next up previous
Next: 2.5.4 GI/G/1-type processes Up: 2.5 Markov chains with Previous: 2.5.2 GI/M/1-type processes


2.5.3 M/G/1-type processes

Consider again the example of the M/Hypo(2)/1 queue. Now suppose that the arrivals can occur in bulk, i.e., one job arrives with rate $\lambda $, two jobs arrive simultaneously with rate $\lambda/2$, three jobs arrive simultaneously with rate $\lambda/4$, i.e., the bulk sizes decrease geometrically. The service process consists of two exponential phases in series. The state transition diagram of this process is illustrated in Figure 2.4.

Figure 2.4: The state transition diagram of a M/Hypo(2)$1$ queue with bulk arrivals.
\begin{figure}\vspace{-0.2in}
\centerline{\psfig{figure=figs-mam/mg1.eps,width=4.5in}}\end{figure}

The processes with similar patterns in the embedded Markov chain are known as M/G/1-type or skip-free to the left processes. Their infinitesimal generator matrix ${\mathbf{Q}}_{M/G/1}$ can be block-partitioned as:

\begin{displaymath}
{\mathbf{Q}}_{M/G/1} =
\left[ \begin{array}{c c c c c c c }
...
...vdots & \vdots & \vdots & \vdots & \ddots
\end{array}\right] .
\end{displaymath} (2.22)

The infinitesimal generator ${\mathbf{Q}}_{M/G/1}$ of an M/G/1 type process is an upper Hessenberg matrix, i.e., the blocks below the main block-diagonal, but the first one, are all zero matrices. As depicted in Figure 2.4, M/G/1-type processes usually characterize bulk arrivals, i.e., more than one job may arrive in the queueing system at a time [34]. We give details on solution methods for M/G/1-type processes in Chapter 3.


next up previous
Next: 2.5.4 GI/G/1-type processes Up: 2.5 Markov chains with Previous: 2.5.2 GI/M/1-type processes
Alma Riska 2003-01-13