EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan Definition of Discrete-Time Fourier Transform (DTFT) F{x[n]} = X(ejω ) , +∞ X x[n]e−jωn n=−∞ 1 F −1 {X(ejω )} = x[n] , 2π Z X(ejω )ejωn dω 2π • Why use the above awkward notation for the transform? • Answer: It is consistent with the z-transform X(z) , ∞ X x[n]z −n ⇒ X(ejω ) = X(z) |z=ejω n=−∞ • This also enables us to distinguish between the discrete-time (DT) and continuous-time (CT) Fourier transforms. Dr. H. Nguyen Page 203 EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan Some Observations X(ejω ) = ∞ X x[n]e−jωn ; n=−∞ | {z Analysis equation } 1 x[n] = 2π | Z X(ejω )ejωn dω 2π {z } Synthesis equation • Since ejωn = ej(ω+`2π)n , for any pair of integers ` and n ⇒ X(ejω ) is a periodic function of ω with a fundamental period of 2π. • Unlike the DT Fourier series, the frequency ω is continuous (Recall that the CT Fourier transform X(jω) is not periodic in general). • Thus the DT synthesis integral can be taken over any continuous interval of length 2π • This is similar to the CT Fourier series analysis equation Dr. H. Nguyen Page 204 EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan Frequency Concept for Discrete-Time Signals • Recall that ej(ω+`2π)n = ejωn for any integer ` • If seemingly very high-frequency discrete-time signals, cos[(ω + `2π)n], are equal to low-frequency discrete-time signals, cos(ωn), what do low and high frequencies mean for discrete-time signals? • Note that the unit of ω is “radians per sample” • A sinusoid with a frequency of 0.1 radians per sample is the same as one with a frequency of 0.1 + 2π radian per sample • No DT signal can oscillate “faster” between two consecutive samples of opposite magnitudes • No DT signal can oscillate “slower” than 0 radians per sample • Thus (i) ω = π, `(π + 2π) is highest perceivable frequency for DT signals (ii) ω = 0, `(2π) is lowest perceivable frequency Dr. H. Nguyen Page 205 EE351–Spectrum Analysis and Discrete Time Systems 5 Portland State University ECE 223 University of Saskatchewan Discrete-Time Fourier Transform Ver. 1.14 6 Frequency Concept for Discrete-Time Signals (Cont’d) Discrete-Time Frequency Concepts Continued X(ejω ) 1 −4π −3π −2π −π 0 π ω 2π 3π 4π X(ejω ) 1 −4π −3π −2π −π 0 π ω 2π 3π 4π •• Low frequencies are those that Low frequencies are those that are nearare 0 near 0 •• High frequencies are those near ±π High frequencies are those near ±π •• Intermediate frequencies are inthose in between Intermediate frequencies are those between •• Note thatthethe highest frequency, π per radians sample equalperto Note that highest frequency, π radians sampleper is equal to 0.5iscycles 0.5 cycles per sample sample • We will encounter this concept again when we discuss sampling Dr. H. Nguyen 7 Portland State University Page 206 ECE 223 Discrete-Time Fourier Transform Ver. 1.14 8 EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan Examples Example 1: Find the Fourier transform of h[n] = an u[n] where a < 1. Sketch the transform over range of −3π to 3π for a = 0.5 and a = −0.5. Example 2: Find the Fourier transform of the following pulse signal. Sketch the transform over range of −3π to 3π for N = 5, 10. 