State-dependent stochastic networks. I: Approximation and applications with continuous diffusion limits (Q1296747)

In this lengthy and comprehensive article, the authors study the transient evolution of state-dependent open \((\text{M}_{\xi}/\text{M}_{\xi}/1)^k\) queueing networks. These are exponential networks in which the arrival and service rates, as well as routing probabilities, depend on the state, the vector of queue lengths. For properly normalised queue-length processes, they derive functional laws of large numbers (FLLN) and functional central limit theorems (FCLTs). They develop new tools to establish convergence, existence and uniqueness of the limits. Their approach to reflection problems with nonconstant directions of reflections is based upon \textit{P. Dupuis} and \textit{H. Ishii} [Ann. Probab. 21, No. 1, 554-580 (1993; Zbl 0787.60099)]. The fluid limit is the unique solution to a multidimensional autonomous ordinary differential equation with state-dependent reflection. The proof of FLLN is based on Lipschitz property of time-dependent reflection operator. The weak limit given in FCLT is the unique strong solution to a stochastic differential equation with time-dependent reflection. These results unify and generalize existing approximation techniques. This state-dependent model provides a flexible framework for accomodating a wide variety of phenomena in queueing networks. The results obtained support the design, analysis and optimization of various manufacturing service, communication and other systems. In particular, this paper extends the results of the authors [in: Stochastic networks. IMA Vol. Math. Appl. 71, 239-282 (1995; Zbl 0829.60081)], and \textit{W. A. Massey} and \textit{W. Whitt} [Queueing Syst. 13, No. 1-3, 183-250 (1993; Zbl 0778.60068)]. The article contains 85 useful references.