2023/2/5 Signals and Systems Instructor: Chia-Wen Lin (林嘉文) Email: cwlin@ee.nthu.edu.tw Course website: https://eeclass.nthu.edu.tw/course/12470 Chapter 4 The Continuous-Time Fourier Transform 1 2023/2/5 Continuous Periodic Non-periodic Fourier Family Periodic Discrete Non-periodic Fourier Series (Chap. 3) Fourier Transform (Chap. 4) Discrete-time Fourier Series (Discrete Fourier Transform) (Sec. 3-6) Discrete-Time Fourier Transform (Chap. 5) Chapter 4 The Continuous-Time Fourier Transform Sec. 4.0 Introduction Sec. 4.1 Representation of Aperiodic Signals: The ContinuousTime Fourier Transform Sec. 4.2 The Fourier Transform for Periodic Signals Sec. 4.3 Properties of the Continuous-Time Fourier Transform Sec. 4.4 The Convolution Property Sec. 4.5 The Multiplication Property Sec. 4.6 Tables of Fourier Properties and of Basic Fourier Transform Pairs Sec. 4.7 Systems Characterized by Linear Constant Coefficient Differential Equations Sec. 4.8 Summary Basic Properties 2 2023/2/5 Sec. 4.1 Representation of Aperiodic Signals: The Continuous-Time Fourier Transform Key Concepts: (i) definition of the continuous-time Fourier transform; (ii) the Fourier series converges to the continuous-time Fourier transform when 𝑇 → ∞; (iii) the Dirichlet conditions where the continuous-time Fourier transform exists (absolutely integrable, a finite number of maxima and minima, and a finite number of discontinuities); (iv) the sinc function and its relation to the rectangular function P.285 4.1.1 Development of the Fourier Transform Representation of an Aperiodic Signal Fourier series can analyze a periodic signal. How do we analyze a nonperiodic signal? P.287 3 2023/2/5 +∞ 𝑥(𝑡) = 𝑎𝑘 𝑒 1 𝑇/2 −𝑗𝑘𝜔0 𝑡 𝑎𝑘 = න 𝑥(𝑡)𝑒 𝑑𝑡 𝑇 −𝑇/2 where 𝜔0 = 2𝜋/𝑇 𝑗𝑘𝜔0 𝑡 𝑘=−∞ Since 𝑥(𝑡) = 𝑥(𝑡) for |𝑡| < 𝑇/2, and