Presentation of the paper entitled “Detecting Flight Trajectory Anomalies and Predicting Diversions in Freight Transportation” (http://dx.doi.org/10.1016/j.dss.2016.05.004), held at EMISA 2016, Vienna, Austria (https://aic.ai.wu.ac.at/emisa2016/). Abstract: Timely identifying flight diversions is a crucial aspect of efficient multi-modal transportation. When an airplane diverts, logistics providers must promptly adapt their transportation plans in order to ensure proper delivery despite such an unexpected event. In practice, the different parties in a logistics chain do not exchange real-time information related to flights. This calls for a means to detect diversions that just requires publicly available data, thus being independent of the communication between different parties. The dependence on public data results in a challenge to detect anomalous behavior without knowing the planned flight trajectory. Our work addresses this challenge by introducing a prediction model that just requires information on an airplane’s position, velocity, and intended destination. This information is used to distinguish between regular and anomalous behavior. When an airplane displays anomalous behavior for an extended period of time, the model predicts a diversion. A quantitative evaluation shows that this approach is able to detect diverting airplanes with excellent precision and recall even without knowing planned trajectories as required by related research. By utilizing the proposed prediction model, logistics companies gain a significant amount of response time for these cases.

1. Detecting Flight Trajectory Anomalies
and Predicting Diversions in Freight
Transportation
Claudio Di Ciccio, Han van der Aa, Cristina Cabanillas, Jan Mendling, and
Johannes Prescher
EMISA 2016, Vienna, Austria
claudio.di.ciccio@wu.ac.at

2. Business processes in
transport domain
SEITE 2

3. Continuous task
monitoring
SEITE 3

4. Continuous task
monitoring
SEITE 4

5. Continuous task monitoring
in multimodal transport
SEITE 5

6. Continuous task monitoring
in multimodal transport
SEITE 6
Diversion
Diversion airport

7. Dealing with flight diversions
A real-life scenario
Start
End
SEITE 7
©

8. Dealing with flight diversions
A real-life scenario
SEITE 8
©

9. Dealing with flight diversions
A real-life scenario
SEITE 9
©

10. Dealing with flight diversions
A real-life scenario
SEITE 10
©

11. Dealing with flight diversions
A real-life scenario
SEITE 11
©

12. Dealing with flight diversions
A real-life scenario
SEITE 12
©

13. Dealing with flight diversions
A real-life scenario
SEITE 13
©

14. Dealing with flight diversions
A real-life scenario
SEITE 14
©

15. Dealing with flight diversions
A real-life scenario
SEITE 15
©

16. Dealing with flight diversions
A real-life scenario
SEITE 16
©

17. Dealing with flight diversions
A real-life scenario
SEITE 17
©

18. Dealing with flight diversions
A real-life scenario
SEITE 18
©

19. Dealing with flight diversions
A real-life scenario
SEITE 19
©

20. Dealing with flight diversions
A real-life scenario
SEITE 20
©

21. Dealing with flight diversions
A real-life scenario
SEITE 21
©

22. Dealing with flight diversions
A real-life scenario
Modern technology
comes
into play
SEITE 22
©

23. Dealing with flight diversions
A real-life scenario
SEITE 23
©

24. Objective:
monitor the
continuous task
and, in case of anomalies,
raise an alert
at this time:
not at this time:
Motivation
SEITE 24
©

25. Flight diversion
Flight diversion is an example of
continuous task execution anomaly

26. Flight diversion
Flight diversion is an example of
continuous task execution anomaly

27. Which is going to be diverted?
Source: http://www.flightradar24.com/
SEITE 27

28. Motivation
Objective:
monitor the
continuous task
and, in case of anomalies,
raise an alert
at this time:
not at this time:
… with an automated
integrated system
SEITE 28

29. Solution sketch
 Gather and buffer flight data information
 Slice data into time-based intervals
 Extract flight features (deltas) representing the
flight in the interval
SEITE 29

