllcorner: x component of the lower left coordinates of the metric grids in Lambert93 (voronoi_grid/voronoi_boost_grid);
  • yllcorner: y component of the lower left coordinates of the metric grids in Lambert 93 (voronoi_grid/voronoi_boost_grid);

The attached document contains more details on accessing and interpreting the data frame.

Funding Agency: 
Comunidad de Madrid
Grant Number: 
2019-T1/TIC-16037