also, since 𝑥(𝑡) = 0 outside this interval, 𝑎𝑘 = 1 𝑇/2 1 +∞ න 𝑥(𝑡)𝑒 −𝑗𝑘𝜔0 𝑡 𝑑𝑡 = න 𝑥(𝑡)𝑒 −𝑗𝑘𝜔0 𝑡 𝑑𝑡 𝑇 −𝑇/2 𝑇 −∞ Define +∞ 𝑋(𝑗𝜔) = න −∞ then 𝑎𝑘 = P.286-287 𝑥(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡 1 𝑋(𝑗𝑘𝜔0 ) 𝑇 𝑇 = 4𝑇1 𝑇 = 8𝑇1 𝑇 = 16𝑇1 P.286-287 4 2023/2/5 Definition 4.1 The Continuous-Time Fourier Transform and Its Inverse +∞ 𝑥(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡 𝑋(𝑗𝜔) = න Fourier transform: −∞ Inverse Fourier transform: 𝑥(𝑡) = 1 +∞ න 𝑋(𝑗𝜔)𝑒 𝑗𝜔𝑡 𝑑𝜔 2𝜋 −∞ These two equations are referred to as the Fourier transform pair. The Fourier transform can be viewed as the Fourier series in the case where the period 𝑇 → ∞ (the period is infinite) P.288 Supplement Alternative Definitions of Fourier Transform Pairs (1) Fourier transform: 𝑥(𝑡) = Inverse Fourier transform: 1 +∞ න 𝑋(𝑗𝜔)𝑒 −𝑗𝜔𝑡 𝑑𝜔 2𝜋 −∞ +∞ 𝑥(𝑡)𝑒 𝑗𝜔𝑡 𝑑𝑡 𝑋(𝑗𝜔) = න −∞ (2) Fourier transform: 1 +∞ න 𝑋(𝑗𝜔)𝑒 𝑗𝜔𝑡 𝑑𝜔 2𝜋 −∞ 𝑥(𝑡) = 1 +∞ න 𝑥(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡 2𝜋 −∞ Inverse Fourier transform: 𝑋(𝑗𝜔) = (3) Fourier transform: +∞ 𝑥(𝑡) = න 𝑋(𝑗𝑓)𝑒 𝑗2𝜋𝑓𝑡 𝑑𝑓 −∞ Inverse Fourier transform: +∞ 𝑋(𝑗𝑓) = න 𝑥(𝑡)𝑒 −𝑗2𝜋𝑓𝑡 𝑑𝑡 −∞ 5 2023/2/5 4.1.2 Convergence of Fourier Transforms Dirichlet conditions (the constraints for the convergence of Fourier transforms) 1. 𝑥(𝑡) be absolutely integrable +∞ න 𝑥(𝑡) 𝑑𝑡 < ∞ −∞ 2. 𝑥(𝑡) have a finite number of maxima and minima within any finite interval. 3. 𝑥(𝑡) have a finite number of discontinuities within any finite interval. Furthermore, each of these discontinuities must be finite. P.289 4.1.3 Examples of Continuous-Time Fourier Transforms [Example 4.1] 𝑥(𝑡) = 𝑒 −𝑎𝑡 𝑢(𝑡) 𝑎>0 ∞ ∞ 𝑋(𝑗𝜔) = න 𝑒 −𝑎𝑡 −𝑗𝜔𝑡 0 𝑋(𝑗𝜔) = 1 , 𝑎 + 𝑗𝜔 𝑒 1 𝑑𝑡 = − 𝑒 −(𝑎+𝑗𝜔)𝑡 ቤ 𝑎 + 𝑗𝜔 0 𝑎>0 P.290 6 2023/2/5 𝑋(𝑗𝜔) = 1 , 𝑎 + 𝑗𝜔 𝑎>0 P.291 [Example 4.3] 𝑥(𝑡) = 𝛿(𝑡) ∞ 𝑋(𝑗𝜔) = න 