scientific article; zbMATH DE number 3874381

From MaRDI portal
Publication:3339910





[For the entire collection see Zbl 0539.00025.] The construction of stationary solutions of several queueing systems of different types are discussed in the author's framework for queueing processes. The following is an example. The random measure \(N(\omega,.)=\sum_{n\in Z}\sigma_ n(\omega)\epsilon_{T_ n}^{(\omega)}\) is defined on a probability space (\(\Omega\), \({\mathcal A},P)\) with measurable flow \(\theta =(\theta_ t\), \(t\in R)\), with respect to which P is invariant and ergodic. Clients arrive at instants \(T_ n: ...,T_{-1}<T_ 0\leq 0<T_ 1...\), \(\sigma_ n\) is the demand of service of the client arriving at \(T_ n\). Let D(t) be the number of the servers who are working at the moment t, assuming the working load is divided equally among them. The random function D(t) is supposed to be stationary with respect to \(\theta\). The question is to find a stationary solution of the following equation W(t), which represents the charges of the system: \(dW(t)+D(t)I_{[W(t)>0]}dt=dN(t).\) Let \(i=EN(]0,1])\), then if \(i<ED(.)\) there exists a unique stationary solution; if \(i=ED(\cdot)\), there exist infinitely many stationary solutions; if \(i>ED(.)\), there exists no stationary solution.




Cited in
(23)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3339910)