NASwide network modeling software for traffic flow management

NAS-wide network modeling software for traffic flow management Alex Bayen Dengfeng Sun, Charles Robelin, Jessica Pannequin, Issam Strub Alaa Hilal, Elie El. Khoury, Ibtissam Ezzedine, Sam Yang, Abdul-Hamid Ghandour University of California, Berkeley Department of Civil Engineering Systems Engineering October 5 th, 2006 Workshop on High Confidence Software for Critical Aviation Systems, Alexandria, VA

Modeling software for NAS-wide network analysis Automated network model building Graph identification techniques (pattern recognition, clustering algorithms) Aggregate flow modeling (partial differential equations, numerical analysis) Mathematical network analysis Delay multiplier analysis (Mixed Integer Linear Programming, MILP) Network sensitivity analysis (adjoint-based optimization) Automated bottleneck identification (multicommodity flows, combinatorial opt. ) Decentralized ARTCC/sector optimization (dual decomposition) Control synthesis for NGATS Dynamic sectorization [collaboration with ENAC/Eurocontrol] Reservoir management Airborne delay mitigation Delay multiplier mitigation

Infrastructure data (sectors, VORs, etc. )

Traffic data (ETMS/ASDI)

Clustering algorithms

Network model

Software specificity: abstractions Data processing capabilities • 1. 5 terabytes / year of data (offline) • I/O: 500 Megs/day (online) CNA Infrastructure data • ETMS/ASDI data structures • CNA oracle / database • ARTCC / FAA infrastructure data NASA, FAA C/C++ Software interfaces • NASA FACET (ATM) • CPLEX (combinatorial optimization) • Matlab (visualization) Metron Aviation

NAS-wide model

Type of predictions

Software architecture and use Infrastructure data -Analysis -Control synthesis -Simulation -Validation -Model assessments weather data, winds FACET OD data, historical data MATLAB CPLEX Function specific software ATM analysis software