Presentation is loading. Please wait.

Presentation is loading. Please wait.

Challenges and opportunities for ATM metric development

Similar presentations


Presentation on theme: "Challenges and opportunities for ATM metric development"— Presentation transcript:

1 Challenges and opportunities for ATM metric development
Big data challenges and non-conventional metrics Complex World Event – 7th-8th April 2015 Marco Ducci – Deep Blue

2 Complex World Event 7th-8th April 2015
ELSA Objectives Motivation Lack of suitable means to analyse and monitor the emerging characteristics at the system level, e.g. EU-wide performance scheme Application of methods used for the data analysis in complex system science to ATM Three main objectives: characterisation of statistical regularities in the current scenario simulation of the emergent properties of the trajectory-based SESAR scenario inform the design of a tool to monitor, predict and intervene on the ATM system Complex World Event 7th-8th April 2015

3 ELSA: The Partners ATM domain knowledge, Human Factors, Validation
Deep Blue ATM domain knowledge, Human Factors, Validation University of Palermo Scuola Normale Superiore di Pisa Complex network experts, Computer scientists, Experience with big data and big data bases Complex World Event 7th-8th April 2015

4 15 months of ECAC wide traffic data
ELSA: Dataset 15 months of ECAC wide traffic data Traffic data from Demand Data Repository (DDR) Historical data about 4-dimensional trajectories over the European Airspace. Both last filed flight plans and radar-updated trajectories are present. Safety events (through ASMT): STCAs and Losses of Separation over Rome FIR Complex World Event 7th-8th April 2015

5 Working with Big Data: challenges…
Hard to get to a uniform understanding within the project team Hard to find a shared taxonomy/glossary Data set often not compliant with the project expectations Data sets often incomplete Hard to SELECT the relevant/meaningful information …and opportunities big data are big! Merge independent data sets Traffic data (DDR data sets) & Safety data (from ASMT) Build databases improves data accessibility and speed up the analysis Improve data availability and data sharing among projects Complex World Event 7th-8th April 2015

6 The Navigation Point Network
Using data about flight plans in a certain airspace it is possible to build a Navigation Point Network: Nodes: Navpoints crossed by the aircraft Links: Two nodes are connected if at least an aircraft has flown from one to the other in the considered period. Navigation point networks are suited to reproduce the structure of the airways in an airspace.   Complex World Event 7th-8th April 2015

7 Network metrics: “classical” vs “non-conventional” metrics
“Classical” metrics: Degree: the number of a node’s neighbours. A node with high degree implies many intersecting trajectories at the same point Strength: The strength of a node is essentially a measure of the traffic load on its adjacent links. It is the most direct measure of the traffic load on the node. Nodes with high strength are also the most busy. Betweenness: It is a measure of a node's centrality in a network. A node with a high value of betweenness means that there are a lot of airways passing through it. New “non-conventional” metrics (computed comparing planned and actual trajectories): Fork: fraction of flights for which a deviation begins after a selected point; that point is the last common point in planned and actual piece of trajectory. Frac: fraction of flights which should have passed by a selected point and have not. Alt: absolute difference of altitude at a selected point between the planned and actual trajectories.  Complex World Event 7th-8th April 2015

8 Using “classical” and “non conventional” metrics: overexpressed nodes
Null Hypothesis: Deviations of flights at a given NVP occur in a random way that however takes into account the heterogeneity of NVPs, i.e. the fact that some NVPs are more used than others. The nodes which reject this hypothesis are special nodes that we call “overexpressed”: we do not expect to have such a high number of flights deviated at this point only by chance. The formulation of an appropriate null hypothesis is crucial when trying to select relevant information from large datasets. Complex World Event 7th-8th April 2015

9 Complex World Event 7th-8th April 2015
Using “classical” and “non conventional” metrics: overexpressed nodes This microarray-like plot indicates that some NVPs are specifically used in specific hours of the day. Complex World Event 7th-8th April 2015

10 “Non-conventional” metrics: challenges…
Choose the right metrics to describe the ATM system Need to involve ATM experts to fully understand the metrics’ meaning in terms of operational impact and relations to industry KPIs …and opportunities Define statistical null models and analyze overexpression Analyse outliers to identify areas of attention Use the ELSA ABM to generate synthetic data to be used as an empirically grounded numerical null hypothesis for the study of specific problems. Complex World Event 7th-8th April 2015

11 The ELSA Agent-Based Model
Illustration of the re-routing sub-module for the conflict resolution Snapshot of the time evolution of the trajectories Complex World Event 7th-8th April 2015

12 Key challenges and opportunities
How to manage and efficiently use big data Improve data accessibility Build databases for efficient data management Share data and lessons learnt Developing “non-conventional” metrics Need for case studies: from the general behaviour to the analysis of outliers Using models to link these metrics with forecasts of KPIs Outliers analysis can support in the identification of areas of attention. They require more in-depth analysis with the collaboration of operational experts to identify: Negative dynamics: improvement actions (Safety I) Positive behaviours: find ingredients to reproduce (Safety II) Complex World Event 7th-8th April 2015

13 Thanks for your attention!
Marco Ducci – Ducci, Gurtner, Monechi, Beato, Tedeschi, Pozzi – DBL Miccichè, Bongiorno, Vitali, Cipolla, Affronti, Mantegna – UniPa Valori, Lillo - SNS


Download ppt "Challenges and opportunities for ATM metric development"

Similar presentations


Ads by Google