Добавил:
Upload Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

Diss / 10

.pdf
Скачиваний:
143
Добавлен:
27.03.2016
Размер:
18.05 Mб
Скачать

Estimation of Stochastic Process Variance

453

Let us define the variance of the variance estimate caused by the finite observation time interval of stochastic process. Let the investigated stochastic process be the stationary one and we use the ideal integrator as the smoothing filter. In this case, the variance estimate is carried out in accordance with (13.46) and the variance of the variance estimate takes the following form:

 

1

 

 

T T

 

2

T T T

 

 

 

 

 

 

 

 

 

Var{Var*} =

 

 

∫∫ x2 (t1)x2 (t2 ) dt1dt2

 

∫∫∫ x2 (t1)x(t2 )x(t3 ) dt1dt2dt3

 

 

 

 

T 2

 

T 3

 

 

 

 

 

 

 

 

 

0

0

 

 

0

0

0

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T T T T

 

 

 

 

 

 

 

T

 

 

τ

 

2

 

 

 

1

 

 

 

 

 

4 2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

+

 

 

∫∫∫∫

x(t1)x(t2 )x(t3 )x(t4 ) dt1dt2dt3dt4 − σ

1

 

 

 

(τ)dτ .

 

T

4

T

T

 

 

 

 

 

 

 

1

 

 

 

 

 

 

 

 

0 0 0 0

 

 

 

 

 

 

 

0

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(13.60)

Determination of moments in the integrands (13.60) is impossible for the arbitrary pdf of the investigated stochastic process. For this reason, to define the main regularities we assume that the investigated­ stochastic process is Gaussian with the unknown mathematical expectation, the true value of which is E0. Then, after the corresponding transformations, we have

 

4σ4

T

 

 

 

τ

 

 

Var{Var*} =

 

 

 

 

2

 

 

 

 

1

 

 

 

 

T

 

 

 

 

 

 

 

T

 

 

 

 

 

0

 

 

 

 

 

 

 

 

1

T T T (t1 t2

 

T 2

 

 

 

0

0

0

 

 

 

 

 

 

2

T

 

 

τ

 

(τ)dτ +

 

(τ

 

 

1

 

 

 

T

 

 

 

 

 

 

T

 

 

 

0

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

) (t1 t3 )dt1dt2dt3

 

 

.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 )dτ

(13.61)

If the mathematical expectation of the investigated stochastic process is known accurately, then according to (13.46) the variance estimate is unbiased and the variance of the variance estimate takes the following form:

Var{Var*} =

4σ4 T

1

τ

 

2

(τ)dτ.

(13.62)

 

 

 

 

 

T

 

 

 

 

 

T

 

 

 

 

 

 

0

 

 

 

 

 

 

 

 

When the observation time interval is sufficiently large, that is, the condition T τcor is satisfied, (13.62) is simplified and takes the following form:

 

4

2

 

4τcor

 

 

Var{Var*} ≈

 

(τ)dτ ≤

 

.

(13.63)

T

T

 

 

0

 

 

 

 

 

Using (12.27) and (12.122), at the condition T τcor, variance of the variance estimate can be presented in the following form:

Var{Var*} ≈

2

R2 (τ)dτ =

2

S2 (ω)dω.

(13.64)

T

πT

 

 

−∞

0

 

454

Signal Processing in Radar Systems

As applied to the exponential correlation function given by (12.13), the normalized variance of the variance estimate can be presented in the following form:

Var{Var*}

=

2p − 1+ exp{−2p}

,

(13.65)

σ4

 

p2

 

 

 

where p is given by (12.48). Analogous formulae for the variance of the variance estimate can be obtained when the stochastic process is sampling. Naturally, at that time, the simplest expressions are obtained in the case of independent readings of the investigated stochastic process.

Determine the variance of the variance estimate of stochastic process according to (13.51) for the case of independent samples. For this case, we transform (13.51) in the following form:

 

1

 

 

N

 

1

 

 

 

2

 

Var* =

 

 

yi

 

N 1

N

 

 

 

 

 

 

 

 

 

 

 

 

 

i=1

 

 

N

2

 

 

 

 

 

 

,

(13.66)

yp

 

 

 

 

 

 

p=1

 

 

 

 

where yi = xi E. As we can see, the variance estimate is unbiased. The variance of the variance estimate of stochastic process can be determined in the following form:

Var{Var*} =

1

(N 1)2

 

N

i=1, j =1

 

2

N

 

1

N

yi2 y2j

yi2 yp yq +

N

N 2

 

 

i=1, p=1,q=1

 

 

i=1, j =1, p=1,q

 

 

 

 

 

 

 

− σ

4

.

