Distributed Infrastructure for Data Analysis in Aviation

Project SafeClouds

Air traffic is increasing worldwide. The constantly growing volume of traffic poses major challenges for aviation security organisations as well as airports and airlines to guarantee the highest possible level of safety. Large amounts of data from various sources such as flight data recorders and radar stations are already being recorded and evaluated.

The project SafeClouds aims at breaking data silos by integrating them into an European wide cloud infrastructure. With the help of machine learning and statistical analysis on such merged datasets the safety of air traffic in Europe will be further increased.

© SafeClouds

Distributed Data Analysis with DART

In our division we are building a multi-tier hybrid cloud infrastructure based on Amazon AWS and combining open-source software such as Apache Kafka with our own tools such as the Distributed Analytics Runtime (DART). DART is based on GPI-Space which is optimized for parallel processing of data in distributed systems. The focus is on data and failure safety as well as easy scalability in terms of number of users, memory and computing power.

The data of the following scenarios is analyzed as an example:

  • Unstable approach: A predefined range for various parameters such as altitude, speed, sink rate etc. is not maintained and can lead to a hard landing, landing abort or similar.
  • Off-road safety warnings: The specified minimum amount was not reached due to geographical conditions.
  • AIRPROX (Aircraft Proximity Hazard): Safety has been compromised by the minimum distance between aircraft being undercut.

Runways are also considered: the aim is, of course, their optimum capacity utilisation; in doing so, the exits to the terminals must also be taken into account while maintaining the minimum safety distances.