Stochastic processes Lecture 7 Linear time invariant systems 1.

59
Stochastic processes Lecture 7 Linear time invariant systems 1

Transcript of Stochastic processes Lecture 7 Linear time invariant systems 1.

Page 1: Stochastic processes Lecture 7 Linear time invariant systems 1.

1

Stochastic processes

Lecture 7Linear time invariant systems

Page 2: Stochastic processes Lecture 7 Linear time invariant systems 1.

2

Random process

Page 3: Stochastic processes Lecture 7 Linear time invariant systems 1.

3

1st order Distribution & density function

First-order distribution

First-order density function

Page 4: Stochastic processes Lecture 7 Linear time invariant systems 1.

4

2end order Distribution & density function

2end order distribution

2end order density function

Page 5: Stochastic processes Lecture 7 Linear time invariant systems 1.

5

EXPECTATIONS

β€’ Expected value

β€’ The autocorrelation

Page 6: Stochastic processes Lecture 7 Linear time invariant systems 1.

6

Some random processes

β€’ Single pulseβ€’ Multiple pulsesβ€’ Periodic Random Processesβ€’ The Gaussian Processβ€’ The Poisson Processβ€’ Bernoulli and Binomial Processesβ€’ The Random Walk Wiener Processesβ€’ The Markov Process

Page 7: Stochastic processes Lecture 7 Linear time invariant systems 1.

7

Recap: Power spectrum density

f

Sxx

(f)

Page 8: Stochastic processes Lecture 7 Linear time invariant systems 1.

8

Power spectrum density

β€’ Since the integral of the squared absolute Fourier transform contains the full power of the signal it is a density function.

β€’ So the power spectral density of a random process is:

β€’ Due to absolute factor the PSD is always real

𝑆π‘₯π‘₯ ( 𝑓 )= 𝑙 π‘–π‘šπ‘‡β†’βˆž

𝐸 [|βˆ«βˆ’π‘‡π‘‡ 𝑠 (𝑑 )π‘’βˆ’ 𝑗 2πœ‹ 𝑓𝑑𝑑𝑑|22𝑇 ]

Page 9: Stochastic processes Lecture 7 Linear time invariant systems 1.

9

Power spectrum density

β€’ The PSD is a density function.– In the case of the random process the PSD is the density

function of the random process and not necessarily the frequency spectrum of a single realization.

β€’ Example– A random process is defined as

– Where Ο‰r is a unifom distributed random variable wiht a range from 0-Ο€

– What is the PSD for the process and – The power sepctrum for a single realization

X (𝑑 )=sin (πœ”π‘Ÿ 𝑑)

Page 10: Stochastic processes Lecture 7 Linear time invariant systems 1.

10

Properties of the PSD

1. Sxx(f) is real and nonnegative

2. The average power in X(t) is given by:

3. If X(t) is real Rxx(Ο„) and Sxx(f) are also even

4. If X(t) has periodic components Sxx(f)has impulses

5. Independent on phase

Page 11: Stochastic processes Lecture 7 Linear time invariant systems 1.

11

Wiener-Khinchin 1

β€’ If the X(t) is stationary in the wide-sense the PSD is the Fourier transform of the Autocorrelation

Page 12: Stochastic processes Lecture 7 Linear time invariant systems 1.

12

Wiener-Khinchin Two method for estimation of the PSD

X(t)

Fourier Transform

|X(f)|2

Sxx(f)

Autocorrelation

Fourier Transformt

X(t

)

f

X(f

)

Rxx

()

f

Sxx

(f)

Page 13: Stochastic processes Lecture 7 Linear time invariant systems 1.

13

The inverse Fourier Transform of the PSD

β€’ Since the PSD is the Fourier transformed autocorrelation

β€’ The inverse Fourier transform of the PSD is the autocorrelation

Page 14: Stochastic processes Lecture 7 Linear time invariant systems 1.

14

Cross spectral densities

β€’ If X(t) and Y(t) are two jointly wide-sense stationary processes, is the Cross spectral densities

β€’ Or

Page 15: Stochastic processes Lecture 7 Linear time invariant systems 1.

15

Properties of Cross spectral densities

1. Since is

2. Syx(f) is not necessary real

3. If X(t) and Y(t) are orthogonal Sxy(f)=0

4. If X(t) and Y(t) are independent Sxy(f)=E[X(t)] E[Y(t)] Ξ΄(f)

Page 16: Stochastic processes Lecture 7 Linear time invariant systems 1.

