7.2 Ergodic Random Processes a
te Chaper 6, Problem 69, we defined a random process of de voltage waveforms, where the prob. | Example 7-2
ability density function for the de voltage was a uniform probability deasity from zero o 96 that
‘ean was a/2. The ensemble of member functions is shown in Figure 6.25. Tis process is
Strict sense stationary. The mean estimated by time averaging a member function depends onthe
member function, For example, the mean estimated by time averaging might be a, or ay, which
f22y not be 2. In other words, the slatisties for a member function are not the same us those for
ensemble. Thus, the mean estimated by time averaging will not always converge to the en.
able average mean. Although this random process strict sense stationary, itis not ergodic in
An alternative definition for ergodic in the mean is given next.
Definition 7.2. A wide sense stationary random process X(t) is ergodi
if and only if
Taf eee ee
insef (: ar} Covsxle) dr =o 221
Recall from Example 6.28 thatthe reason forthe limits arises from the product of
%o integrals, This definition is the same as saying that the estimate for the mean
onverges to the true mean in the mean square sense, namely
1
imp BL x — wx
fim yp Bll tx — ws 721
Definition 7.2 canbe stated as a theorem and proven from [7.2.3] in a manne
to that shown in Chapter 6, Example 6.28 and Problem 6.35.
sm Definition 7.2 one can define other definitions of ergodicity
Definition 7.3, A wide sense stationary random process X(t) is ergodic in the auto-
‘ecrelation if and only if for all +
if
X()X(t+ 7) > = Rex(r) Finag[ ,XCOX( + dt = Ryy(r) (7.2.4)
PE= < X(OXit+7) > denotes the estimate of the autocorrelation function by
Bie averaging.
over the Rxx(7) is used to emphasize tha this is an estimate ofthe autocorre-
Section, which in this case, is calculated by time averaging as distinguished from
raging. Various symbols ate used to denote an estimate of the autocorrelation
cluding &, R, and, 31. Note that the above definitions can be easily modified to
sedate random sequences. For example, the definition for ergodic in the mean
Beas follows.
tion 7.4. A wide sense stationary random sequence X(n) is ergodic in the
‘and only if