CHAPTER 3 Discrete-Time Signals in the Transform-Domain Wang Weilian wlwang@ynu.edu.cn School of Information Science and Technology Yunnan University Outline • The Discrete-Time Fourier Transform • The Discrete Fourier Transform • Relation between the DTFT and the DFT, and Their Inverses • Discrete Fourier Transform Properties • Computation of the DFT of Real Sequences • Linear Convolution Using the DFT • The z-Transform 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 2 Outline • Region of Convergence of a Rational z-Transform • Inverse z-Transform • z-Transform Properties 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 3 The Discrete-Time Fourier Transform • The discrete-time Fourier transform (DTFT) or, simply, the Fourier transform of a discrete–time sequence x[n] is a representation of the sequence in terms of the complex exponential sequence e j x where is the real frequency variable. • The discrete-time Fourier transform sequence x[n] is defined by X e j x [ n ]e X e j of a j n n 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 4 The Discrete-Time Fourier Transform • In general X e j is a complex function of the real variable and can be written in rectangular form as X e j X re e where X re e j and imaginary parts of X im e X e j j j jX im e j are, respectively, the real and , and are real functions of . • Polar form X e j X e j w h ere 云南大学滇池学院课程:数字信号处理 e arg X e j j Discrete-Time Signals in the Transform-Domain 5 The Discrete-Time Fourier Transform • Convergence Condition: If x[n] is an absolutely summable sequence, i.e., if x n n th en X e j n x ne j n x n n Thus the equation is a sufficient condition for the existence of the DTFT. 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 6 The Discrete-Time Fourier Transform • Bandlimited Signals: – A full-band discrete-time signal has a spectrum occupying the whole frequency rang 0 . – If the spectrum is limited to a portion of the frequency range 0 , it is called a bandlimited signal. – A lowpass discrete-time signal has a spectrum occupying the frequency range 0 p , where p is called the bandwidth of the signal. – A bandpass discrete-time signal has a spectrum occupying the frequency range 0 L H , where H L is its bandwidth. 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 7 The Discrete-Time Fourier Transform • Discrete-Time Fourier Transform Properties There are a number of important properties of the discrete-time Fourier transform which are useful in digital signal processing applications. We list the general properties in Table 3.2, and the symmetry properties in Tables 3.3 and 3.4. 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 8 The Discrete-Time Fourier Transform • Energy Density Spectrum P arseval's relation : g [ n ]h [ n ] * n - 1 2 G (e j * ) H (e j )d T otal en ergy of a fin ite-en ergy seq u en ce g [ n ] : g g n 2 n If h[ n ] g [ n ], th en from P arseval's relation w e ob ser ve g | g[ n ] | 2 n - T h e q u an tity : S g g e j 1 2 G e j | G (e j ) | d 2 2 is called th e en ergy d en sity sp ectru m of th e seq u en ce g [ n ]. 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 9 The Discrete Fourier Transform • DTFT Computation Using MATLAB – The Signal Processing Toolbox in MATLAB – Functions: • freqz • abs • Angle – The forms of freqz: • H = freqz(num, den, w) • [H, w] = freqz(num, den, k, ’whole’) – Example 3.8: 云南大学滇池学院课程:数字信号处理 Program 3_1 Discrete-Time Signals in the Transform-Domain 10 The Discrete Fourier Transform • Definition The simplest relation between a finite-length sequence x[n], defined for 0 n N 1 , and its DTFT X e j is obtained by uniformly sampling X e j on the -axis between k 2 k / N , F rom X e j 0 2 at 0 k N 1 . x [ n ]e j n n X k X e j 云南大学滇池学院课程:数字信号处理 N 1 2 k / N x [ n ]e j 2 kn / N , 0 