Discrete-time Random Signals Until now, we have assumed that the signals are deterministic, i.e., each value of a sequence is uniquely determined. In many situations, the processes that generate signals are so complex as to make precise description of a signal extremely difficult or undesirable. A random or stochastic signal is considered to be characterized by a set of probability density functions. Stochastic Processes Random (or stochastic) process (or signal) A random process is an indexed family of random variables characterized by a set of probability distribution function. A sequence x[n], <n< . Each individual sample x[n] is assumed to be an outcome of some underlying random variable Xn. The difference between a single random variable and a random process is that for a random variable the outcome of a random-sampling experiment is mapped into a number, whereas for a random process the outcome is mapped into a sequence. Stochastic Processes (continue) Probability density function of x[n]: pxn , n Joint distribution of x[n] and x[m]: pxn , n , xm , m Eg., x1[n] = Ancos(wn+n), where An and n are random variables for all < n < , then x1[n] is a random process. Independence and Stationary x[n] and x[m] are independent iff pxn , n , xm , m pxn , n pxm , m x is a stationary process iff p x n k , n k , x m k , m k p x n , n , x m , m for all k. That is, the joint distribution of x[n] and x[m] depends only on the time difference m n. Stationary (continue) Particularly, when m = n for a stationary process: pxn k , n k pxn , n It implies that x[n] is shift invariant. Stochastic Processes vs. Deterministic Signal In many of the applications of discrete-time signal processing, random processes serve as models for signals in the sense that a particular signal can be considered a sample sequence of a random process. Although such a signals are unpredictable – making a deterministic approach to signal representation is inappropriate – certain average properties of the ensemble can be determined, given the probability law of the process. Expectation Mean (or average) m xn xn xn p xn , n dxn denotes the expectation operator g xn g xn pxn , n dxn For independent random variables xn ym xn ym Mean Square Value and Variance Mean squared value { xn } 2 xn pxn , n dxn 2 Variance 2 varxn xn mxn Autocorrelation and Autocovariance Autocorrelation xx {n,m} xn xm xn xm pxn , n , xm , m dxn dxm Autocovariance xx {n,m} xn m xn xm m xm xx {n,m} m xn mxm * Stationary Process For a stationary process, the autocorrelation is dependent on the time difference m n. Thus, for stationary process, we can write mx mxn xn 2 x x n m x 2 If we denote the time difference by k, we have xx n k , n xx k xnk xn Wide-sense Stationary In many instances, we encounter random processes that are not stationary in the strict sense. If the following equations hold, we call the process wide-sense stationary (w. s. s.). mx mxn xn 2 x k x x n m x xx n k , n xx 2 n k xn Time Averages For any single sample sequence x[n], define their time average to be L 1 xn lim xn l 2 L 1 n L Similarly, time-average autocorrelation is xn mxn L 1 lim xn mx n l 2 L 1 n L Ergodic Process A stationary random process for which time averages equal ensemble averages is called an ergodic process: xn mx xn mxn xx m Ergodic Process (continue) It is common to assume that a given sequence is a sample sequence of an ergodic random process, so that averages can be computed from a single sequence. 1 L 1 ˆx m xn L n 0 In practice, we cannot compute with the limits, but L 1 instead the quantities. 1 2 2 ˆ x n m x x Similar quantities are often L n 0 computed as estimates of the L 1 1 mean, variance, and xn mx n xn mx n L L n 0 autocorrelation. Properties of correlation and covariance sequences xx m xn m xn xx m xn m m x xn m x xy m xn m y n xy m xn m m x y n m y Property 1: xx m xx m m x xy m xy m 2 mx m y Properties of correlation and covariance sequences (continue) Property 2: 2 xx 0 E xn Mean Squared Value xy 0 Property 3 2 x Variance xx m m xx m xx m xx xy m m xy m xy m xy Properties of correlation and covariance sequences (continue) Property 4: xy m xx 0 yy 0 2 xy m xx 0 yy 0 2 xx m xx 0 xx m xx 0 Properties of correlation and covariance sequences (continue) Property 5: If yn xnn0 yy m xx m yy m xx m Fourier Transform Representation of Random Signals Since autocorrelation and autocovariance