Robust, integrated arrival-departure-surface scheduling based on Bayesian networks

Airports and terminal airspaces in busy metropolitan areas are key capacity bottlenecks within the national air transportation network. Two key aeronautics challenges relevant to traffic management in these constrained flight domains are [1] - (i) developing the capability to perform four-dimensional trajectory (4DT)-based planning, and (ii) increasing airport approach, surface and departure capacity. The key to addressing both of these challenges simultaneously is to generate globally-optimal trajectory optimization and de-confliction strategies that span multiple flight domains of the National Airspace System (NAS), while efficiently addressing the strong and uncertain interactions between component subsystems. To approach the problems of uncertainty and interaction effects, current ATM research typically uses piecewise control (i.e., separate planning functions for each flight domain) and the use of frequent re-planning on a local level, without a clear connection to their effects on overall system performance. This paper presents a novel multidisciplinary design approach for integrated arrival-departure-surface spacing and scheduling, covering two flight domains: terminal airspace and airport surface. We develop a spacing-and-scheduling decision support tool (DST) called PROCAST (Probabilistic Robust Optimization of Complex Aeronautics Systems Technology). PROCAST combines the attributes of modern developments in the disciplines of Complex Adaptive Systems (namely, NextGen AeroSciences' Continuous Re-planning Engine, NACRE) and Probability Theory (namely, Probabilistic Graphical Models, (PGMs)). We provide a preliminary proof-of-concept for PROCAST by applying it to the simulation of arrival and departure traffic on the ground at the John F. Kennedy International Airport (JFK), with future plans for extending it to cover the entire New York metroplex (all major airport surfaces and the TRACON airspace).