16

Cross spectral densities example

β€’ 1 Hz Sinus curves in white noise

Where w(t) is Gaussian noise

0 5 10 15 20-10

0

10

X(t

)

t (s)

Signal X(t)

0 5 10 15 20-10

0

10

Y(t

)

t (s)

Signal Y(t)

𝑋 (𝑑 )=sin (2πœ‹ 𝑑 )+3𝑀 (𝑑)π‘Œ (𝑑 )=sin(2πœ‹π‘‘+πœ‹2 )+3𝑀(𝑑)

0 5 10 15 20 25-30

-25

-20

-15

-10

-5

0

5

Frequency (Hz)

Pow

er/f

requ

ency

(dB

/Hz)

Welch Cross Power Spectral Density Estimate

Page 17: Stochastic processes Lecture 7 Linear time invariant systems 1.

17

The periodogramThe estimate of the PSD

β€’ The PSD can be estimate from the autocorrelation

β€’ Or directly from the signal

𝑆 π‘₯π‘₯ [Ο‰ ]= βˆ‘π‘š=βˆ’π‘+1

π‘βˆ’ 1

𝑅π‘₯π‘₯ [π‘š]π‘’βˆ’ 𝑗 Ο‰π‘š  

𝑆 π‘₯π‘₯ [Ο‰ ]= 1𝑁 |βˆ‘

𝑛=0

π‘βˆ’ 1

π‘₯ [𝑛]π‘’βˆ’ 𝑗ω𝑛  |2

Page 18: Stochastic processes Lecture 7 Linear time invariant systems 1.

18

Bias in the estimates of the autocorrelation

N=12𝑅π‘₯π‘₯ [π‘š ]= βˆ‘

𝑛=0

π‘βˆ’|π‘š|βˆ’ 1

π‘₯ [𝑛 ] π‘₯ [𝑛+π‘š]

-10 -5 0 5 10 15 20-2

-1

0

1

2

n

-10 -5 0 5 10 15 20-2

-1

0

1

2

n+m-15 -10 -5 0 5 10 15

-6

-4

-2

0

2

4

6

8Autocorrelation

M=-10

-10 -5 0 5 10 15 20-2

-1

0

1

2

n

-10 -5 0 5 10 15 20-2

-1

0

1

2

n+m-15 -10 -5 0 5 10 15

-6

-4

-2

0

2

4

6

8Autocorrelation

M=0

-10 -5 0 5 10 15 20-2

-1

0

1

2

n

-10 -5 0 5 10 15 20-2

-1

0

1

2

n+m-15 -10 -5 0 5 10 15

-6

-4

-2

0

2

4

6

8Autocorrelation

M=4

Page 19: Stochastic processes Lecture 7 Linear time invariant systems 1.

19

Variance in the PSD

β€’ The variance of the periodogram is estimated to the power of two of PSD

π‘‰π‘Žπ‘Ÿ (𝑆π‘₯π‘₯ [πœ” ] )=𝑆 π‘₯π‘₯(πœ”)  2

0 5 10-5

0

5Realization 1

t (s)0 50 100 150 200

0

5

10PSD: Realization 1

f (Hz)

0 5 10-5

0

5

t (s)

Realization 2

0 50 100 150 2000

5

10

f (Hz)

PSD: Realization 2

0 5 10-5

0

5

t (s)

Realization 3

0 50 100 150 2000

5

10

f (Hz)

PSD: Realization 3 0 50 100 150 200

0

0.2

0.4

0.6

0.8

1

f (Hz)

Sxx

(f)

True PSD

Page 20: Stochastic processes Lecture 7 Linear time invariant systems 1.

20

Averaging

β€’ Divide the signal into K segments of M length

β€’ Calculate the periodogram of each segment

β€’ Calculate the average periodogram

Page 21: Stochastic processes Lecture 7 Linear time invariant systems 1.

21

Illustrations of Averaging

0 1 2 3 4 5 6 7 8 9 10-4

-2

0

2X

(t)

0 100 2000

5

10

0 100 2000

2

4

0 100 2000

5

10

0 100 2000

2

4

6

0 50 100 150 2000

5

10

f (Hz)

Page 22: Stochastic processes Lecture 7 Linear time invariant systems 1.