 

yi yj yp yq

 

=1

 

 

 

 

 

 

 

 

(13.67)

To compute the sums in the brackets we select the terms with the same indices. Taking into consideration the independence of the samples, we have

N

N

N

 

 

yi2 y2j

= yi4

+ yi2

y2j = Nµ4 + N(N − 1)µ22 ,

(13.68)

i=1, j =1

i=1

i=1, j =1

 

 

 

 

ij

 

 

where

 

 

 

 

 

µνi

= yiν = (xi

E)ν

(13.69)

is the central moment of the νth order and, naturally, μ2 = σ2. Analogously, we obtain

 

 

N

 

 

 

 

yi2 yp yq = Nµ4 + N(N − 1)µ22;

(13.70)

 

i=1, p=1,q=1

 

 

 

 

N

 

 

 

 

yi yj yp yq = Nµ4 + 3N(N − 1)µ22.

(13.71)

i=1, j =1, p=1,q=1

Estimation of Stochastic Process Variance

455

Substituting (13.70) and (13.71) into (13.67), we obtain

Var{Var*} = µ4 {1 − (2/(N − 1))}σ2 .

N

In the case of Gaussian stochastic process we have μ4 = 3σ4. As a result we obtain

Var{Var*} = 4 .

N − 1

If the mathematical expectation is known, then the unbiased variance estimate is defined as

 

 

N

Var* =

1

(xi E0 )2 ,

N

 

i=1

 

 

and the variance of the variance estimate takes the following form:

Var{Var*} = µ4 − σ4 .

N

(13.72)

(13.73)

(13.74)

(13.75)

In the case of Gaussian stochastic process, variation in the variance estimate has the following form:

*

4

(13.76)

N .

Var{Var } =

 

Comparing (13.76) with (13.73), we can see that at N 1 these formulas are coincided.

As applied to the iterative variance estimate of stochastic process with zero mathematical expectation, based on (12.342) and (12.345) we obtain

• In the case of discrete stochastic process

Var*[N] = Var*[N − 1] + γ[N]{x2[N] − Var*[N − 1]};

(13.77)

• In the case of continuous stochastic process

d Var*(t)

= γ (t){x2 (t) − Var*(t)}.

(13.78)

dt

 

 

As shown in Section 12.9, we can show that the optimal value of the factor γ[N] is equal to N−1.

13.3  ERRORS UNDER STOCHASTIC PROCESS VARIANCE ESTIMATE

As we can see from formulas for the variance estimate of stochastic process, a squaring of the stochastic process realization (or its samples) is a very essential operation. The device carrying out this operation is called the quadratic transformer or quadrator. The difference in quadrator performance from square-law function leads to additional errors arising under measurement of the stochastic process variance. To define a character of these errors we assume that the stochastic

456

Signal Processing in Radar Systems

process possesses zero mathematical expectations and the variance estimate is carried out based on investigation of independent samples. The characteristic of transformation y = g(x) can be presented by the polynomial function of the μth order

 

 

 

y = g(x) = ak xk .

(13.79)

k =

0

 

Substituting (13.79) into (13.59) instead of xi and carrying out averaging, we obtain the mathematical expectation of variance estimate:

 

 

 

 

 

Var* = ak xk = a0 + a2σ2 + ak xk .

(13.80)

k =

0

k =

3

 

The estimate bias caused by the difference between the transformation characteristic and the squarelaw function can be presented in the following form:

 

 

 

b{Var*} = σ4 (a2 1) + a0 + ak xk .

(13.81)

k =

3

 

For the given transformation characteristic the coefficients ak can be defined before. For this reason, the bias of variance estimate caused by the coefficients a0 and a2 can be taken into consideration. The problem is to take into consideration the sum in (13.81) because this sum depends on the shape of pdf of the investigated stochastic process:

xk = xk f ( x)dx.

(13.82)

−∞

 

In the case of Gaussian stochastic process, the moment of the kth order can be presented in the following form:

 

k

 

1

3 5 (k 1)σ0.5k ,

if

k is even;

x

 

 

 

 

(13.83)

 

=

 

 

 

 

 

0,

 

if

k is odd.

 

 

 

 

 

 

 

Difference between the transformation characteristic and the square-law function can lead to high errors under definition of the stochastic process variance. Because of this, while defining the stochastic process variance we need to pay serious attention to transformation performance. Verification of the transformation performance is carried out, as a rule, by sending the harmonic signal of known amplitude at the measurer input.

