Graph Signal Processing Techniques for Analyzing Aviation Disruptions

Abstract

Understanding the characteristics of air-traffic delays and disruptions is critical for developing ways to mitigate their significant economic and environmental impacts. Conventional delay-performance metrics reflect only the magnitude of incurred flightdelays at airports; in this work, we show that it is also important to characterize the spatial distribution of delays across a network of airports. We analyze graph-supported signals, leveraging techniques from spectral theory and graph-signal processing to compute analytical and simulation-driven bounds for identifying outliers in spatial distribution. We then apply these methods to the case of airport-delay networks and demonstrate the applicability of our methods by analyzing U.S. airport delays from 2008 through 2017. We also perform an airline-specific analysis, deriving insights into the delay dynamics of individual airline subnetworks. Through our analysis, we highlight key differences in delay dynamics between different types of disruptions, ranging from nor’easters and hurricanes to airport outages. We also examine delay interactions between airline subnetworks and the system-wide network and compile an inventory of outlier days that could guide future aviation operations and research. In doing so, we demonstrate how our approach can provide operational insights in an air-transportation setting. Our analysis provides a complementary metric to conventional aviation-delay benchmarks and aids airlines, traffic-flow managers, and transportation-system planners in quantifying offnominal system performance.

Dynamics of Disruption and Recovery in Air Transportation Networks

“…However, these recovery definitions and strategies are airline-specific and of limited use in defining and measuring system-wide characteristics. We expand on this line of work by incorporating system-wide spatio-temporal information [7,10], leading to our comprehensive definitions of disruptions and recoveries.…”

“…where r ij denotes the correlation coefficient between historical airport delays at airports i and j. For a more in-depth discussion, we refer readers to [10].…”

“…This partition distinguishes a nominal region (i.e., nominal TD and TV levels), a high-TD region (i.e., delay magnitudes are high), and an unexpected TV region (i.e., the spatial distribution of delays is unexpected). Prior work has shown that unexpected spatial distributions correspond to regions with very high or low TV [7,10]. In particular, given an observation x(t) belonging to a certain hour, we can compute the bounds on the TV for identifying delay distributions that are spatially perturbed at time t, denoted as Θ, Θ hour(t)…”

“…, using the methods proposed in [7,11,10]. We also compute a delay threshold f hour(t) TD ∈ R ≥0 for each hour to identify delay distributions that have a very high magnitude of TD.…”

“…The inherent variability of factors such as weather in the operating environment, along with the complex interconnectivity of the system, make it difficult to extract actionable insights from past events. Our work focuses on formalizing and analyzing disruptions and recoveries in the air transportation system by leveraging techniques that evaluate aviation disruptions from the perspective of signal processing in networked systems [10].…”