22

PSD units

β€’ Typical units:β€’ Electrical measurements: V2/Hz or dB V/Hzβ€’ Sound: Pa2/Hz or dB/Hz

β€’ How to calculate dB I a power spectrum:PSDdB(f) = 10 log10 { PSD(f) }

.

-100 -50 0 50 1000

1

2

3

4

5

6x 10

8

f (Hz)

Sxx

(f)

V2 /

Hz

-100 -50 0 50 10040

50

60

70

80

90

f (Hz)

Sxx

(f)

dB V

/H

z

Page 23: Stochastic processes Lecture 7 Linear time invariant systems 1.

23

Agenda (Lec. 7)

β€’ Recap: Linear time invariant systemsβ€’ Stochastic signals and LTI systems

– Mean Value function– Mean square value – Cross correlation function between input and output– Autocorrelation function and spectrum output

β€’ Filter examples β€’ Intro to system identification

Page 24: Stochastic processes Lecture 7 Linear time invariant systems 1.

24

Focus continuous signals and system

Continuous signal:

Discrete signal:

0 20 40 60 80 100-1

0

1

t (s)

x(t)

0 2 4 6 8 10-1

-0.5

0

0.5

1

n

x[n]

Page 25: Stochastic processes Lecture 7 Linear time invariant systems 1.

25

Systems

Page 26: Stochastic processes Lecture 7 Linear time invariant systems 1.

26

Recap: Linear time invariant systems (LTI)

β€’ What is a Linear system:– The system applies to superposition

)()()()( 2121 txTbtxTatxbtxaT

0 1 2 3 4 50

2

4

6

8

10

12

14

16

18

20Linear system

x(t)

y(t)

0 1 2 3 4 5-20

-15

-10

-5

0

5

10

15

20

25Nonlinear systems

x(t)

y(t)

x[n]2

Γ–x[n]

20 log(x[n])

Page 27: Stochastic processes Lecture 7 Linear time invariant systems 1.

27

Recap: Linear time invariant systems (LTI)

β€’ Time invariant:β€’ A time invariant systems is independent on explicit time

– (The coefficient are independent on time)

β€’ That means If: y2(t)=f[x1(t)]

Then: y2(t+t0)=f[x1(t+t0)]

The same to Day tomorrow and in 1000 years

70 years45 years20 yearsA non Time invariant

Page 28: Stochastic processes Lecture 7 Linear time invariant systems 1.

28

Examples

β€’ A linear systemy(t)=3 x(t)

β€’ A nonlinear systemy(t)=3 x(t)2

β€’ A time invariant system y(t)=3 x(t)

β€’ A time variant system y(t)=3t x(t)

Page 29: Stochastic processes Lecture 7 Linear time invariant systems 1.

The impulse response

T{βˆ™}

)]([][ tfnh )(th)(t

The output of a system if Dirac delta is input

-10 -5 0 5 10 15 20

0

t

y(t)

Impuls response

-10 -5 0 5 10 15 200

inf

t

x(t)

Impuls

Page 30: Stochastic processes Lecture 7 Linear time invariant systems 1.

30

Convolution

β€’ The output of LTI system can be determined by the convoluting the input with the impulse response

Page 31: Stochastic processes Lecture 7 Linear time invariant systems 1.

31

Fourier transform of the impulse response

β€’ The Transfer function (System function) is the Fourier transformed impulse response

β€’ The impulse response can be determined from the Transfer function with the invers Fourier transform

Page 32: Stochastic processes Lecture 7 Linear time invariant systems 1.

32

Fourier transform of LTI systems

β€’ Convolution corresponds to multiplication in the frequency domain

-10 -5 0 5 10 15 20

0

t

y(t)

Impuls response

-2 -1 0 1 20

0.5

1

1.5

f (Hz)

|H(f

)|

-10 -5 0 5 10 15 20-3

-2

-1

0

1

2

3

t

x(t)

Input

-2 -1 0 1 20

500

1000

1500

2000

2500

3000

f (Hz)

|X(f

)|

-2 -1 0 1 20

200

400

600

800

1000

1200

f (Hz)

|Y(f

)|

-10 -5 0 5 10 15 20-3

-2

-1

0

1

2

3

t

y(t)

Output

Time domain

Frequency domain

* =

x =

Page 33: Stochastic processes Lecture 7 Linear time invariant systems 1.

