Automatic Dependent Broadcast-Surveillance (ADS- B) was introduced to increase the safety of air travel. Since then it has proven to be a useful source of air traffic data for academic research because it is publicly available. Another source of information are the communications between air traffic control (ATC) and aircraft. Methods are (being) developed to enhance the existing radio communications with digital text messages. These are not publicly available while access to these could further the research in this field.
The lack of publicly available machine-readable ATC data hinders research in the field of air traffic management. Therefore the objective of this study is to develop a method that uses open- source data to infer ATC commands. This two-part exploratory study is presented to address this objective.
The first part explains the rule-based analysis used to infer ATC commands. This method uses a set of rules to detect lateral and vertical changes by approaching the ADS-B data as a time series. This approach can detect changes using the turn rate and rate of climb which can correspond to ’heading’, ’direct to’ and ’altitude’ commands. It was not possible to detect changes in speed due to the variability in the wind.
The second part of this paper tests the found method in an air traffic simulator. For 111 flights, ADS-B data was gathered using the Opensky network. The previous method was applied to this data to create a list of possible ATC commands. For every unique flight, two flight plans were made. The first was based on the found list of commands. The second flight plan used the original ADS-B data. These flight plans were simulated and differences between trajectories were used to test the fidelity of the rule-based analysis.
The results showed that the mean absolute horizontal error was below 15km for 75% of the flights and the mean absolute vertical error was below 100m for 75% of the flights. A remark- able correlation was found between the availability of ADS-B data points along the trajectory and the error in the horizontal distance. When more datapoints are available a smaller error in horizontal distance is observed.
It was found that a method could be developed using a rule- based analysis that produces a list of possible ATC commands. The could be done for both lateral and vertical commands, but not speed commands.