Ward Whitt (born in 1942) spent his early years inBozeman, Montana, before attending Dartmouth (ABin mathematics) and Cornell (PhD in operations re-search, 1969). His doctoral thesis, Weak ConvergenceTheorems for Queues in Heavy Traffic, advised byD. Iglehart, set the direction for his future research.Whitt’s career has three phases: In 1969–1977, he wasat Yale; in 1977–2002, he was at AT&T, first in BellLabs and then AT&T Labs; since 2002, he has been atColumbia in the IEOR Department.

Whitt’s research has focused on stochastic queueingmodels and their application to communication, com-puter, manufacturing and service systems. In his 330publications he has also contributed more broadly tostochastic modeling, e.g., to stochastic-process limits,stochastic comparisons, simulation methodology, nu-merical transform inversion, approximate dynamicprogramming and heavy-tailed probability distribu-tions.

Whitt’s 1970 papers with Iglehart were seminalcontributions on the use of asymptotic methods todevelop tractable approximations describing the per-formance of complex queueing systems. These papersshow the power of functional limit theorems, the con-tinuous mapping theorem and state space collapse toestablish heavy-traffic limits for multi-server queueswith non-exponential distributions and networks ofsuch queues. Whitt’s work culminated in his 2002book, Stochastic-Process Limits, for which he wasawarded the INFORMS Lanchester Prize. As statedin the preface, ‘‘Since the converging processes areconstructed by appropriately scaling time and space,they provide a macroscopic view of uncertainty. Thestochastic-process limits strip away unessential detailsand reveal key features determining performance.’’

Other notable stochastic-process limits are Whitt’s1981 paper with S. Halfin initiating the ‘‘Halfin-Whitt’’many-server heavy-traffic regime, his 1985 paper es-tablishing the asymptotic correctness of the Erlangfixed-point approximation for stochastic loss net-works, complementing work by F. Kelly, his 1991paper with P. Glynn establishing hydrodynamic limitsfor the output from many empty queues in series, andhis 1994 paper with P. Glynn establishing the funda-mental large deviations theory for non-Markovianqueueing models, supporting the concept of effectivebandwidths for communication networks, whichWhitt studied further with A. Berger.

Whitt was strongly influenced by the practical en-gineering environment of Bell Labs. In the early 1980’sthat motivated him to develop the Queueing Network

Analyzer (QNA) software package, implementing adecomposition approximation to describe the steady-state performance of a non-Markovian open networkof queues. The arrival process and the general servicedistribution at each queue are each approximatelycharacterized by two parameters, one describing therate or mean and the other describing the variability.The arrival-process variability parameters are ob-tained by solving a system of equations, like thetraffic rate equations for the arrival rates, based on a‘‘network calculus’’ to represent the operations of su-perposition, splitting and flow through a queue. Thattool motivated Whitt’s research during that period,including studies of how to approximate a stationarypoint processes by a renewal process. Papers with K.Sriram and K. Fendick showed how to capture theperformance impact of dependence among interarriv-al and service times, as occurs with superpositionarrival processes.

The need for useful practical results also motivatedWhitt to work with J. Abate and other Bell Labs col-leagues G. Choudhury, D. Lucantoni and K. Leung todevelop effective algorithms for numerically invertingtransforms and apply them to a wide array of queue-ing models. These results refuted the conventionalwisdom that many previously derived transformswere of little value. Their work received honorablemention for the 1997 INFORMS Lanchester Prize.

Whitt investigated many aspects of queueing sys-tems. With W. A. Massey and others, he studied theperformance of multi-server queues with time-vary-ing arrival rates. That work is surveyed in a 2007paper in this journal, co-authored with L. V. Greenand P. J. Kolesar. He helped improve understanding offundamental queueing principles: Little’s law (with P.Glynn) and Poisson Arrivals See Time Averages (withB. Melamed).

Since joining Columbia in 2002, Whitt has devel-oped models to analyze the performance of large-scaleservice systems, such as customer contact centers andhealthcare systems, mostly in collaboration with stu-dents: on simulation-based staffing algorithms with R.Wallace, on ratio controls with I. Gurvich and O.Perry, on two-parameter processes with R. Talreja, G.Pang and Y. Liu, and on delay prediction with R.Ibrahim.

Whitt was recognized for his research contributionsby several awards: the National Academy of Engi-neering in 1996, AT&T Fellow in 1997, INFORMS vonNeumann Prize in 2001 and INFORMS Inaugural Fel-low in 2002.

Ward Whitt

vii

Ward Whitt