33

Causal systems

β€’ Independent on the future signal

-10 -5 0 5 10 15 20

0

t

y(t)

Impuls response

Page 34: Stochastic processes Lecture 7 Linear time invariant systems 1.

34

Stochastic signals and LTI systems

β€’ Estimation of the output from a LTI system when the input is a stochastic process

Ξ‘ is a delay factor like Ο„

Page 35: Stochastic processes Lecture 7 Linear time invariant systems 1.

35

Statistical estimates of output

β€’ The specific distribution function fX(x,t) is difficult to estimate. Therefor we stick to– Mean – Autocorrelation – PSD – Mean square value.

Page 36: Stochastic processes Lecture 7 Linear time invariant systems 1.

36

Expected Value of Y(t) (1/2)

β€’ How do we estimate the mean of the output?

𝐸 [π‘Œ (𝑑 ) ]=𝐸[βˆ«βˆ’βˆž

∞

𝑋 (π‘‘βˆ’π›Ό )h (𝛼 )𝑑𝛼 ]𝐸 [π‘Œ (𝑑 ) ]=∫

βˆ’βˆž

∞

𝐸 [ 𝑋 (π‘‘βˆ’π›Ό ) ] h (𝛼 ) 𝑑𝛼

𝐸 [π‘Œ (𝑑 ) ]=βˆ«βˆ’βˆž

∞

π‘šπ‘₯ (π‘‘βˆ’π›Ό)h (𝛼 )𝑑𝛼

If mean of x(t) is defined as mx(t)

π‘Œ (𝑑)=βˆ«βˆ’βˆž

∞

𝑋 (π‘‘βˆ’π›Ό )h (𝛼 )𝑑𝛼

Page 37: Stochastic processes Lecture 7 Linear time invariant systems 1.

37

Expected Value of Y(t) (2/2)

If x(t) is wide sense stationary

π‘šπ‘₯ (π‘‘βˆ’π›Ό )=π‘šπ‘₯ (𝑑 )=π‘šπ‘₯(π‘šπ‘₯π‘–π‘ π‘Žπ‘π‘œπ‘›π‘ π‘‘π‘Žπ‘›π‘‘)

Alternative estimate:At 0 Hz the transfer function is equal to the DC gain

βˆ«βˆ’βˆž

∞

h (𝛼 )𝑑𝛼=𝐻 (0)

Therefor: π‘šπ‘¦=𝐸 [π‘Œ (𝑑 ) ]=π‘šπ‘₯𝐻 (0)

Page 38: Stochastic processes Lecture 7 Linear time invariant systems 1.

38

Expected Mean square value (1/2)

𝐸 [π‘Œ (𝑑 )2 ]=𝐸 [π‘Œ (𝑑 )π‘Œ (𝑑 ) ] π‘Œ (𝑑)=βˆ«βˆ’βˆž

∞

𝑋 (π‘‘βˆ’π›Ό )h (𝛼 )𝑑𝛼

𝐸 [π‘Œ (𝑑 )2 ]=𝐸[(βˆ«βˆ’βˆžβˆž

𝑋 (π‘‘βˆ’π›Ό1 )h (𝛼1 ) 𝑑𝛼1)(βˆ«βˆ’βˆž

∞

𝑋 (π‘‘βˆ’π›Ό2 )h (𝛼2 )𝑑𝛼2) ]𝐸 [π‘Œ (𝑑 )2 ]=𝐸[∫

βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝑋 (π‘‘βˆ’π›Ό1 )𝑋 (π‘‘βˆ’π›Ό2 )h (𝛼1 )h (𝛼2 )𝑑𝛼1𝑑𝛼2 ]𝐸 [π‘Œ (𝑑 )2 ]=∫

βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝐸 [𝑋 (π‘‘βˆ’π›Ό1 ) 𝑋 (π‘‘βˆ’π›Ό2 ) ] h (𝛼1 )h (𝛼2 )𝑑𝛼1𝑑𝛼2

𝐸 [π‘Œ (𝑑 )2 ]=βˆ«βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝑅π‘₯π‘₯ (π‘‘βˆ’π›Ό1, π‘‘βˆ’π›Ό2)h (𝛼1 )h (𝛼2 )𝑑𝛼1𝑑𝛼2

𝐸 [π‘Œ (𝑑 )2 ]=βˆ«βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝑅π‘₯π‘₯ (𝛼1 ,𝛼2)h (π‘‘βˆ’π›Ό1 )h (π‘‘βˆ’π›Ό2 )𝑑𝛼1𝑑𝛼2

Page 39: Stochastic processes Lecture 7 Linear time invariant systems 1.