𝛿(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡 = 1 −∞ Note: Here, we apply the sifting property of the unit impulse function 𝑏 න 𝑦(𝑡)𝛿(𝑡 − 𝑡0 )𝑑𝑡 = 𝑦 𝑡0 if 𝑎 < 𝑡0 < 𝑏 𝑎 P.292 7 2023/2/5 [Example 4.4] 𝑥(𝑡) = ൝ 1, |𝑡| < 𝑇1 0, 𝑇1 > |𝑡| 𝑇1 𝑋(𝑗𝜔) = න 𝑒 −𝑗𝜔𝑡 𝑑𝑡 = 2 −𝑇1 sin 𝜔 𝑇1 𝜔 Fourier transform P.293 [Example 4.5] 𝑋(𝑗𝜔) = ൝ 𝑥(𝑡) = 1, |𝜔| < 𝑊 0, |𝜔| > 𝑊 1 𝑊 𝑗𝜔𝑡 sin 𝑊 𝑡 න 𝑒 𝑑𝜔 = 2𝜋 −𝑊 𝜋𝑡 Fourier transform P.294-295 8 2023/2/5 Fourier transform for different 𝑊 Fourier transform Fourier transform Fourier transform P.296 Definition 4.2 Sinc Function sinc(𝜃) = Specially, sinc(0) = 1 sin 𝜋 𝜃 𝜋𝜃 sinc(𝑘) = 0 where 𝑘 is a nonzero integer. P.295 9 2023/2/5 ∞ න 𝑒 −∞ −𝑗𝜔𝑡 ∞ න 1 ∞ 𝑗𝜔𝑡 𝑡 න 𝑒 sin 𝑐 𝑑𝑡 = Π 𝜔 2𝜋 −∞ 2𝜋 𝜔 Π 𝑡 𝑑𝑡 = sinc , 2𝜋 𝑒 −𝑗𝜔𝑡 sinc −∞ 𝑡 𝑑𝑡 = 2𝜋Π 𝜔 2𝜋 ∞ න 𝑒 −𝑗𝜔𝑡 sinc 𝑡 𝑑𝑡 = Π −∞ 𝜔 2𝜋 where Π(𝑡) means the rectangular function Π(𝑡) = 1 for |𝑡| < 1/2, Π(𝑡) = 0 otherwise. P.295-296 Sec. 4.2 The Fourier Transform for Periodic Signals Key concepts: (i) The Fourier transform for a periodic signal is equivalent to the Fourier series. The output is a linear combination of unit impulse functions. periodic function 𝔉 impulse train (ii) If 𝑥(𝑡) is a linear combination of 𝛿(𝑡 − 𝑘𝑇) where 𝑘 is some integer, then the Fourier transform of 𝑥(𝑡) periodic. impulse train 𝔉 periodic function P.296-297 10 2023/2/5 If 𝑥(𝑡) is periodic +∞ 𝑥(𝑡) = 𝑎𝑘 𝑒 𝑗𝑘𝜔0 𝑡 𝑘=−∞ then +∞ 𝑋(𝑗𝜔) = 2𝜋𝑎𝑘 𝛿(𝜔 − 𝑘𝜔0 ) 𝑘=−∞ P.297 [Example 4.6] Fourier series coefficients sin 𝑘 𝜔0 𝑇1 𝑎𝑘 = 𝜋𝑘 +∞ 𝑋(𝑗𝜔) = 𝑘=−∞ 2 sin 𝑘 𝜔0 𝑇1 𝛿(𝜔 − 𝜔0 ) 𝑘 P.297-298 11 2023/2/5 [Example 4.7] 𝑥(𝑡) = sin𝜔0 𝑡 𝑎1 = 1 2𝑗 𝑎−1 = − 1 2𝑗 𝑎𝑘 = 0, 𝑘 ≠ 1 or − 1 1 2 𝑎𝑘 = 0, 𝑘 ≠ 1 or − 1 𝑥(𝑡) = cos𝜔0 𝑡 𝑎1 = 1 2 𝑎−1 = P.298-299 [Example 4.8] +∞ 𝑥(𝑡) = 𝛿(𝑡 − 𝑘𝑇) 𝑘=−∞ 𝑎𝑘 = 1 +𝑇/2 1 න 𝛿(𝑡)𝑒 −𝑗𝑘𝜔0 𝑡 𝑑𝑡 = 𝑇 −𝑇/2 𝑇 (sifting property in Table 2.1) +∞ 2𝜋 2𝜋𝑘 