George Mason Univeristy
SYS 798/OR 680
Professor Thomas Speller (Professor for Course),
Professor Alex Levis (Sponsor and Mentor)
Problem Statement: Crowding in rapid transit systems worldwide is a major source of crime, dissatisfaction of ridership,
stress on the system, and danger to the public. Overcrowding is indicative of insufficiencies and/or inefficiencies related to:
Human Factors (bikes/strollers or hurrying to catch train)
Schedule/configuration flexibility (reversing escalators etc)
Worldwide, rapid transit control systems have been created, modified, analyzed, and optimized on an ad hoc basis. Considering
each system as an independent design as is currently done leads to inefficiency and waste. Furthermore, failing to consider the
commonality of such systems prevents finding the best control solutions for the whole and also prevents future design
evolution to improve the whole.
AMART will compare a minimum of three different systems to develop a common language to identify network architecture
(nodes, edges, flows etc) for each zoom level in addition to the relationships between the zoom levels. We will model the
common network flow pattern at multiple chosen related layers of abstraction, concentrating on the minimization of crowds
within the station. Starting from the station level and zooming out iteratively.