1 |n| ≤ N pN [n] = 0 |n| ≤ N Dr. H. Nguyen Page 207 14 EE351–Spectrum Analysis and Discrete Time Systems 9 Portland State University ECE 223 Discrete-Time Fourier Transform University of Saskatchewan Ver. 1.14 10 Example 1: First-Order Filter a = 0.5 Example 3: First-Order Filter a = 0.5 n Fourier Transform of (0.5) u[n] 1 x[n] 0.5 0 −6 −4 −2 0 2 4 6 8 −6.2832 −3.1416 0 3.1416 6.2832 9.4248 −6.2832 −3.1416 0 3.1416 Frequency (rad/sample) 6.2832 9.4248 jω |X(e )| −0.5 −8 2 ∠ X(ejω) (o) .14 1 0 −9.4248 50 0 −50 −9.4248 11 Portland State University Dr. H. Nguyen ECE 223 Discrete-Time Fourier Transform Ver. 1.14 12 Page 208 EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan Example First-Order Filter Filteraa==−0.5 −0.5 Example 1: 4: First-Order Find tran n Fourier Transform of (−0.5) u[n] 1 x[n] 0.5 0 −0.5 −8 −6 −4 −2 0 2 4 6 8 ∠ X(ejω) (o) jω |X(e )| 2 1 0 −9.4248 50 −6.2832 −3.1416 0 3.1416 6.2832 9.4248 −6.2832 −3.1416 0 3.1416 Frequency (rad/sample) 6.2832 9.4248 0 −50 −9.4248 Portland State University ECE 223 Discrete-Time Fourier Transform Dr. H. Nguyen Ver. 1.14 13 Portland Page 209 Example 5: Workspace 14 EE351–Spectrum Analysis and Discrete Time Systems 13 Portland State University ECE 223 Discrete-Time Fourier Transform University of Saskatchewan Ver. 1.14 14 Example 2: Fourier Transform of A Pulse N = 5 Example 6: Pulse Transform for N = 5 Fourier Transform of p5[n] 10 8 6 jω P(e ) 14 4 2 0 −2 −9.4248 −6.2832 15 Portland State University Dr. H. Nguyen −3.1416 ECE 223 0 3.1416 6.2832 Discrete-Time Fourier Transform 9.4248 Ver. 1.14 16 Page 210 EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan Example 2: Fourier Transform of for A Pulse N = 10 Example 7: Pulse Transform N = 10 Fourier Transform of p [n] 10 20 10 P(ejω) P(ejω) 15 5 0 −9.4248 −6.2832 Portland State University −3.1416 ECE 223 0 3.1416 6.2832 Discrete-Time Fourier Transform Dr. H. Nguyen 9.4248 Ver. 1.14 17 Portland Page 211 Example 9: Inverse of Impulse Train EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan Convergence X(ejω ) = +∞ X x[n]e−jωn ; n=−∞ 1 x[n] = 2π Z X(ejω )ejωn dω 2π • Sufficient conditions of the convergence of the discrete time Fourier transform of a bounded discrete-time signal: – Finite duration: There exits an N such that x[n] = 0 for |n| > N . P∞ – Absolutely summable: n=−∞ |x[n]| < ∞. P∞ 2 – Finite energy: n=−∞ |x[n]| < ∞ • The synthesis equation always converges • There is no Gibb’s phenomenon Dr. H. Nguyen Page 212 EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan Properties Periodicity X(ej(ω+`2π) ) = X(ejω ) Linearity F a1 x1 [n] + a2 x2 [n] ⇐⇒ a1 X1 (ejω ) + a2 X2 (ejω ) Time Shifting F x[n − n0 ] ⇐⇒ e−jωn0 X(ejω ) Frequency Shifting F ejω0 n x[n] ⇐⇒ X(ej(ω−ω0 ) ) Conjugation F x∗ [n] ⇐⇒ X ∗ (e−jω ) Conjugate Symmetry If x[n] is real, then X(ejω ) = X ∗ (e−jω ) Differencing F x[n] − x[n − 1] ⇐⇒ (1 − e−jω )X(ejω ) This is similar to a continuous-time derivative. Dr. H. Nguyen Page 213 EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan Properties (Cont’d) Accumulation n X ∞ X 1 jω j0 x[n] ⇐⇒ X(e ) + πX(e ) δ(ω − 2π`) −jω 1 − e m=−∞ F `=−∞ This is similar to a continuous-time integration. Time Reversal F x[−n] ⇐⇒ X(e−jω ) Frequency Differentiation dX(ejω ) n · x[n] ⇐⇒ j dω F Parseval’s Relation +∞ X 1 |x[n]| = 2π n=−∞ 2 Z |X(ejω |2 dω 2π Convolution F y[n] = x[n] ∗ h[n] ⇐⇒ Y (ejω ) = X(ejω )H(ejω ) Multiplication 1 x[n] · w[n] ⇐⇒ 2π F Dr. H. Nguyen Z X(eju )W (ej(ω−u) )du 2π Page 214 EE351–Spectrum Analysis and Discrete Time Systems University of Saskatchewan DT Fourier Transform: Summary • X(ejω ) is a periodic function of ω with a fundamental period of 2π • Two discrete-time complex exponentials with frequencies that differ by a multiple of 2π are equal: ejωn = ej(ω+2`π)n • The highest perceivable discrete-time frequency is π radians per sample • The lowest perceivable discrete-time frequency is 0 radians per sample • The analysis equation may or may not converge, the synthesis equation always converges • The energy of the signal in the time-domain is proportional to energy of the DTFT in an interval of 2π • The DTFT shares many of the properties of CTFT and Fourier series Dr. H. Nguyen Page 215