Professional Documents
Culture Documents
MVU
E () =
for all
Unbiased E ( ) =
g (x ) p(x; )dx =
n = 0, 1, , N-1
unbiased!
NCTU EE
P.1
MVU
biased!
Bias of estimator = E () = b( )
NCTU EE
P.2
MVU
MSE is composed of errors due to the variance of the estimator as well as the
bias!!
for some constant a.
Example:
NCTU EE
P.3
MVU
Any criterion that depends on the bias will lead to an unrealizable estimator!!
Generally, we constrain the bias to be zero and find the estimator which
minimizes the variance.
Minimum Variance Unbiased (MVU) estimator.
are unbiased.
NCTU EE
P.4
MVU
Remark: Its possible that there may not exist even a single unbiased estimator.
Remarks:
1. If an estimator exists whose variance equals the CRLB for each value of
, then it must be the MVU estimator.
2. It may happen that no estimator exists whose variance equals the bound.
NCTU EE
P.5
MVU
= [
An estimator
1
2
" p ]T is unbiased if
By define
NCTU EE
P.6
MVU
P.7
MVU
When the pdf is viewed as a function of the unknown parameter (with x fixed), it
is termed the likelihood function.
The sharpness of the likelihood function determines how accurately we can
estimate the unknown parameter.
Note:
In general,
2 ln p
is random variable.
2
.
the average curvature of the log-likelihood function.
NCTU EE
P.8
MVU
Remarks:
1.
2. Regularity Condition is satisfied if the order of differentiation and
integration may be interchanged.:
This condition is generally true except when the domain of the pdf for
which it is nonzero depends on the unknown parameter.
An example that does not satisfy the regularity condition:
x[n] are iid according to U[0,], for n = 0, 1, , N-1.
3.
NCTU EE
P.9
MVU
4. Fisher Information
9 Nonnegative
9 Additive for independent observations.
where
As N , CRLB 0 for iid distribution
Proof of Scalar Parameter CRLB:
We consider all unbiased estimator with
That is,
Since
,
we have
.
We now apply the Cauchy-Schwarz inequality
NCTU EE
P.10
MVU
with
If = g() = ,
NCTU EE
P.11
MVU
Q.E.D.
NCTU EE
P.12
MVU
The sample mean estimator attains the bound and must be the MVU
estimator.
Summary: CRLB theorem gives us a lower bound on variance of any unbiased
estimator but also finds estimator that attains bound (if it exists)
Estimators that attains bound are termed efficient (good use of data)
efficient MVU
MVU > efficient
NCTU EE
P.13
MVU
NCTU EE
P.14
MVU
Example:
NCTU EE
P.15
MVU
~ Transformation of Parameters
Assume we wish to estimate = g(), a function of .
Not efficient!
NCTU EE
P.16
MVU
The CRLB is
But we have
achieved!
asymptotically unbiased!
asymptotically efficient!
NCTU EE
P.17
MVU
where
I()
Remark:
w[n]: WGN
=
B
Likelihood function
NCTU EE
P.18
MVU
Remarks:
1. When B is known, we have a lower CRLB.
NCTU EE
P.19
MVU
Remark:
(positive semi-definite)
NCTU EE
P.20
MVU
w[n]: WGN
=
B
NCTU EE
P.21
MVU
g: r-dimensional function
g
g 1
var( i ) [
I ()
] ii
or
r p Jacobian matrix
NCTU EE
P.22
A2
MVU
NCTU EE
P.23
MVU
w[n]: WGN
NCTU EE
P.24
MVU
As N , CRLB 2A4
NCTU EE
P.25
MVU
Pxx(f;) = F{rxx[k]}
NCTU EE
P.26
MVU
For
and Q(f) >> 2,
NCTU EE
P.27
MVU
NCTU EE
P.28
MVU
NCTU EE
P.29
MVU
S(F) = F{s(t)}
F
Ts 2
1 2
exp{2 2 F 2 / F2 } .
If s (t ) = exp{ F (t ) } , S ( F ) =
2
2
2
F =
2
F2
2
NCTU EE
P.30
MVU
Want to estimate A, f0, and , with A > 0 and 0 < f0 < 0.5.
where = [A
where
NCTU EE
P.31
f0
]T
MVU
Linear Models
~ Definition and Properties
Example: Line Fitting
observation matrix
Linear Model:
Assume w ~ N(0,2I).
x ~ N(H,2I)
(A is symmetric)
NCTU EE
P.32
MVU
MVU estimator
Covariance matrix:
Remark: HTH is invertible the columns of H are linearly independent.
Counterexample:
The model parameters will not be identifiable even in the absence of noise.
NCTU EE
P.33
MVU
Assume w(tn) are i.i.d. Gaussian random variables with zero mean and
variance 2.
NCTU EE
P.34
MVU
w[n]: WGN
N 2M
NCTU EE
P.35
(N > 2M)
MVU
~ Gaussian
P.36
MVU
We provide u[n] and measure x[n]. Wish to estimate the weights h[n].
