scientific article; zbMATH DE number 1409852
The author considers a one-channel queueing system which has a Poisson entry, intensity \(\lambda\), arbitrary distribution \(F(x)\) of a queueing time of demands and a finite mean value \(\tau.\) The queueing device is working at ``heating condition at the end of busy period, i.e., when a queueing of all demands, appearing in the system, has been ended, it immediately outgoes to ``heating which duration has an arbitrary cumulative distribution function \(G(x)\) with a finite mean value \(\nu^{-1}\) and a variance \(\sigma_1^2.\) Let \(\xi_{n}\) be the number of demands which remain in the system after the end of queueing of the \(n\)th demand, \(P_{nk}=P\{\xi_{n}=k\}\), \(P_{n}(z)= \sum_{k\geq 0}P_{nk}z^{k}.\) The author proves existence of \(\lim_{n\to\infty}P_{n}(z).\) For example the following assertion is proved: If \(|z|< |f(\lambda-\lambda z)|\), where \(f(\lambda)=\int_0^{+\infty}\exp\{- \lambda t\} dF(t)\), for all \(z\) satisfying the condition \(0<|z|<1\), then \[ \lim_{n}P_{n}(z)=\sum_{x\geq 0}P_{k}z^{k}= {1-g(\lambda-\lambda z)+(1-z)g(\lambda)\over f(\lambda-\lambda z) -z}P_0 f(\lambda-\lambda z), \] where \[ P_{k}=\lim_{n}P_{nk},\quad g(\lambda)=\int_0^{\infty} \exp\{ -\lambda t\} dG(t), \quad P_0={\nu(1-\lambda\tau)\over\lambda+\nu g(\lambda)}. \] Otherwise, \(\lim_{n}P_{n}(z)=0.\)
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 Q4940262)
