Published in:
16-03-2022
First exit time for a discrete-time parallel queue
Author:
Zbigniew Palmowski
Published in:
Queueing Systems
|
Issue 3-4/2022
Log in
Excerpt
We consider a discrete-time parallel queue, which is a two-queue network, with batch arrivals and services. At the beginning of the
nth time slot,
\(A_n\) customers arrive to both queues. Then after this,
\(S_n^i\in {\mathbb {N}}\cup \{0\}\) customers can be served in slot
n by server
i for
\(i=1,2\). We assume that
\(\{A_n\}_{\{n\in {\mathbb {N}}\cup \{0\}\}}\),
\(\{S_n^i\}_{\{n\in {\mathbb {N}}\cup \{0\}\}}\) are independent sequences of i.i.d. random variables with support in
\({\mathbb {N}}\cup \{0\}\) and
\({\mathbb {E}}A_n<{\mathbb {E}}S_n^i\) for
\(i=1,2\). Let
\(Q_n^i\) with
\(i=1,2\) be the queue length after the service
\(S_{n-1}^i\) and before the arrival
\(A_n\). Then, it satisfies the Lindley recursion
\(Q_{n+1}^i=(Q_n^i+A_n-S_n^i)_+\) for
\(i=1,2\) and its stationary law
\((Q_\infty ^1, Q_\infty ^2)\) is given by
\((\max _{n\in {\mathbb {N}}\cup \{0\}}T_n^1, \max _{n\in {\mathbb {N}}\cup \{0\}} T_n^2)\) where
\(T_n^i=\sum _{k=1}^{n} (A_k-S_k^i),\quad T_0^i=0, \quad i=1,2.\) For
$$\begin{aligned} {\mathscr {A}}_n=\sum _{k=1}^n A_k, \quad {\mathscr {S}}_n^i=\sum _{k=1}^n S_k^i, \quad i=1,2, \end{aligned}$$
and
\(x,y\in {\mathbb {N}}\cup \{0\}\), we are interested in
$$\begin{aligned} H(x,y)= & {} {\mathbb {P}}(Q_\infty ^1>x, Q_\infty ^2>y)\\= & {} {\mathbb {P}}(\text {there exists }n\in {\mathbb {N}}\text { such that}\quad {\mathscr {A}}_n>\max \{x+{\mathscr {S}}_n^1, y+{\mathscr {S}}_n^2\}). \end{aligned}$$
Note that
H(
x,
y) equals the probability that the first entrance time of the two-dimensional random walk
\(\{({\mathscr {A}}_n-{\mathscr {S}}_n^1, {\mathscr {A}}_n-{\mathscr {S}}_n^2), n\in {\mathbb {N}}\cup \{0\}\}\) to the set
\([x, \infty ]\times [y, \infty ]\) is finite. In the context of actuarial science,
H(
x,
y) corresponds to ruin probability for a discrete-time two-dimensional insurance risk model where each business line faces common simultaneous losses
\(A_k\) and
\(S_k^i\), for
\(i=1,2\), is the premium derived within the
kth period of times (see [
1] for similar considerations). …