k N 1 n0 Discrete-Time Signals in the Transform-Domain 11 The Discrete Fourier Transform • The sequence X[k] is called the discrete Fourier transform (DFT) of the sequence x[n]. • Using the commonly used notation WN e j 2 / N • We can rewrite as N 1 X [k ] 0 k N 1 kn x [ n ]W N , n0 • Inverse discrete Fourier transform (IDFT) N 1 x[ n ] X [ k ]e n0 云南大学滇池学院课程:数字信号处理 j 2 kn / N N 1 kn X [ k ]W N , 0 n N 1 n0 Discrete-Time Signals in the Transform-Domain 12 The Discrete Fourier Transform • Matrix Relations N 1 The DFT samples defined in X [k ] kn x [ n ]W N can n0 be expressed in matrix form as X DN x where X is the vector composed of the N DFT samples, T X X 0 X 1 X N 1 x is the vector of N input samples, x x [0] x [1] 云南大学滇池学院课程:数字信号处理 x[ N 1] T Discrete-Time Signals in the Transform-Domain 13 The Discrete Fourier Transform • D N is the N N DN DFT matrix given by 1 1 1 1 1 2 N- 1 1 W W W N N N 2 4 2( N- 1) 1 WN WN WN N- 1 2( N- 1) ( N- 1) ( N- 1) WN 1 WN WN • IDFT relations 1 N x D X 云南大学滇池学院课程:数字信号处理 1 N * DN X Discrete-Time Signals in the Transform-Domain 14 The Discrete Fourier Transform • DFT computation Using MATLAB – MATLAB functions: fft(x), fft(x,N), ifft(X), ifft(X,N) – X = fft(x, N) If N < R=length(x), truncate (截短) to the first N samples. If N > R=length(x), zero-padded (补零) at the end. – Example 3.11, 3.12, 3.13, Program 3_2, 3_3, 3_4. 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 15 Relation between the DTFT and the DFT, and their Inverses • DTFT from DFT by Interpolation We could express X e j 1 N 1 N 1 N N 1 x [ n ]e j n k0 N 1 k0 1 N n0 N 1 X e N 1 X [ k ] e j in terms of X[k]: N 1 j 2 kn / N n0 e N 1 j n kn e X [ k ] W N k0 j n n0 N 2 k sin 2 X k N 2 k sin 2N 云南大学滇池学院课程:数字信号处理 j 2 k / N N 1 / 2 e Discrete-Time Signals in the Transform-Domain 16 Relation between the DTFT and the DFT, and their Inverses • Sampling the DTFT – Consider the following question DTFT x[ n ] X (e j ) k 2 k / N , 0 k N - 1 ? y [ n ], 0 n N - 1 Y [ k ] X ( e DFT j k ), 0 k N - 1 – We obtain the relation y[ n ] x [ n m N ], 0 n N -1 m – Example 3.14 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 17 Relation between the DTFT and the DFT, and their Inverses • Numerical Computation of the DTFT Using the DFT – Let X ( e j ) be the DTFT of length-N sequence x[n]. We wish to evaluate X ( e j ) at a dense grid of frequencies: k 2 k / M , 0 k M 1, w h ere M X e j k X e j N 1 k x [ n ]e j k n N N 1 n0 x [ n ]e j 2 kn / M n0 x[n ] 0 n N 1 D efin e a n ew seq u en ce x e [ n ] N n M -1 0 th en X e j k DFT { x M 1 [ n ]} e x e [ n ]e j 2 kn / M n0 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 18 Discrete Fourier Transform Properties • Discrete Fourier Transform Properties Like the DTFT, the DFT also satisfies a number of properties that are useful in signal processing application. A summary of the DFT properties are included in Tables 3.5, 3.6, and 3.7. 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 19 Discrete Fourier Transform Properties • Circular Shift of a Sequence – Time-shifting property of the DTFT x 1 [ n ] x [ n n 0 ] X 1 ( e DTFT j ) e j n 0 X (e j ) – Circular shifting property of the DFT x [ n ], 0 n N - 1 X [ k ], 0 k N - 1 DFT ? x c [ n ], 0 n N - 1 X c [ k ] W N 0 X [ k ], 0 k N - 1 DFT W e ob tain kn x c [ n ] x [ n n 0 N ] x [( n n 0 )% N ] n0 n N 1 x[ n n0 ] F or n 0 > 0, x c [ n ] 0 n n0 x[n n0 N ] 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 20 Computation of the DFT of Real Sequences • Computation of the DFT of Real Sequences Tow N-point DFTs can be computed efficiently using a single N-point DFT X[k] of a complex length-N sequence x[n] defined by x n g n jh n where, g n R e{ x [ n ]} and h[ n ] Im { x[ n ]} 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 21 Computation of the