sequences are all (aperiodic) one-dimensional sequences, there Fourier transform exist and are bounded in |w|. Let the Fourier transform of the autocorrelation and autocovariance sequences be e e xx m xx e jw xy m xy e jw xx m xx xy m xy jw jw Fourier Transform Representation of Random Signals (continue) Consider the inverse Fourier Transforms: 1 xx m 2 1 xx m 2 e e xx e jw e jwn dw xx jw jwn dw Fourier Transform Representation of Random Signals (continue) Consequently, 1 xn xx 0 xx e jw dw 2 1 2 jw jwn x xx 0 xx e e dw 2 2 Denote Pxx w xx e to be the power density spectrum (or power spectrum) of the random process x. jw Power Density Spectrum xn 2 1 2 Pxx wdw The total area under power density in [,] is the total energy of the signal. Pxx(w) is always real-valued since xx(n) is conjugate symmetric For real-valued random processes, Pxx(w) = xx(ejw) is both real and even. Mean and Linear System Consider a linear system with frequency response h[n]. If x[n] is a stationary random signal with mean mx, then the output y[n] is also a stationary random signal with mean mx equaling to m y n yn k k hk xn k hk mx n k Since the input is stationary, mx[nk] = mx , and consequently, m y mx j0 h k H e mx k Stationary and Linear System If x[n] is a real and stationary random signal, the autocorrelation function of the output process is yy n , n m ynyn m hk hr xn k xn m r k r k r hk hr xn k xn m r Since x[n] is stationary , {x[nk]x[n+mr] } depends only on the time difference m+kr. Stationary and Linear System (continue) Therefore, yy n , n m k r hk hr xx m k r yy m The output power density is also stationary. Generally, for a LTI system having a wide-sense stationary input, the output is also wide-sense stationary. Power Density Spectrum and Linear System By substituting l = rk, l k yy m where xx m l hk hk hl k xx m l chh l l chh l hk hl k k A sequence of the form of chh[l] is called a deterministic autocorrelation sequence. Power Density Spectrum and Linear System (continue) A sequence of the form of Chh[l] l = rk, C e e yy e jw jw hh jw xx where Chh(ejw) is the Fourier transform of chh[l]. For real h, chh l hl h l H e H e e H e Chh e Thus Chh jw jw jw jw 2 jw Power Density Spectrum and Linear System (continue) We have the relation of the input and the output power spectrums to be the following: H e e yy e jw jw 2 jw xx 1 jw xn xx 0 e dw total average power of the input xx 2 2 1 2 jw jw yn yy 0 H e e dw xx 2 total average power of the output 2 Power Density Property Key property: The area over a band of frequencies, wa<|w|<wb, is proportional to the power in the signal in that band. To show this, consider an ideal band-pass filter. Let H(ejw) be the frequency of the ideal band pass filter for the band wa<|w|<wb. Note that |H(ejw)|2 and xx(ejw) are both even functions. Hence, 0 averagepowerin output yy 1 2 wa w b e He jw 2 jw xx 1 dw 2 wb w a e dw He jw 2 jw xx White Noise (or White Gaussian Noise) A white noise signal is a signal for which xx m x2 m Hence, its samples at different instants of time are uncorrelated. The power spectrum of a white noise signal is a constant jw 2 xx e x The concept of white noise is very useful in quantization error analysis. White Noise (continue) The average power of a white-noise is therefore 1 1 2 jw xx 0 xx e dw x dw x2 2 2 White noise is also useful in the representation of random signals whose power spectra are not constant with frequency. A random signal y[n] with power spectrum yy(ejw) can be assumed to be the output of a linear time-invariant system with a white-noise input. H e yy e jw jw 2 2 x Cross-correlation The cross-correlation between input and output of a LTI system: m xn yn m xy xn hk xn m k k hk xx m k k That is, the cross-correlation between the input output is the convolution of the impulse response with the input autocorrelation sequence. Cross-correlation (continue) By further taking the Fourier transform on both sides of the above equation, we have xy e jw H e jw xx e jw This result has a useful application when the input is white noise with variance x2. 2 xy m x hm, xy e H e jw 2 x jw These equations serve as the bases for estimating the impulse or frequency response of a LTI system if it is possible to observe the output of the system in response to a white-noise input. Remained Materials Not Included From Chap. 4, the materials will be taught in the class without using slides