𝑋(𝑗𝜔) = 𝛿 𝜔− 𝑇 𝑇 𝑘=−∞ Fourier transform P.299-300 12 2023/2/5 Sec. 4.3 Properties of the CTFT Key concepts Learning several properties of the continuous-time Fourier transform, including (i) linearity, (ii) time shifting, (iii) frequency shifting, (iv) conjugate symmetry, (v) symmetry property for a real input, (vi) symmetry property for a real and even(or odd) input, (vii) differentiation in time, (viii) differentiation in frequency, (ix) integration, (x) scaling, (xi) time reversal, and (xii) duality, and (xiii) Parseval’s relation. These properties will be summarized in Sec. 4.6. Note that the timeshifting and the frequency-shifting properties form a duality pair. The differentiation in time and the differentiation in frequency properties also form a duality pair. P.300 +∞ 𝑥(𝑡) 𝑒 −𝑗𝜔𝑡 𝑑𝑡 𝑋(𝑗𝜔) = න 𝑥(𝑡) = −∞ 𝑥(𝑡) 𝔉 1 +∞ න 𝑋(𝑗𝜔) 𝑒 𝑗𝜔𝑡 𝑑𝜔 2𝜋 −∞ 𝑋(𝑗𝜔) 4.3.1 Linearity If 𝑥(𝑡) 𝔉 𝑦(𝑡) 𝑋(𝑗𝜔) 𝔉 𝑌(𝑗𝜔) then 𝑎𝑥(𝑡) + 𝑏𝑦(𝑡) 𝔉 𝑎𝑋(𝑗𝜔) + 𝑏𝑌(𝑗𝜔) P.301 13 2023/2/5 4.3.2 Time Shifting 𝑥(𝑡 − 𝑡0 ) 𝔉 𝑒 −𝑗𝜔𝑡0 𝑋(𝑗𝜔) (Proof): 1 ∞ 𝑥(𝑡) = න 𝑋(𝑗𝜔) 𝑒 𝑗𝜔𝑡 𝑑𝜔 2𝜋 −∞ 1 +∞ 𝑥(𝑡 − 𝑡0 ) = න 𝑋(𝑗𝜔)𝑒 𝑗𝜔(𝑡−𝑡0 ) 𝑑𝜔 2𝜋 −∞ 1 +∞ −𝑗𝜔𝑡 0 𝑋(𝑗𝜔) 𝑒 𝑗𝜔𝑡 𝑑𝜔 = න 𝑒 2𝜋 −∞ P.301-302 [Example 4.9] 𝑥(𝑡) = 1 𝑥 (𝑡 − 2.5) + 𝑥2 (𝑡 − 2.5) 2 1 From [Example 4.4] 𝑋1 (𝑗𝜔) = 2 sin( 𝜔/2) 𝜔 𝑋(𝑗𝜔) = 𝑒 −𝑗5𝜔/2 𝑋2 (𝑗𝜔) = 2 sin( 3𝜔/2) 𝜔 sin( 𝜔/2) + 2 sin( 3𝜔/2) 𝜔 P.302-303 14 2023/2/5 4.3.3 Conjugation and Conjugate Symmetry If 𝔉 𝑥(𝑡) 𝑋(𝑗𝜔) then 𝔉 𝑥 ∗ (𝑡) 𝑋 ∗ (−𝑗𝜔) (Proof): ∗ +∞ ∗ 𝑋 (𝑗𝜔) = න 𝑥(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡 +∞ +∞ =න −∞ 𝑥 ∗ (𝑡)𝑒 𝑗𝜔𝑡 𝑑𝑡 𝑋 ∗ (−𝑗𝜔) = න 𝑥 ∗ (𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡 −∞ −∞ Moreover, if 𝑥(𝑡) is real, then 𝑋(−𝑗𝜔) = 𝑋 ∗ (𝑗𝜔) [𝑥 𝑡 : real] P.303-304 If 𝑥(𝑡) is real 𝑋(−𝑗𝜔) = 𝑋 ∗ (𝑗𝜔) [𝑥 𝑡 : real] ℛ𝑒{𝑋(𝑗𝜔)} = Re{𝑋(−𝑗𝜔)} Even{𝑥(𝑡)} 𝔉 Re{𝑋(𝑗𝜔)} ℐ𝑚{𝑋(𝑗𝜔)} = −Im{𝑋(−𝑗𝜔)} Odd{𝑥(𝑡)} 