While measuring the stochastic process variance, we can avoid the squaring operation. For this purpose, we need to consider two sign functions (12.208)

η1(t) = sgn[ξ(t) − µ1

(t)]

(13.84)

 

 

,

η2

(t) = sgn[ξ(t) − µ2 (t)]

 

 

 

 

 

Estimation of Stochastic Process Variance

457

instead of the initial stochastic process ξ(t) with zero mathematical expectation, where μ1(t) and μ2(t) are additional independent stationary stochastic processes with zero mathematical expectations and the same pdf given by (12.205). At the same time, the condition given by (12.206) is satisfied.

The functions η1(t) and η2(t) are the stationary stochastic processes with zero mathematical expectations. In doing so, if the fixed value ξ(t) = x the conditional stochastic processes η1(t|x) and η2(t|x) are statistically independent, that is,

η1(t|x)η2 (t|x) = η1(t|x) η2 (t|x) .

(13.85)

Taking into consideration (12.209), we obtain

 

 

 

η1(t|x)η2 (t|x) =

x2

.

(13.86)

 

 

A2

 

The unconditional mathematical expectation of product can be presented in the following form:

 

1

σ2

 

η1(t|x)η2 (t|x) =

x2 p(x)dx =

 

 

 

.

(13.87)

A2

A2

 

 

−∞

 

 

 

Denote the realizations of stochastic processes η1(t) and η2(t) by the functions y1(t) and y2(t), respectively. Consequently, if we consider the following value as the variance estimate,

 

A2

T

 

Var =

 

y1(t)y2 (t)dt,

(13.88)

T

 

 

0

 

then the variance estimate is unbiased.

The realizations y1(t) and y2(t) take the values equal to ±1. For this reason, the multiplication and integration in (13.88) can be replaced by individual integration of new unit functions obtained as a result of coincidence and noncoincidence of polarities of the realizations y1(t) and y2(t):

T

y1(t)y2 (t)dt = T z1(t)dt T z2 (t)dt,

(13.89)

0

 

 

 

0

0

 

where

 

 

 

 

 

 

 

 

 

if

y1(t) > 0, y2 (t) > 0;

 

 

 

1

 

 

 

z1(t) =

 

 

y1(t) < 0,

y2 (t) < 0;

(13.90)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0

otherwise;

 

 

 

 

 

 

 

 

 

 

 

 

if

y1(t) > 0, y2 (t) < 0;

 

 

 

1

 

 

 

z2

(t) =

 

 

y1(t) < 0,

y2 (t) > 0;

(13.91)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0

otherwise.

 

 

 

 

 

 

 

 

 

458

Signal Processing in Radar Systems

u1(t)

y1(t)

z1(t)

 

 

 

 

 

0

 

+

sgn

1

 

 

 

Matching

 

 

 

 

 

T T z1(t)dt

k = A2

 

 

 

 

 

+

 

 

 

 

 

>

 

 

 

z2(t)

 

 

 

sgn

 

1

0

 

+

y2(t)

No matching

z2(t)dt

Var

 

 

T T

 

u2(t)

x(t)

FIGURE 13.4  Measurer based on additional signals.

The flowchart of measuring device based on the implementation of additional signals is shown in Figure 13.4.

In the case of discrete process, the variance estimate can be presented in the following form:

 

 

N

 

 

A2

y1i y2i ,

 

Var =

 

(13.92)

N

 

i=1

 

 

 

 

where y1i = y1(ti) and y2i = y2(ti). In this case, the integrator can be replaced by the summator in Figure 13.4.

Determine the variance of the variance estimate Var applied to investigation of stochastic process at discrete instants for the purpose of simplifying further analysis that the samples y1i and y2i are independent. Otherwise, we should know the two-dimensional pdfs of additional uniformly distributed stochastic processes μ1(t) and μ2(t). The variance of the variance estimate can be presented in the following form:

 

 

A4

N

 

 

 

 

 

 

2

 

 

Var{Var}

=

 

y1i y1j y2i y2 j Var

 

.

(13.93)

N 2

 

 

 

 

i=1, j =1

 

 

 

 

In double sum, we can select the terms with i = j. Then

 

 

 

 

N

 

 

N

N

 

 

 

y1i y1j y2i y2 j = (y1i y2i )2 +

y1i y1j

y2i y2 j ,

(13.94)

i=1, j =1

 

 

i=1

i=1, j =1

 

 

 

 

 

 

 

ij

 

 

 

where

 

 

 

 

 

 

 

y1i

y1j y2i

y2 j = η1i η1j η2i η2 j .

 

 

(13.95)

Based on a definition of sign functions, we can obtain from (13.84) the following condition:

[ y1(t)y2 (t)]2 = 1.