30. Interval-based
progress features
 Features are extracted out of data
 Clustered into fixed-length time intervals
SEITE 30
Gather flight
data events
along a time
interval
Interpolate
attribute
values
Redo

31. Solution sketch
 Gather and buffer flight data information
 Slice data into time-based intervals
 Extract flight features (deltas) representing the
flight in the interval
 Let an automated classifier establish whether
the features are anomalous
 In our implementation:
Support Vector Machines (SVMs)
 After a given number of consecutive
anomalous intervals, raise an alert
SEITE 31

32. Interval-based
checking
latitude
longitude
velocity (speed)
height (altitude)
timestamp
<lat,lon,v,h,t>
SEITE 32

33. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
[features] [SVM]
SEITE 33

34. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
SEITE 34

35. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
SEITE 35

36. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
SEITE 36

37. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
SEITE 37

38. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
SEITE 38

39. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
SEITE 39

40. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
SEITE 40

41. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
SEITE 41

42. Interval-based
checking
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
∆ 𝑑
gain
, ∆ 𝑑
cmpl
, ∆ 𝑑
ph
, ∆ 𝑣, ∆ℎ 
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
<lat,lon,v,h,t>
SEITE 42

43. Evaluation:
Flight data
 Flight data gathered from FlightStats.com and
FlightRadar24.com
 July-August 2013
 (Semi-)publicly available
 K-fold cross validation
Area Diverted Regular Overall
EU 46 746 792
US 22 316 338
Total 68 1,062 1,130
* Thanks to Han van der Aa for his contributionSEITE 43

44. Evaluation:
Train & validation (tuning)
F-score, Precision, Recall F-Score v. time-to-predict
* Thanks to Han van der Aa for his contributionSEITE 44

45. Evaluation:
Test results
SEITE 45

46. Further reading
 Claudio Di Ciccio, Han van der Aa, Cristina Cabanillas, Jan Mendling, and
Johannes Prescher (2016)
Detecting flight trajectory anomalies and predicting diversions in freight
transportation
Decision Support Systems, 88, 1 - 17
http://dx.doi.org/10.1016/j.dss.2016.05.004
 Cristina Cabanillas, Claudio Di Ciccio, Jan Mendling, and
Anne Baumgrass (2014)
Predictive Task Monitoring for Business Processes
BPM 2014, Springer
http://dx.doi.org/10.1007/978-3-319-10172-9_31
 Anne Baumgrass, Cristina Cabanillas, and Claudio Di Ciccio (2015)
A Conceptual Architecture for an Event-based Information Aggregation
Engine in Smart Logitics
EMISA 2015 (GI)
http://subs.emis.de/LNI/Proceedings/Proceedings248/109.pdf
SEITE 46

47. Detecting Flight Trajectory Anomalies
and Predicting Diversions in Freight
Transportation
Claudio Di Ciccio, Han van der Aa, Cristina Cabanillas, Jan Mendling, and
Johannes Prescher
EMISA 2016, Vienna, Austria
claudio.di.ciccio@wu.ac.at

48. Detecting Flight Trajectory Anomalies
and Predicting Diversions in Freight
Transportation
Extra slides

49. System Architecture:
Which component does what

  

50. Further reading
 Claudio Di Ciccio, Han van der Aa, Cristina Cabanillas, Jan Mendling, and
Johannes Prescher (2016)
Detecting flight trajectory anomalies and predicting diversions in freight
transportation
Decision Support Systems, 88, 1 - 17
http://dx.doi.org/10.1016/j.dss.2016.05.004
 Cristina Cabanillas, Claudio Di Ciccio, Jan Mendling, and
Anne Baumgrass (2014)
Predictive Task Monitoring for Business Processes
BPM 2014, Springer
http://dx.doi.org/10.1007/978-3-319-10172-9_31
 Anne Baumgrass, Cristina Cabanillas, and Claudio Di Ciccio (2015)
A Conceptual Architecture for an Event-based Information Aggregation
Engine in Smart Logitics
EMISA 2015 (GI)
http://subs.emis.de/LNI/Proceedings/Proceedings248/109.pdf
SEITE 50