𝔉 𝑗Im{𝑋(𝑗𝜔)} Examples: In Example 4.1, 𝑥(𝑡) = 𝑒 −𝑎𝑡 𝑢(𝑡) 𝑋(𝑗𝜔) = 1 𝑎 + 𝑗𝜔 𝑋(−𝑗𝜔) = 1 = 𝑋 ∗ (𝑗𝜔) 𝑎 − 𝑗𝜔 P.304 15 2023/2/5 [Example 4.10] 𝑥 𝑡 = 𝑒 −𝑎 𝑡 = 𝑒 −𝑎𝑡 𝑢(𝑡) + 𝑒 𝑎𝑡 𝑢(−𝑡) From Example 4.1, 𝑒 −𝑎𝑡 𝑢 𝑡 𝔉 1 𝑎 + 𝑗𝜔 If 𝑒 −𝑎𝑡 𝑢(𝑡) + 𝑒 𝑎𝑡 𝑢(−𝑡) 2 −𝑎𝑡 𝑎𝑡 −𝑎|𝑡| = 𝑒 𝑢(𝑡) + 𝑒 𝑢(−𝑡) = 𝑒 𝑥(𝑡) = 2 Ev 𝑒 −𝑎𝑡 𝑢(𝑡) = 2 𝑋(𝑗𝜔) = 2Re 1 𝑎 − 𝑗𝜔 2𝑎 = 2Re 2 = 2 2 𝑎 + 𝑗𝜔 𝑎 +𝜔 𝑎 + 𝜔2 P.305-306 16 2023/2/5 4.3.4 Differentiation and Integration Differentiation Property Proof: 𝑑𝑥(𝑡) 𝑑𝑡 𝔉 𝑗𝜔𝑋(𝑗𝜔) 𝑑𝑥(𝑡) 1 +∞ = න 𝑗𝜔𝑋(𝑗𝜔)𝑒 𝑗𝜔𝑡 𝑑𝜔 𝑑𝑡 2𝜋 −∞ Integration Property 𝑡 න 𝑥(𝜏)𝑑𝜏 −∞ 𝔉 1 𝑋(𝑗𝜔) + 𝜋𝑋(0)𝛿(𝜔) 𝑗𝜔 P.306 [Example 4.11] 𝑔(𝑡) = 𝛿(𝑡) 𝔉 𝐺(𝑗𝜔) = 1 𝑡 𝑥(𝑡) = න 𝑔(𝜏)𝑑𝜏 = 𝑢(𝑡) (unit step function) −∞ 𝑋(𝑗𝜔) = 𝐺(𝑗𝜔) 1 + 𝜋𝐺(0)𝛿(𝜔) = + 𝜋𝛿(𝜔) 𝑗𝜔 𝑗𝜔 P.307 17 2023/2/5 [Example 4.12] 𝐺(𝑗𝜔) = 2 sin 𝜔 − 𝑒 𝑗𝜔 − 𝑒 −𝑗𝜔 𝜔 𝑋(𝑗𝜔) = 𝐺(𝑗𝜔) + 𝜋𝐺(0)𝛿(𝜔) 𝑗𝜔 𝑋(𝑗𝜔) = 2 sin 𝜔 2 cos 𝜔 − 𝑗𝜔 2 𝑗𝜔 since 𝐺(0) = 0 P.307-308 4.3.5 Time and Frequency Scaling If 𝑥(𝑡) then since 𝔉 𝑋(𝑗𝜔) 𝑥(𝑡) = +∞ 𝔉{𝑥(𝑎𝑡)} = න after substituting 𝑎𝑡 by 𝜏 1 +∞ න 𝑋(𝑗𝜔)𝑒 𝑗𝜔𝑡 𝑑𝜔 2𝜋 −∞ 𝑥(𝑎𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡 −∞ 1 +∞ න 𝑥(𝜏)𝑒 −𝑗(𝜔/𝑎)𝜏 𝑑𝜏, 𝑎 > 0 𝑎 −∞ 𝔉{𝑥(𝑎𝑡)} = 1 +∞ − න 𝑥(𝜏)𝑒 −𝑗(𝜔/𝑎)𝜏 𝑑𝜏, 𝑎 < 0 𝑎 −∞ 𝑥(𝑎𝑡) P.308-309 𝔉 1 𝑗𝜔 𝑋 |𝑎| 𝑎 Specially, when 𝑎 = −1, 𝑥(−𝑡) 𝔉 𝑋(−𝑗𝜔) (time reversal property) 18 2023/2/5 4.3.6 Duality (1) Duality for Transform Pairs In summary, if 𝔉 𝑥 𝑡 𝑋 𝑗𝜔 𝑋 𝑗𝑡 then 1 𝑋 𝑗𝑡 2𝜋 𝔉 𝔉 2𝜋𝑥 −𝜔 𝑥 −𝜔 P.309-310 [Example 4.13] From Example 4.2 Therefore, (Proof): 2 1 + 𝜔2 𝑒 −|𝑡| 𝔉 𝑒 −|𝑡| 2 = 2𝜋𝑒 −|𝜔| 1 + 𝑡2 1 ∞ 2 = න 𝑒 𝑗𝜔𝑡 𝑑𝜔 2𝜋 −∞ 1 + 𝜔 2 ∞ 2𝜋𝑒 −|𝑡| = න −∞ ∞ 2𝜋𝑒 P.310-311 −|𝜔| =න −∞ 2 𝑒 −𝑗𝜔𝑡 𝑑𝜔 1 + 𝜔2 2 𝑒 −𝑗𝜔𝑡 𝑑𝑡 1 + 𝑡2 19 2023/2/5 (2) Duality for Properties 𝑑𝑥(𝑡) 𝑑𝑡 𝑥(𝑡 − 𝑡0 ) 𝑡 න 𝑥(𝜏)𝑑𝜏 𝔉 −∞ 𝔉 𝔉 Duality 𝑗𝜔𝑋(𝑗𝜔) −𝑗𝑡𝑥(𝑡) 𝑒 −𝑗𝜔𝑡0 𝑋(𝑗𝜔) 𝑒 𝑗𝜔0 𝑡 𝑥(𝑡) 1 𝑋(𝑗𝜔) + 𝜋𝑋(0)𝛿(𝜔) 𝑗𝜔 − 𝔉 𝔉 𝑑𝑋(𝑗𝜔) 𝑑𝜔 𝑋(𝑗(𝜔 − 𝜔0 )) 1 𝑥(𝑡) + 