39

Expected Mean square value (2/2)

𝐸 [π‘Œ (𝑑 )2 ]=βˆ«βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝑅π‘₯π‘₯ (𝛼1 ,𝛼2)h (π‘‘βˆ’π›Ό1 )h (π‘‘βˆ’π›Ό2 )𝑑𝛼1𝑑𝛼2

𝐸 [π‘Œ (𝑑 )2 ]=βˆ«βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝑅π‘₯π‘₯ (π›Όβˆ’ 𝛽)h (𝛼 )h (𝛽 )𝑑𝛼1𝑑𝛼2

By substitution:

𝐸 [π‘Œ (𝑑 )2 ]=βˆ«βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝑅π‘₯π‘₯ (π‘‘βˆ’π›Ό ,π‘‘βˆ’ 𝛽)h (𝛼 )h ( 𝛽)𝑑𝛼1𝑑𝛼2

If X(t)is WSS

Thereby the Expected Mean square value is independent on time

Page 40: Stochastic processes Lecture 7 Linear time invariant systems 1.

40

Cross correlation function between input and output

β€’ Can we estimate the Cross correlation between input and out if X(t) is wide sense stationary

𝑅𝑦π‘₯ (𝑑+𝜏 , 𝑑 )=𝐸 [π‘Œ (𝑑+𝜏 )π‘‹βˆ—(𝑑)]

𝑅𝑦π‘₯ (𝑑+𝜏 , 𝑑 )=𝐸[(βˆ«βˆ’βˆžβˆž

𝑋 (π‘‘βˆ’π›Ό+𝜏 )h (𝛼 ) 𝑑𝛼) π‘‹βˆ—(𝑑)]𝑅𝑦π‘₯ (𝑑+𝜏 , 𝑑 )=𝐸[∫

βˆ’βˆž

∞

𝑋 (π‘‘βˆ’π›Ό+𝜏 ) π‘‹βˆ—(𝑑)h (𝛼 )𝑑𝛼 ]

𝑅𝑦 π‘₯ (𝜏 )=βˆ«βˆ’βˆž

∞

𝑅π‘₯π‘₯ (πœβˆ’π›Ό )h (𝛼 ) 𝑑𝛼=𝑅π‘₯π‘₯ (𝜏 )βˆ—h (𝜏)  

-10 -5 0 5 10 15 20-3

-2

-1

0

1

2

3

t

x(t)

Input

-10 -5 0 5 10 15 20-3

-2

-1

0

1

2

3

t

y(t)

Output

𝑅π‘₯π‘₯ (𝜏 )=𝐸 [ 𝑋 (𝑑+𝜏 )𝑋 (𝑑)]

-30 -20 -10 0 10 20 30-1500

-1000

-500

0

500

1000

1500

(s)

Rxy

()

Cross-correlation between y(t) and x(t)

Thereby the cross-correlation is the convolution between the auto-correlation of x(t) and the impulse response

Page 41: Stochastic processes Lecture 7 Linear time invariant systems 1.

41

Autocorrelation of the output (1/2)

𝑅𝑦𝑦 (𝜏 )=𝑅𝑦𝑦 (𝑑+𝜏 , 𝑑 )=𝐸 [π‘Œ (𝑑+𝜏 )π‘Œ (𝑑) ]

𝑅𝑦𝑦 (𝜏 )=βˆ«βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝐸 [ 𝑋 (𝑑+πœβˆ’π›Ό ) 𝑋 (π‘‘βˆ’ 𝛽 )]h (𝛼 )h (𝛽 )𝑑𝛼 𝑑𝛽

π‘Œ (𝑑+𝜏)=βˆ«βˆ’βˆž

∞

𝑋 (𝑑+πœβˆ’π›Ό )h (𝛼 )𝑑𝛼

π‘Œ (𝑑)=βˆ«βˆ’βˆž

∞

𝑋 (π‘‘βˆ’ 𝛽)h (𝛽 )𝑑 𝛽

Y(t) and Y(t+Ο„) is :

𝑅𝑦𝑦 (𝜏 )=βˆ«βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝑅π‘₯π‘₯(πœβˆ’π›Ό+𝛽)h (𝛼 )h (𝛽 )𝑑𝛼 𝑑𝛽

Page 42: Stochastic processes Lecture 7 Linear time invariant systems 1.