(13.96)

Define the cumulative moment  η1iη1jη2iη2j|xixj  at the condition that ξ(ti) = xi and ξ(tj) = xj. Taking into consideration the statistical independence of the samples η1 and η2, the statistical mutual

Estimation of Stochastic Process Variance

459

independence of the stochastic values η1(t|x) and η2(t|x), and (12.209), the conditional cumulative moment can be presented in the following form:

η1

η1

 

η2

 

η2

 

|xi x j =

xi2 x2j

.

(13.97)

 

 

 

 

i

 

j

 

i

 

j

 

A4

 

Averaging (13.97) by possible values of independent random variables xi and xj, we obtain

η1i η1j η2i η2 j =

xi2 x

2j

=

xi2 x2j

=

σ4

(13.98)

A4

 

A4

A4 .

 

 

 

 

 

 

 

 

Substituting (13.96) and (13.98) into (13.93), we obtain the variance of the variance estimate (13.92)

Var{Var} =

A4

 

 

σ4

 

(13.99)

 

 

1

 

 

.

N

A

4

 

 

 

 

 

 

 

The variance estimate according to (13.92) envisages that the condition (12.206) is satisfied. Consequently, the inequality σ2 A2 must be satisfied. Because of this, as in the case of estimation of the mathematical expectation with employment of additional signals, the variance of the variance estimate Var is completely defined by the half intervals of possible values of additional random sequences. Comparing the obtained variance (13.99) with the variance of the variance estimate in (13.76) by N independent samples, we can find that

 

 

A

4

 

 

 

σ

4

 

 

Var{Var}

=

 

 

 

 

 

 

 

 

 

 

1

 

 

.

(13.100)

2σ

4

A

4

Var{Var }

 

 

 

 

 

 

 

 

Thus, we can conclude that the method of measurement of the stochastic process variance using the additional signals is characterized by the higher variance compared to the algorithm (13.74). This is based on the example of definition of the variance estimate of stochastic sample with the uniform pdf coinciding with (12.205), σ2 = A2/3. As a result, we have

Var{Var} = 4. (13.101)

Var{Var }

The methods that are used to measure the variance assume the absence of limitations of instantaneous values of the investigated stochastic process. The presence of these limitations leads to additional errors while measuring the variance. Determine the bias of variance estimate of the Gaussian stochastic process when there is a limiter of the type (12.151) presented in Figure 12.6 applied to zero mathematical expectation and the true value of the variance σ2. The variance estimate is defined by (13.41). The ideal integrator h(t) = T−1 plays a role in averaging or smoothing filter. Substituting the realization y(t) = g[x(t)] into (13.41) instead of x(t) and carrying out averaging of the obtained formula with respect to x(t), the bias of variance estimate of Gaussian stochastic process can be written in the following form:

 

 

2

 

γ

 

 

 

 

2

2

2

2

 

*

 

(x) p(x)dx − σ

= −2σ (1 − γ )Q(γ ) −

 

exp{−0.5γ } ,

(13.102)

 

 

b{Var } = g

 

 

−∞

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

460

Signal Processing in Radar Systems

b{Var*} σ2

1.0

0.8

0.6

0.4

0.2

00.1

 

 

 

 

 

 

 

 

 

γ

 

 

 

 

 

 

 

 

 

 

0.2

0.5

1.0

2.0

3.0

FIGURE 13.5  Bias of variance estimate of Gaussian stochastic process as a function of the normalized limitation level.

where

γ =

a

(13.103)

σ

 

 

is the normalized limitation level. At γ  1, we obtain that the bias of variance estimate tends to approach zero, which is expected. The bias of variance estimate of Gaussian stochastic process as a function of the normalized limitation level is shown in Figure 13.5.

13.4  ESTIMATE OF TIME-VARYING STOCHASTIC PROCESS VARIANCE

To define the current value of nonstationary stochastic process variance there is a need to have a set of realizations xi(t) of this process. Then the variance estimate of stochastic process at the instant t0 can be presented in the following form:

N

Var*(t0 ) = N1 [xi (t0 ) − E(t0 )]2 , (13.104)

i=1

where

N is the number of realizations xi(t) of stochastic process

E(t0) is the mathematical expectation of stochastic process at the instant t0

As we can see from (13.104), the variance estimate is unbiased and the variance of the variance estimate under observation of N independent realizations can be presented in the following form:

Var{Var*(t0 )} =

Var2 (t0 )

;

(13.105)

 

N

 

 

that is, the variance estimate according to (13.104) is the consistent estimate.

