Scribd
Upload
242 views20 pages

Advanced DSP for Engineering Students

This document provides an overview of discrete random signal processing and spectrum estimation techniques. It introduces key concepts such as random variables, mean, variance, correlation, autocorrelation, power spectrum, white noise, wide sense stationary processes, ergodicity, filtering of random processes, and the Wiener-Khintchine theorem. It also describes modeling random signals using autoregressive (AR), moving average (MA), and autoregressive moving average (ARMA) processes. Finally, it outlines methods for spectrum estimation including non-parametric techniques like the periodogram and parametric modeling using AR, MA, and ARMA processes.

Uploaded by

Puspha Vasanth R
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
242 views20 pages

Advanced DSP for Engineering Students

This document provides an overview of discrete random signal processing and spectrum estimation techniques. It introduces key concepts such as random variables, mean, variance, correlation, autocorrelation, power spectrum, white noise, wide sense stationary processes, ergodicity, filtering of random processes, and the Wiener-Khintchine theorem. It also describes modeling random signals using autoregressive (AR), moving average (MA), and autoregressive moving average (ARMA) processes. Finally, it outlines methods for spectrum estimation including non-parametric techniques like the periodogram and parametric modeling using AR, MA, and ARMA processes.

Uploaded by

Puspha Vasanth R
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 20

AP7101-ADVANCED DIGITAL SIGNAL

PROCESSING

K. SUBHASHINI
ASSISTANT PROFESSOR
DEPARTMENT OF ELECTRONICS AND
COMMUNICATION ENGINEERING
SRI SAIRAM ENGINEERING COLLEGE

UNIT I

DISCRETE RANDOM SIGNAL PROCESSING

Signals-Deterministic
-Random
Ensemble averages
-Mean
-Variance
Expected value is mean of Random
variable x
Expected value of x2 is mean square
value

Mean square value is a measure for


the quality of an estimate
Ensemble averages provide
important and useful characterization
of jointly distributed random variable
-Correlation
-Covariance

If x & y have zero mean


covariance=correlation
Normalized covariance is correlation
coefficient
|xy|1
If rxy =0 -orthogonal
Random Process
-Mean
-Variance
Autocovariance

Autocorrelation
5

Wide Sense Stationary Random


Process
-Mean is a constant
- depends on the difference k,l
-Variance is finite
Ergodicity
-If sample mean of a WSS process
converges to mx in mean square sense,
then process is said to be ergodic in mean
6

White Noise
-sequence of uncorrelated random
variable having variance
Power Spectrum
-shows how signal is distributed as a
function of frequency

Compute inverse DTFT of power


spectrum

Z-transform

The Wiener-Khintchine theorem states a


relationship between two important
characteristics of a random process: the power
spectrum of the process and the correlation
function of the process.

Filtering random process


-relationship between mean and
autocorrelation of input process to
mean and autocorrelation of output
process
Applications
-signal detection and estimation
-synthesis
10

Spectral Factorization
may be factored into a product
-Process that can be factored is called
regular process
Inverse filter(Whitening filter)
x(n)

1/H(z)

w(n)
11

Wold Decomposition
-Random process is decomposed into sum of two
orthogonal process
-x(n)=xp(n)+xr(n)
Special types of Random process
Process generated by filtering white noise with a
LSI filter that has rational system function
-ARMA
-AR
-MA
12

Autocorrelation sequence of these


process satisfy a set of equation
called Yule-Walker equation
Yule-Walker equation relate to
parameter of the filter

13

ARMA process(p,q)
system function H(z)=

-Power spectrum

14

AR process(p,0)
system function

power spectrum

15

MA process(0,q)
system function

power spectrum

16

Modeling Deterministic signals


Least squares method
-In this method of signal modeling the
squared error is to be minimized
-the partial derivative w.r.t each of the
coefficients vanish
Disadvantage
Leads to mathematically intractable
solution
17

Stochastic modeling
-ARMA model
-AR model
-MA model

18

UNIT II
SPECTRUM ESTIMATION

19

Spectral Estimation

Non Parametric

Periodogram
Modified
Periodogram
Barletts Method
Welch Method
Blackman-Tukey
Method

Parametric

ARMA
AR
MA

20

You might also like