Crowds is an algorithm for dynamic communication detection in VANETs. It is an epidemic label propagation in which each vertex (vehicle) chooses a community with the maximum score of it’s neighbors communities.
The score is computed based on:

  • Topological information of the vehicle’s neighborhood
  • Relative mobility of the vehicle and its neighborhood

Below, the are presented communities detected by the algorithm from vehicular traces over the area of Luxembourg country. You can interactively see and explore the structure and the evolution of the dynamic communities:

Source code of the algorihtm: Crowds github project

The screenshot below presents another result of the basic community detection algorithm that uses only topological information. Grey dots indicate separated vehicles (without any neighbour in the communication range) and color dots indicate vehicles assigned to a community:

Screen Shot 2013-10-05 at 2.48.37 PM