Estimation of Stochastic Process Variance

461

In most cases, the researcher defines the variance estimate on the basis of investigation of a single realization of stochastic process. While estimating time-varying variance based on a single realization of time-varying stochastic process, there are similar problems as seen with the estimation of the time-varying mathematical expectation. On the one hand, to decrease the variance estimate caused by the finite observation time interval, the last must be as soon as large. On the other hand, we need to choose the integration time as short as possible for the best definition of variations in variance. Evidently, there must be a compromise.

The simplest way to define the time-varying stochastic process variance at the instant t0 is averaging the transformed input data of stochastic process within the limits of finite time interval. Thus, let x(t) be the realization of stochastic process ξ(t) with zero mathematical expectation. Measurement of variance of this stochastic process at the instant t0 is carried out by averaging the quadrature ordinates x(t) within the limits of the interval about the given value of argument [t0 − 0.5T, t0 + 0.5T]. In this case, the variance estimate takes the following form:

Var*(t0 ,T ) =

1

t0

+0.5T x2 (t)dt =

1

0.5T

x2 (t + t0 )dt.

(13.106)

T

T

 

 

t0

− 0.5T

−0.5T

 

 

Averaging the variance estimate by realizations, we obtain

 

1

0.5T

 

Var*(t0 ,T ) =

Var(t + t0 )dt.

(13.107)

T

 

 

−0.5T

 

Thus, as in the case of time-varying mathematical expectation, the estimate of the variance of the time-varying mathematical expectation does not coincide with its true value in a general case, but it is obtained by smoothing the variance within the limits of finite time interval [t0 − 0.5T, t0 + 0.5T]. As a result of such averaging, the bias of variance of stochastic process can be presented in the following form:

 

1

0.5T

 

b{Var*(t0 ,T )}=

[Var(t + t0 ) − Var(t0 )]dt.

(13.108)

T

 

 

−0.5T

 

If we would like to have the of time interval [t0 − 0.5T, t0 function of time

unbiased estimate, a variance of the variance Var(t) within the limits + 0.5T] must be the odd function and, in the simplest case, the linear

Var(t0 + t) ≈ Var(t0 ) + Var′(t0 )t, −0.5T t ≤ 0.5T.

(13.109)

Evaluate the influence of deviation of the current variance Var(t) from linear function on its estimate. In doing so, we assume that the estimated current variance has the continuous first and second derivatives with respect to time. Then, according to the Taylor expansion we can write

Var(t) = Var(t0 ) + (t t0 )Var′(t) + 0.5(t t0 )2 Var′′[t0 + θ(t t0 )],

(13.110)

462

 

Signal Processing in Radar Systems

where 0 < θ < 1. Substituting (13.110) into (13.108), we obtain

 

 

1

0.5T

 

b{Var*(t0 ,T )}=

t2Var′′(t0 + tθ)dt.

(13.111)

2T

 

 

−0.5T

 

Denoting MVar the maximal absolute value of the second derivative of the current variance Var(t) with respect to time, we obtain the high bound value of the variance estimate bias by absolute value

b{Var*(t0 ,T )}

T 2 MVar .

(13.112)

 

24

 

The maximum value of the second derivative of the time-varying variance Var(t) can be evaluated, as a rule, based on the analysis of specific physical problems.

To estimate the difference in the time-varying variance a square of the investigated realization of the stochastic process can be presented in the following form of two sums:

ξ2 (t) = Var(t) + ζ(t),

(13.113)

where

Var(t) is the true value of the stochastic process variance at the instant t

ζ(t) are the fluctuations of square of the stochastic process realization with respect to its mathematical expectation at the same instant t

Then the variation in the variance estimate of the investigated nonstationary stochastic process can be presented in the following form:

 

1

0.5T

0.5T

 

Var{Var*(t0 ,T )} =

 

Rζ (t1 + t0 ,t2 + t0 )dt1dt2 ,

(13.114)

T 2

 

 

−0.5T −0.5T

 

where

R (t1,t2 ) =

x2

(t ) Var(t ) x2

(t

2

) Var(t

2

)

(13.115)

ζ

 

1

1

 

 

 

is the correlation function of the random component of the investigated squared stochastic process with respect to its variance at the instant t.

To define the approximate value of the variance of slowly varying in time variance estimate we can assume that the centralized stochastic process ζ(t) within the limits of the finite time interval [t0 − 0.5T, t0 + 0.5T] is the stationary stochastic process with the correlation function defined as

Rζ (t, t + τ) ≈ Varζ (t0 ) ζ (τ).

(13.116)

Соседние файлы в папке Diss