𝜋𝑥(0)𝛿(𝑡) 𝑗𝑡 𝔉 𝜔 න 𝑋(𝑗𝜂)𝑑𝜂 −∞ P.311 4.3.7 Parseval’s Relation Parseval’s Relation (Energy Preservation) +∞ |𝑥(𝑡)|2 𝑑𝑡 = න −∞ (Proof) +∞ න 1 +∞ න |𝑋(𝑗𝜔)|2 𝑑𝜔 2𝜋 −∞ +∞ 𝑥(𝑡) 2 𝑑𝑡 = න −∞ 𝑥(𝑡)𝑥 ∗ (𝑡)𝑑𝑡 −∞ +∞ =න −∞ 𝑥(𝑡) 1 +∞ ∗ න 𝑋 (𝑗𝜔)𝑒 −𝑗𝜔𝑡 𝑑𝜔 𝑑𝑡. 2𝜋 −∞ +∞ 1 +∞ ∗ 1 +∞ −𝑗𝜔𝑡 = න 𝑋 (𝑗𝜔) න 𝑥(𝑡)𝑒 𝑑𝑡 𝑑𝜔 = න |𝑋(𝑗𝜔)|2 𝑑𝜔 2𝜋 −∞ 2𝜋 −∞ −∞ P.312 20 2023/2/5 [Example 4.14] If 𝑋(𝑗𝜔) is ∞ evaluate (i) 𝐸 = න 𝑥(𝑡) 2 𝑑𝑡 and (ii) 𝐷 = −∞ ∞ 𝑥(𝑡) 2 𝑑𝑡 = 𝐸=න −∞ 𝑑 𝑥(𝑡)|𝑡 = 0 𝑑𝑡 1 ∞ 5 න 𝑋(𝑗𝜔) 2 𝑑𝜔 = 2𝜋 −∞ 8 𝔉 𝑑 𝑥(𝑡) 𝑗𝜔𝑋(𝑗𝜔) = 𝐺(𝑗𝜔) 𝑑𝑡 1 ∞ 1 ∞ 𝐷 = 𝑔(0) = න 𝐺(𝑗𝜔)𝑑𝜔 = න 𝑗𝜔𝑋(𝑗𝜔)𝑑𝜔 = 0 2𝜋 −∞ 2𝜋 −∞ 𝑔(𝑡) = P.312-313 If 𝑋(𝑗𝜔) is evaluate (i) 𝐸 = න ∞ −∞ ∞ 𝐸=න −∞ 𝑥(𝑡) 2 𝑑𝑡 and (ii) 𝐷 = 𝑑 𝑥(𝑡)|𝑡 = 0 𝑑𝑡 1 ∞ 𝑥(𝑡) 𝑑𝑡 = න 𝑋(𝑗𝜔) 2 𝑑𝜔 = 1 2𝜋 −∞ 2 𝔉 𝑑 𝑥(𝑡) 𝑗𝜔𝑋(𝑗𝜔) = 𝐺(𝑗𝜔) 𝑑𝑡 1 ∞ 1 ∞ −1 𝐷 = 𝑔(0) = න 𝐺(𝑗𝜔)𝑑𝜔 = න 𝑗𝜔𝑋(𝑗𝜔)𝑑𝜔 = 2𝜋 −∞ 2𝜋 −∞ 2 𝜋 𝑔(𝑡) = P.313 21 2023/2/5 Sec. 4.4 The Convolution Property Key concepts: (i) The convolution property; (ii) the application of the convolution property for determining the output of an LTI system; (iii) convolution with an impulse function is equal to an identity or a delay operation P.314 Convolution Property 𝔉 𝑦(𝑡) = ℎ(𝑡) ∗ 𝑥(𝑡) (Proof): 𝑌(𝑗𝜔) = 𝐻(𝑗𝜔)𝑋(𝑗𝜔) +∞ If then 𝑦(𝑡) = න 𝑥(𝜏)ℎ(𝑡 − 𝜏)𝑑𝜏 −∞ +∞ +∞ 𝑌(𝑗𝜔) = න න −∞ 𝑥(𝜏)ℎ(𝑡 − 𝜏)𝑑𝜏 𝑒 −𝑗𝜔𝑡 𝑑𝑡. −∞ +∞ 𝑌(𝑗𝜔) = න +∞ 𝑥(𝜏) න −∞ ℎ(𝑡 − 𝜏)𝑒 −𝑗𝜔𝑡 𝑑𝑡 𝑑𝜏 −∞ +∞ 𝑌(𝑗𝜔) = න +∞ 𝑥(𝜏)𝑒 −𝑗𝜔𝑡 𝐻(𝑗𝜔)𝑑𝜏 = 𝐻(𝑗𝜔) න −∞ 𝑥(𝜏)𝑒 −𝑗𝜔𝑡 𝑑𝜏 −∞ = 𝑋(𝑗𝜔)𝐻(𝑗𝜔) P.314-315 22 2023/2/5 Three equivalent LTI systems. Here, each block represents an LTI system with the indicated frequency response. P.316 4.4.1 Examples [Example 4.15] Consider a continuous-time LTI system with impulse response ℎ(𝑡) = 𝛿(𝑡 − 𝑡0 ) 𝐻(𝑗𝜔) = 𝑒 −𝑗𝜔𝑡0 𝑌(𝑗𝜔) = 𝐻(𝑗𝜔)𝑋(𝑗𝜔) = 𝑒 −𝑗𝜔𝑡0 𝑋(𝑗𝜔). 