Chapter 2

Poisson Processes

2.1 INTRODUCTION
A Poisson process is a simple and widely used stochastic process for modeling the
times at which arrivals enter a system. We usually look at arrivals after some starting
time. say t=O. Figure 2.1 illustrates some of the different ways to characterize random
arrivals over the positive time axis. The sequence of times at which arrivals occur is
denoted by the random variables {S J, S2 • ... }. We usually refer to a point on the time
axis at which something happens as an epoch, and thus we refer to Sn as the epoch of
the nth arrival, or the nth arrival epoch. In principle, an arrival process can be character-
ized by a rule specifying the joint distribution functions of {SI> S2' ... , Sn} for all n~l.
but usually these distribution functions are derived in tenns of other random variables.

Jt
X3
Xz
4
~
~
·1 J<o
0
• Xl
~
SI S2 S3
1

Figure 2.1. An arrival process and its arrival epochs (5" 52, ... ), its inter-arrival intervals
(X" X2, ... ), and its counting process ((N(t); t~O}).

An arrival process over the positive time axis can also be described by the inter-
arrival intervals, denoted {XI> X2, ... }. For n::::2, Xn is the interval between the n- 181 and
the nth arrival epoch. i.e .• Xn = Sn-Sn_I ' By convention. XI=SI' It follows that the nth
arrival epoch can be expressed in tenns of the inter-arrival intervals as

S = "",n X. (I)
n LJi=! 1

A rule specifying the joint distribution function of {Xl> .... Xn} for all n~l specifies the
arrival process. Renewal processes. the topic of Chapter 3. are usually specified di-

R. G. Gallager, Discrete Stochastic Processes 31

reetly in this way. In particular, a renewal process is an arrival process in which the
inter-arrival intervals are independent identically distributed (lID) positive random
variables. Thus, a renewal process is specified via the distribution function of the inter-
arrival intervals. A Poisson process is a renewal process in which the inter-arrival in-
tervals have an exponential distribution function: i.e., each Xi has the density fx(x) =
A.exp(-h) for x~.1 The parameter A. is called the rate of the process. The following
section demonstrates the special properties that follow from the exponential density
function.
Figure 2.1 also shows a family of random variables (N(t); ~O} where N(t), for
each 1>0, is the number of arrivals in the interval from 0 to t. Whether the end points are
included in these intervals is sometimes important, and we use parentheses to represent
intervals without end points and square brackets to represent inclusion of the end point.
Thus (a,b) denotes the interval {t: a<t<b}, and (a,b] denotes {t: a<t!>b}. The counting
random variables N(t) for each 1>0 are then defined as the number of arrivals in the
interval (O,t]. N(O) is defined to be 0 with probability 1, which means that we are
considering only arrivals at strictly positive times.
A counting process (N(t), t~} is a family of non-negative integer valued random
variables, one for each real number t~, with the properties that N('t) ;:: N(t) for all 'C;::t
(i.e., N('C)-N(t) is a non-negative random variable) and N(O) = 0 with probability I. We
interpret the random variable N(t), for each 1>0, as the number of arrivals in (O,t], and
thus, the formal definition of counting process above is simply another way to charac-
terize an arrival process. In summary, then, an arrival process can be specified by the
joint distributions of the arrival epochs, or of the inter-arrival intervals, or of the count-
ing random variables. In principle, specifying anyone of these specifies the others
also. We shall often refer to arrival processes as counting processes, since stochastic
processes are often characterized by a family of random variables, (N(t); r;::0), and a
counting process is then a special case of a stochastic process in which N(t) is non-
negative, integer, and non-decreasing in t.
For any given integer n;::1 and time ~, the nth arrival epoch, So, and the counting
random variable, NCt), are related by

{So:5 t} = {N(t);:: n} (2)

To see this, note that {Sn :5 t} is the event that the nth arrival occurs by time t. This event
implies that N(t), the number of arrivals up to and including t, must be at least n; i.e., it
implies the event (N(t) ;:: n}. Similarly, (N(t) ;:: n} implies {So :5 t}, yielding the equal-
ity in (2). This equation is essentially obvious, but is one of those peculiar obvious
things that is often difficult to see. One should be sure to understand it, since it is
fundamental in going back and forth between arrival epochs and counting random vari-
ables.
Although we have referred to these processes as arrival processes, we could equally
well be modeling departures from a system, or any other sequence of incidents. Al-
though it is quite common, especially in the simulation field, to refer to incidents or
arrivals as events, we shall avoid that here. The nth arrival epoch Sn is a random vari-