Let be the magnitude-squared of the DFT bins
of a size- DFT. Then, .
Let have mean
. Then,
the distribution of the zero and Nyquist
frequency bins () follows the Chi-squared distribution
with 1 degree of freedom scaled by ,
DFT bins 1 through are complex
so have the Chi-squared distribution with 2 degrees of
freedom scaled by
, or equivalently
the exponential distribution with mean :
The joint PDF of
is obtained from the product
of the above bin densities. Interestingly, maximizing
over
will also maximize the PDF of the input data to the
DFT. Let be the length- input to the DFT,
then
|
(5.6) |
where are the DFT bins. Therefore, we can replace
with :
|
(5.7) |
It is mathematically simpler to maximize (5.7) instead
of
.
Although (5.7) is a function of
, we need to
evaluate it for a particular value of the parameters
and .
In Section 5.2.5, we showed how to compute
from the AR parameters.