𝑦(𝑡) = 𝑥(𝑡 − 𝑡0 ) In other words, 𝛿(𝑡 − 𝑡0 ) ∗ 𝑥(𝑡) = 𝑥(𝑡 − 𝑡0 ) P.317 23 2023/2/5 [Example 4.16] 𝑦(𝑡) = 𝑑𝑥(𝑡) 𝑑𝑡 𝑌(𝑗𝜔) = 𝑗𝜔𝑋(𝑗𝜔) The frequency response of a differentiator is 𝐻(𝑗𝜔) = 𝑗𝜔 P.317 [Example 4.18] ideal lowpass filter 𝐻(𝑗𝜔) = ൝ 1 |𝜔| < 𝜔𝑐 0 |𝜔| > 𝜔𝑐 ℎ(𝑡) = sin𝜔𝑐 𝑡 𝜋𝑡 P.318-319 24 2023/2/5 [Example 4.19] 𝑥(𝑡) = 𝑒 −𝑏𝑡 𝑢(𝑡), 𝑋(𝑗𝜔) = If 𝑏>0 1 𝑏 + 𝑗𝜔 ℎ(𝑡) = 𝑒 −𝑎𝑡 𝑢(𝑡), 𝐻(𝑗𝜔) = 𝑎>0 1 𝑎 + 𝑗𝜔 𝑦(𝑡) = ℎ(𝑡) ∗ 𝑥(𝑡) then 𝑌(𝑗𝜔) = 1 (𝑎 + 𝑗𝜔)(𝑏 + 𝑗𝜔) 𝑦(𝑡) = 𝑌(𝑗𝜔) = 1 1 1 − 𝑏 − 𝑎 𝑎 + 𝑗𝜔 𝑏 + 𝑗𝜔 1 𝑒 −𝑎𝑡 𝑢(𝑡) −𝑒 −𝑏𝑡 𝑢(𝑡) 𝑏−𝑎 P.320-321 [Example 4.20] 𝑥(𝑡) = 𝑋(𝑗𝜔) = ቊ If then P.321-322 sin𝜔𝑐 𝑡 𝜋𝑡 sin𝜔𝑖 𝑡 𝜋𝑡 ℎ(𝑡) = 1, 𝜔 ≤ 𝜔𝑖 0, elsewhere 𝐻(𝑗𝜔) = ቊ 1, 0, 𝜔 ≤ 𝜔𝑐 elsewhere 𝑦(𝑡) = ℎ(𝑡) ∗ 𝑥(𝑡) 𝑌(𝑗𝜔) = ቊ 1, 0, 𝜔 ≤ min 𝜔𝑖 , 𝜔𝑐 elsewhere sin𝜔𝑐 𝑡 , 𝜋𝑡 𝑦 𝑡 = sin𝜔𝑖 𝑡 , 𝜋𝑡 if 𝜔𝑐 ≤ 𝜔𝑖 if 𝜔𝑖 ≤ 𝜔𝑐 25 2023/2/5 Sec. 4.5 The Multiplication Property Key concept: The multiplication property. It and the convolution property form a duality pair. 1 +∞ 𝑟(𝑡) = 𝑠(𝑡)𝑝(𝑡) ↔ 𝑅(𝑗𝜔) = න 𝑆(𝑗𝜃)𝑃(𝑗(𝜔 − 𝜃))𝑑𝜃 2𝜋 −∞ P.322 𝑟(𝑡) = 𝑠(𝑡)𝑝(𝑡) ↔ 𝑅(𝑗𝜔) = [Example 4.21] 𝑝(𝑡) = cos𝜔0 𝑡 𝑅(𝑗𝜔) = 1 +∞ න 𝑆(𝑗𝜃)𝑃(𝑗(𝜔 − 𝜃))𝑑𝜃 2𝜋 −∞ 𝑃(𝑗𝜔) = 𝜋𝛿(𝜔 − 𝜔0 ) + 𝜋𝛿(𝜔 + 𝜔0 ) 1 +∞ 1 1 න 𝑆(𝑗𝜃)𝑃(𝑗(𝜔 − 𝜃))𝑑𝜃 = 𝑆(𝑗(𝜔 − 𝜔0 )) + 𝑆(𝑗(𝜔 + 𝜔0 )), 2𝜋 −∞ 2 2 P.323-324 26 2023/2/5 [Example 4.21] 1 +∞ 𝑟(𝑡) = 𝑠(𝑡)𝑝(𝑡) ↔ 𝑅(𝑗𝜔) = න 𝑆(𝑗𝜃)𝑃(𝑗(𝜔 − 𝜃))𝑑𝜃 2𝜋 −∞ 𝑝(𝑡) = cos𝜔0 𝑡 𝑃(𝑗𝜔) = 𝜋𝛿(𝜔 − 𝜔0 ) + 𝜋𝛿(𝜔 + 𝜔0 ) 1 +∞ 1 1 𝑅(𝑗𝜔) = න 𝑆(𝑗𝜃)𝑃(𝑗(𝜔 − 𝜃))𝑑𝜃 = 𝑆(𝑗(𝜔 − 𝜔0 )) + 𝑆(𝑗(𝜔 + 𝜔0 )), 2𝜋 −∞ 2 2 P.323-324 [Example 4.22] 𝑔(𝑡) = 𝑟(𝑡)𝑝(𝑡) 𝑟(𝑡) is the output of Example 4.21 𝑝(𝑡) = cos𝜔0 𝑡 +∞ 𝐺(𝑗𝜔) = 1 න 𝑅(𝑗𝜃)𝑃(𝑗(𝜔 − 𝜃))𝑑𝜃 2𝜋 −∞ P.324-325 27 2023/2/5 [Example 4.23] 𝑥(𝑡) = 𝑥(𝑡) = 𝜋 sin( 𝑡) 𝜋𝑡 sin( 𝑡) sin( 𝑡/2) 𝜋𝑡 2 sin( 𝑡/2) 𝜋𝑡 1 sin( 𝑡) sin( 𝑡/2) 𝑋(𝑗𝜔) = 𝔉 ∗𝔉 2 𝜋𝑡 𝜋𝑡 P.325 4.5.1 Frequency-Selective Filtering with Variable Center Frequency Implementation of a bandpass filter using amplitude modulation with a complex exponential carrier. P.326 28 2023/2/5 equivalent bandpass filter P.326-327 