If u[n] is provided for n = 0, 1, .., N-1 and u[n] = 0 for n < 0, we observe
NCTU EE
P.37
MVU
Note that how well we can estimate the tap weights depends on the probing
signal u[n].
The signal should be chosen to be a pseudorandom noise (PRN).
Proof:
Note that
Let
and
NCTU EE
P.38
MVU
Since
For large N,
NCTU EE
P.39
MVU
ln p( x; )
derivation
NCTU EE
P.40
MVU
Recall
The data are first prewhitened to form x[n] and then averaged using
prewhitened averaging weights dn.
NCTU EE
P.41
MVU
NCTU EE
P.42
MVU
as an estimator.
E[ A] = A , Var[ A] = 2
The variance is much larger because the data {x[1], x[2], , x[N-1]} are
not used.
Which data are important or sufficient for estimation?
Consider
may be found.
Once the sufficient statistics are known, we no longer need the individual data
values. All information has been summarized in the sufficient statistic.
The sufficient data set that contains the least number of elements is called the
minimal one.
NCTU EE
P.43
MVU
and assume
To show that
conditional pdf:
where
NCTU EE
P.44
MVU
doesnt depend on A.
Proof:
() Assume the factorization holds.
NCTU EE
P.45
MVU
where
We can let
NCTU EE
P.46
MVU
where
.
Moreover, the pdf of the sufficient statistic can be found based on the
factorization.
A, f0: known.
NCTU EE
w[n]: WGN
P.47
MVU
T1(x) and T2(x) are jointly sufficient statistics for the estimation of .
The r statistics T1(x), T2(x), , Tr(x) are jointly sufficient statistics if the
conditional pdf p(x | T1(x), T2(x), , Tr(x), )h(x) does not depend on .
NCTU EE
P.48
MVU
A = E[ A | T ] .
(2) Find some function g so that A = g (T ) is an unbiased estimator
of A.
N 1
N 1
NCTU EE
P.49
MVU
N 1
Approach 2: A = g ( x[n])
n =0
1 N 1
A = x[n]
N n =0
Much easier to apply!!
Remark: A statistic is complete if there is only one function of the statistic that
is unbiased.
Assume T(x) is complete.
(1) is unique
(2) All produce the same
(3) But variance must be decreased.
is the MVU estimator.
NCTU EE
P.50
MVU
NCTU EE
P.51
MVU
Remarks:
1. In general, it is difficult to validate the completeness of a sufficient
statistic.
2. For the exponential family of PDFs, sufficient statistics are complete.
Example: Completeness of a Sufficient Statistic
N 1
Let T = x[n] .
n =0
Wish to show that if E[g(T)] = A for all A, then there is only one
solution for g.
Suppose there exists a second function h with E[h(T)] = A, A .
Since T ~ N(NA,N2)
NCTU EE
P.52
MVU
P.53
MVU
NCTU EE
P.54
MVU
where
NCTU EE
P.55
MVU
NCTU EE
P.56
MVU
Remarks:
1. The sample mean is not the MVU estimator of the mean for uniformly
distributed noise. For sample mean,
2. In the phase estimation example, we require two sufficient statistics,
T1(x) and T2(x). We need to evaluate
or find g(T1,
NCTU EE
P.57
MVU
NCTU EE
P.58
MVU
Remark: In the vector parameter case, completeness means that for v(T), an
arbitrary r 1 function of T, if
then
x
n
A
x[n] + NA2 )]
N
2
2 n =0
n=0
(2 2 ) 2
jointly sufficient
r = p = 2 we can find MVU estimator.
NCTU EE
P.59
Since
MVU
P.60
MVU
Pf:
Alternate approach:
where
NCTU EE
P.61
MVU
NCTU EE
P.62
MVU
For some estimation problems, the use of a BLUE can be totally inappropriate.
Example: estimation of the power of WGN.
MVU:
If we force the estimator to be linear,
NCTU EE
P.63
MVU
or
where s = [s[0] s[1] s[N-1]]T
Solution: Use the method of Lagrangian multipliers:
NCTU EE
P.64
MVU
NCTU EE
P.65
MVU
Remarks:
1. The sample mean is the BLUE independent of the pdf of the data.
2. For Gaussian data, the sample mean is the MVU estimator.
NCTU EE
P.66
MVU
The BLUE weights those samples most heavily with smallest variances.
Remark: Our model E(x[n]) = s[n] is actually the general linear model
To be unbiased,
NCTU EE
P.67
MVU
Let
so that
Assume
Remark: The form of the BLUE is identical to the MVU estimator for the
, where w ~ N(0,C)
NCTU EE
P.68
MVU
Assume N antennas have been placed at known locations and the time of
arrival measurements ti for i = 0, 1, , N-1 are available.
The arrival times are corrupted by noise with zero mean and known covariance
but otherwise unknown pdf.
NCTU EE
P.69
MVU
The source is near the nominal position (xn yn)T, which has been
obtained from previous measurements.
Let
, we have
NCTU EE
P.70
MVU
e.g., N = 3
NCTU EE
P.71
NCTU EE
MVU
P.72