DFT of Real Sequences we arrive at: Note that G k 1 H k X k X 2j * { X [ k ] X k 2 * X k 云南大学滇池学院课程:数字信号处理 1 N * N }, k X* N k N N Discrete-Time Signals in the Transform-Domain 22 Linear Convolution Using the DFT • Linear Convolution of Two Finite-Length Sequences Let g[n] and h[n] be finite-length sequences of lengths N and M, respectively. Denote L=M+N-1. Define two length-L sequences, g n , 0 n N 1 ge n 0. N n L 1 h n , 0 n M 1 he n 0. M n L 1 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 23 Linear Convolution Using the DFT obtained by appending g[n] and h[n] with zero-valued samples. Then yL n g n h n yc n g e n linear convolution h e n • Linear Convolution of a Finite-Length Sequence with an Infinite-Length Sequence – Overlap-Add Method – Overlap-Save Method 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 24 The z-Transform • Definition For a given sequence g[n], its z-transform G(z) is defined as G z Z g n g n z n n where z R e z j Im z is a complex variable. If we let z re j , then the right-hand side of the above expression reduces to G re j g n r n e j n n 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 25 The z-Transform For a given sequence, the set R of values of z for which its z-transform converges is called the region of convergence (ROC). If g n r n n In general, the region of convergence R of a ztransform of a sequence g[n] is an annular region of the z-plane: Rg z Rg 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 26 The z-Transform • Rational z-Transforms – An alternate representation as a ration of two polynomials in z: – An alternate representation in factored form as 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 27 Region of Convergence of a Rational zTransform • The ROC of a rational z-transform is bounded by the locations of its poles. – A finite-length sequence ROC: 0 z – A right-sided sequence ROC: Rg z – A left-sided sequence ROC: z Rg – A two-sided sequence ROC: Rg z Rg 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 28 Inverse z-Transform • General Expression – By the inverse Fourier transform relation. We have g n r n 1 2 G re j e j n d j – By making the change of variable z re , the above equation can be converted into a contour integral given by 1 n 1 g n G z z dz ' 2 j C ' Where C is a counterclockwise contour of integration defined by z r 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 29 Inverse z-Transform • Inverse Transform by Partial-Fraction Expansion G z can be expressed as G z P z D z • We can divide P(Z) by D(Z) and re-express G(Z) as G z M N 0 云南大学滇池学院课程:数字信号处理 z P z D z Discrete-Time Signals in the Transform-Domain 30 Inverse z-Transform • Simple Poles p168 • Multiple Poles p169 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 31 z-Transform Properties • P174 Table 3.9 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 32 Summary • Three different frequency-domain representations of an aperiodic discrete-time sequence have been introduced and their properties reviewed .Two of these representations, the discrete-time Fourier transform (DTFT) and the z-transform, are applicable to any arbitrary sequence, whereas the third one , the discrete Fourier transform (DFT), can be applied only to finite-length sequences. • Relation between these three transforms have been established. The chapter ends with a discussion on the transform-domain representation of a random discrete-time sequence. • For future convenience we summarize below these three frequency-domain representations. 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 33 Assignment and Experiment • Assignment – A03: 3.2, 3.12, 3.20, See p180~182 – A04: – A05: • Experiment – E03: Q3.3 See p32 – E04: – E05 云南大学滇池学院课程:数字信号处理 Discrete-Time Signals in the Transform-Domain 34