Sec. 4.6 Tables of Fourier Properties and of Basic Fourier Transform Pairs Key concept: The summaries in Tables 4.1 and 4.2. P.328 29 2023/2/5 (followed by the next page) P.328 general form of Parseval’s relation: +∞ න −∞ 𝑥(𝑡)𝑦 ∗ (𝑡)𝑑𝑡 = 1 +∞ න 𝑋(𝑗𝜔)𝑌 ∗ (𝑡)𝑑𝜔 2𝜋 −∞ P.359 30 2023/2/5 P.329 (followed by the next page) P.329 31 2023/2/5 Sec. 4.7 Systems Characterized by Linear Constant Coefficient Differential Equations Key concept: Using the differentiation property of the Fourier transform to analyze the linear differential equation with constant coefficients P.330 linear constant-coefficient differential equation 𝑁 𝑀 𝑘=0 𝑘=0 𝑀 𝑑𝑘 𝑦(𝑡) 𝑑 𝑘 𝑥(𝑡) 𝑎𝑘 = 𝑏 𝑘 𝑑𝑡𝑘 𝑑𝑡𝑘 𝑁 𝑎𝑘 (𝑗𝜔)𝑘 𝑌(𝑗𝜔) = 𝑏𝑘 (𝑗𝜔)𝑘 𝑋(𝑗𝜔) 𝑘=0 𝑘=0 frequency response 𝐻(𝑗𝜔) for an LTI system 𝐻(𝑗𝜔) = 𝑘 𝑌(𝑗𝜔) σ𝑀 𝑘=0 𝑏𝑘 (𝑗𝜔) = 𝑁 𝑋(𝑗𝜔) σ𝑘=0 𝑎𝑘 (𝑗𝜔)𝑘 P.331-332 32 2023/2/5 [Example 4.25] 𝑑 2 𝑦(𝑡) 𝑑𝑦(𝑡) 𝑑𝑥(𝑡) + 4 + 3𝑦(𝑡) = + 2𝑥(𝑡) 𝑑𝑡 2 𝑑𝑡 𝑑𝑡 𝐻(𝑗𝜔) = (𝑗𝜔) + 2 + 4(𝑗𝜔) + 3 (𝑗𝜔)2 1 1 𝑗𝜔 + 2 2 𝐻(𝑗𝜔) = = + 2 (𝑗𝜔 + 1)(𝑗𝜔 + 3) 𝑗𝜔 + 1 𝑗𝜔 + 3 1 1 ℎ(𝑡) = 𝑒 −𝑡 𝑢(𝑡) + 𝑒 −3𝑡 𝑢(𝑡) 2 2 impulse response P.331-332 [Example 4.26] 𝑑2 𝑦(𝑡) 𝑑𝑦(𝑡) 𝑑𝑥(𝑡) + 4 + 3𝑦(𝑡) = + 2𝑥(𝑡) 𝑑𝑡 2 𝑑𝑡 𝑑𝑡 𝐻(𝑗𝜔) = input (𝑗𝜔) + 2 + 4(𝑗𝜔) + 3 (𝑗𝜔)2 𝑥(𝑡) = 𝑒 −𝑡 𝑢(𝑡) 𝑌(𝑗𝜔) = 𝐻(𝑗𝜔)𝑋(𝑗𝜔) = 𝑌(𝑗𝜔) = 𝑗𝜔 + 2 1 𝑗𝜔 + 2 = . (𝑗𝜔 + 1)(𝑗𝜔 + 3) 𝑗𝜔 + 1 (𝑗𝜔 + 1)2 (𝑗𝜔 + 3) 𝐴11 𝐴12 𝐴21 + + 2 𝑗𝜔 + 1 (𝑗𝜔 + 1) 𝑗𝜔 + 3 𝑦(𝑡) = P.332-333 1 1 ℎ(𝑡) = 𝑒 −𝑡 𝑢(𝑡) + 𝑒 −3𝑡 𝑢(𝑡) 2 2 1 𝐴11 = , 4 1 𝐴12 = , 2 𝐴11 = − 1 4 1 −𝑡 1 −𝑡 1 −3𝑡 𝑒 + 𝑡𝑒 − 𝑒 𝑢(𝑡) 4 2 4 33 2023/2/5 Sec. 4.8 Summary The Fourier transform possesses a wide variety of important properties that describe how different characteristics of signals are reflected in their transforms. Fourier analysis are particularly well suited to the examination of LTI systems characterized by linear constant-coefficient differential equations. P.333 34