42

Autocorrelation of the output (2/2)

𝑅𝑦𝑦 (𝜏 )=βˆ«βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝐸 [ 𝑋 (𝑑+πœβˆ’π›Ό ) 𝑋 (π‘‘βˆ’ 𝛽 )]h (𝛼 )h (𝛽 )𝑑𝛼 𝑑𝛽

By substitution: Ξ±=-Ξ²

𝑅𝑦𝑦 (𝜏 )=βˆ«βˆ’βˆž

∞

βˆ«βˆ’βˆž

∞

𝐸 [ 𝑋 (𝑑+πœβˆ’π›Ό ) 𝑋 (𝑑+𝛼 )] h (𝛼 )h (βˆ’π‘Ž )𝑑𝛼 𝑑𝛼

Remember:

-30 -20 -10 0 10 20 30-1000

-500

0

500

1000

(s)

Rxy

()

Autocorrelation of y(t)

𝑅𝑦𝑦 (𝜏 )=𝑅𝑦 π‘₯ (𝜏 )βˆ—h(βˆ’πœ)

𝑅𝑦𝑦 (𝜏 )=𝑅π‘₯ π‘₯ (𝜏 )βˆ—h (𝜏)βˆ—h(βˆ’πœ )

Page 43: Stochastic processes Lecture 7 Linear time invariant systems 1.

43-2 -1 0 1 2

0

2

4

6

8

10

12x 10

5

f (Hz)

Syy

(f)

Spectrum of output

β€’ Given:

β€’ The power spectrum is

𝑅𝑦 𝑦 (𝜏 )=𝑅π‘₯π‘₯ (𝜏 )βˆ—h (𝜏 )βˆ—h(βˆ’πœ )

-2 -1 0 1 20

0.5

1

1.5

f (Hz)

|H(f

)|2

-2 -1 0 1 20

2

4

6

8

10x 10

6

f (Hz)

Sxx

(f)

x =

¿𝐻 ( 𝑓 )∨¿2=𝐻 ( 𝑓 )π»βˆ—( 𝑓 )ΒΏ

Page 44: Stochastic processes Lecture 7 Linear time invariant systems 1.

44

Filter examples

Page 45: Stochastic processes Lecture 7 Linear time invariant systems 1.

45

Typical LIT filters

β€’ FIR filters (Finite impulse response)β€’ IIR filters (Infinite impulse response)

– Butterworth– Chebyshev– Elliptic

Page 46: Stochastic processes Lecture 7 Linear time invariant systems 1.

Ideal filters

β€’ Highpass filter

β€’ Band stop filter

β€’ Bandpassfilter

Page 47: Stochastic processes Lecture 7 Linear time invariant systems 1.

47

Filter types and rippels

Page 48: Stochastic processes Lecture 7 Linear time invariant systems 1.

Analog lowpass Butterworth filter

β€’ Is ”all pole” filter– Squared frequency transfer function

β€’ N:filter orderβ€’ fc: 3dB cut off frequency

β€’ Estimate PSD from filter

NcfffH 2

2

/1

1)(

Nc

xxyyff

ffS 2/1

1)(S)(

Page 49: Stochastic processes Lecture 7 Linear time invariant systems 1.

Chebyshev filter type I

β€’ Transfer function

β€’ Where Ξ΅ is relateret to ripples in the pass band

β€’ Where TN is a N order polynomium

pN ffTfH

/1

122

2

1

1

)coshcosh(

)coscos()(

1

1

x

x

xN

xNxTN

Page 50: Stochastic processes Lecture 7 Linear time invariant systems 1.

Transformation of a low pass filter to other types (the s-domain)

Filter type Transformation New Cutoff frequency

Lowpas>Lowpas

Lowpas>Highpas

Lowpas>Highpas

Lowpas>Stopband

ssp

p

'

p'

ss pp '

p'

)(

2

lu

ulp s

ss

ul

lup s

ss

2

)(

ul ,

ul ,

Lowest Cutoff frequency

Highest Cutoff frequency:

:

u

l

p

New Cutoff frequencyp'

Old Cutoff frequency

Page 51: Stochastic processes Lecture 7 Linear time invariant systems 1.

