Gérald Gurtner

Physics, Networks and Agent-Based Models

Community Detection — Navpoints and Sectors

Always based on traffic data, we also built the networks of sectors and navigation points. Sectors are volumes of airspace directly controlled by two controllers — one tactical, one strategic — whereas the navpoints are fixed points with latitudes and longitudes designed to help navigation — the flight plan is essentially a sequence of navpoints + altitude.

Beginning with the sectors, we ended up with the following communities using the infomap algorithm.

map_communities_infomap.png

Since sectors are often on top of each other, the whole picture is a bit messy, but local communities are clearly visible. What is interesting is that, with the algorithm we use, there is no a priori on the number of communities, or there average size. Just like the communities of airports looked like a FAB partition, this partition looks like a partition between of ACCs, even though some parts are much bigger, for instance in Eastern Europe.

map_communities_ACC.png

This, once again, could lead to new data-driven division of space which directly into account the existing traffic, instead of a top-down approach lead primarily by political views and expert judgements (which are absolutely not to be discarded!).

We did the same exercise for the navpoint network, producing the following communities (here with OSLOM).map_communities_oslom_new.png

The number of navpoints is much than sectors, and a local community structure appears in this network too.  Quite strikingly, the approximate size of the communities is the ones of the ACCs, even though in some part of Europe, their size is much smaller and fits better the sector partition. map_communities_ACC.png

PreviousBack to Air Traffic ManagementForward to Air Traffic Simulator

%d bloggers like this: