This research focuses on a new technology, the "Queue Inference Engine" (QIE), which is a set of statistical procedures that allows one to infer with precision the performance of many important queues (waiting lines) without ever observing them. Using only transactional data -- service initiation and completion times -- all of the key performance measures for Poisson arrival queueing systems can be obtained. Under NSF SBIR Phase I funding, substantial research progress was made, building on the original QIE mathematics developed by Dr. Richard C. Larson at the Massachusetts Institute of Technology, and addressing research problems likely to occur in practical implementation settings. The OIE has proven to be a uniquely accurate and unobtrusive tool for measuring near real-time queue behavior in a variety of settings. The research will be carried out cooperatively with two major U.S. service organizations and is designed to evolve an installable prototype OIE that is relatively "bulletproof" in typical application settings. Using a sequence of three empirical methods involving a combination of analyses of videotaped observations and Monte Carlo simulations, Q.E.D. will examine light traffic situations, inter-service time gaps, heavy traffic congestion periods and sensor-based data augmentation.