51

Discrete time implantation of filters

β€’ A discrete filter its Transfer function in the z-domain or Fourier domain

– Where bk and ak is the filter coefficients

β€’ In the time domain:

Mm

Mm

zazazaa

zbzbzbbzH

zX

zY

.......Β΄

.......Β΄)(

)(

)(2

21

10

22

110

][......]2[]1[

][......]2[]1[][][

21

210

Mnyanyanya

Mnxanxbnxbnxbny

M

M

Page 52: Stochastic processes Lecture 7 Linear time invariant systems 1.

52

Filtering of a Gaussian process

β€’ Gaussian process– X(t1),X(t2),X(t3),….X(tn) are jointly Gaussian for all t

and n valuesβ€’ Filtering of a Gaussian process

– Where w[n] are independent zero mean Gaussian random variables.

][......]2[]1[

][......]2[]1[][][

21

210

Mnyanyanya

Mnwanwbnwbnwbny

M

M

Page 53: Stochastic processes Lecture 7 Linear time invariant systems 1.

The Gaussian Process

β€’ X(t1),X(t2),X(t3),….X(tn) are jointly Gaussian for all t and n values

β€’ Example: randn() in Matlab

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000-4

-3

-2

-1

0

1

2

3

4

5Gaussian process

-4 -3 -2 -1 0 1 2 3 4 50

100

200

300

400

500

600

700Histogram of Gaussian process

Page 54: Stochastic processes Lecture 7 Linear time invariant systems 1.

The Gaussian Process and a linear time invariant systems

β€’ Out put = convolution between input and impulse response

Gaussian input Gaussian output

Page 55: Stochastic processes Lecture 7 Linear time invariant systems 1.

Example

β€’ x(t):

β€’ h(t): Low pass filterβ€’ y(t):

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000-4

-3

-2

-1

0

1

2

3

4

5Gaussian process

-4 -3 -2 -1 0 1 2 3 4 50

100

200

300

400

500

600

700Histogram of Gaussian process

-1.5 -1 -0.5 0 0.5 1 1.50

100

200

300

400

500

600Histogram of y(t)

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000-1.5

-1

-0.5

0

0.5

1

1.5

Page 56: Stochastic processes Lecture 7 Linear time invariant systems 1.

56

Filtering of a Gaussian process example 2

0 100 200 300 400 500-1000

-500

0

Frequency (Hz)

Pha

se (

degr

ees)

0 100 200 300 400 500-100

-50

0

Frequency (Hz)

Mag

nitu

de (

dB) Transfere function of filter

0 100 200 300 400 500 600 700 800 900 1000-4

-2

0

2

4

t (ms)

x(t)

White noise

Band pass filter

0 100 200 300 400 500 600 700 800 900 1000-1

-0.5

0

0.5

1

t (ms)

y(t)

Output

Page 57: Stochastic processes Lecture 7 Linear time invariant systems 1.

57

Intro to system identification

β€’ Modeling of signals using linear Gaussian models:

β€’ Example: AR models

β€’ The output is modeled by a linear combination of previous samples plus Gaussian noise.

][][......]2[]1[][ 21 nwMnyanyanyany M

Page 58: Stochastic processes Lecture 7 Linear time invariant systems 1.

58

Modeling example

β€’ Estimated 3th order model

][]3[0.7299-]2[2.3903]1[-2.6397][ nwnynynyny

0 100 200 300 400 500 600 700 800 900 1000-1

-0.5

0

0.5

1

t (ms)

y(t)

Output

451 451.5 452 452.5 453 453.5 4540.25

0.3

0.35

0.4

t (ms)y(

t)

Output

signal

points used for predictionPrediction

True point

453.98 453.99 454 454.01 454.02

0.282

0.284

0.286

0.288

0.29

0.292

0.294

t (ms)

y(t)

Output

w[n]

Page 59: Stochastic processes Lecture 7 Linear time invariant systems 1.

59

Agenda (Lec. 7)

β€’ Recap: Linear time invariant systemsβ€’ Stochastic signals and LTI systems

– Mean Value function– Mean square value – Cross correlation function between input and output– Autocorrelation function and spectrum output

β€’ Filter examples β€’ Intro to system identification