ELG3175 Introduction to Communication Systems Frequency-Domain Representation: Fourier Series and Fourier Transform Generalized Fourier Series • We are given a set of functions {φn(t)}n=1,2,…,N which are mutually orthogonal on the interval to ≤ t ≤ to + T. t o +T ∫ to i≠ j 0, φi (t )φ (t )dt = cn , i = j = n * j • If cn = 1 for all n, we call this an orthonormal set of functions. Lecture 2 Generalized Fourier Series (2) • We can find an approximation for any function x(t), on the interval (to, to + T) by xa(t) which is a weighted sum of the functions in the orthogonal set of functions : N x a (t ) = ∑X n φ n (t ) n =1 • We wish to find the approximation that minimises the mean square error between x(t) and xa(t): t o +T εN = ∫ 2 x(t ) − x a (t ) dt to Lecture 2 Generalized Fourier Series (3) • We can show that 1 Xn = cn t o +T ∫ x(t )φ * n (t )dt (1) to • And the best approximation is N x a (t ) = ∑X n φ n (t ) n =1 • Where Xn is given by (1). Lecture 2 The complex exponential function e j 2πnf ot = cos(2πnf o t ) + j sin( 2πnf o t ) n is an integer This is a periodic fiunction with period Tp. e j 2πnf o t =e j 2πnf o (t +T p ) =e j 2πnf o t e j 2πnf oT p The second exponential equals 1 when nfoTp is an integer. Lecture 2 Orthogonality and cn On the interval to ≤ t ≤ to+T, for fo = 1/T. t o +T t o +T ∫ ∫ φ n (t )φ m* (t )dt = t o +T t o +T ∫e to t o +T = ∫ dt = T to ∫ e j 2π ( n − m ) f ot dt to For m≠n For m=n cn = e j 2πnf ot e − j 2πnf ot dt = to to j 2π ( n − m ) f o t t o +T ∫e dt to m=n j 2π ( n − m ) f o t e j 2π ( n − m ) f o (to +T ) − e j 2π ( n − m ) f oto dt = j 2π (n − m) f o e j 2π ( n − m ) f oto e j 2π ( n − m ) f oT − e j 2π ( n − m ) f oto = j 2π (n − m) f o =0 Lecture 2 Complex exponential Fourier Series On the interval to ≤ t ≤ to + T the complex exponential Fourier Series of x(t) is given by : ∞ x(t ) = j 2πnf o t X e = ∑ n n = −∞ where 1 Xn = T ∞ ∑ X ne j 2πnt T n = −∞ t o +T ∫ x(t )e − j 2πnf ot dt to Lecture 2 Complex Exponential Fourier Series for periodic signals ∞ • Consider the signal t ≤ ∞. ∑ X n e j 2πnf ot on the interval -∞ ≤ n = −∞ • We know that complex exponential functions are periodic. • The fundamental period of a complex exponential is T/|n|. • The period of a sum of periodic signals is the lowest common multiple (LCM) of the individual periods • In this case, the fundamental period is T. Lecture 2 Complex Exponential Fourier Series for periodic signals (2) ∞ • ∑ X n e j 2πnf ot is periodic with period T = 1/fo. n = −∞ • The fundamental frequency fo is the inverse of the period. • Therefore if x(t) is periodic with period T, then ∞ ∑X ne j 2πnf o t =x(t) pour -∞ < t < ∞ n = −∞ • Therefore if x(t) is periodic with period T then its Fourier series is an exact representation . In other words, ∞ x(t) = ∑ X n e j 2πnf ot n = −∞ Lecture 2 Complex Exponential Fourier Series for periodic signals (3) • We can use any period to calculate the Fourier coefficients. – Therefore we write 1 Xn = T ∫ x(t )e − j 2πnf o t dt T Lecture 2 Example x(t) … A … 0.25 0.5 0.75 t -A Trouvez la série de Fourier exponentielle complexe du signal périodique x(t) Find the complex exponential Fourier Series of the periodic signal x(t). Lecture 2 Solution • We need to find – The period of x(t) as well as fo. – The coefficients Xn – Then the Fourier series itself ∞ ∑ X n e j 2πnf t o n = −∞ Lecture 2 Solution (2) • In our example, the period is 0.5, therefore fo = 2. • The set of functions is therefore ej4πnt. • Therefore 0.5 ∫ X n = 2 x(t )e − j 4πnt dt 0 0.25 = 2 ∫ Ae 0 0.5 − j 4πnt dt − ∫ Ae 0.25 0.25 − j 4πnt − j 4πnt dt 0.5 1 1 = 2 A − e + e − j 4πnt j 4πn j 4πn 0 0 . 25 1 1 1 1 e − jπn + e − j 2πn − e − jπnt = 2 A − + j 4πn j 4πn j 4πn j 4πn 1 1 A = 2 A − e − jπn = 1 − (− 1)n j 2πn j 2πn jπn [ ] Lecture 2 Solution continued • Forn = 0, we have X0 = 0/0. 0.5 ∫ X 0 = 2 x(t )dt = 0 0 ∞ x(t ) = ∑ n = −∞ n is odd ∞ 2 A j 4πnt 2A e =∑ e j 4π ( 2i −1)t jπn i = −∞ jπ ( 2i − 1) Lecture 2 Hermetian symmetry for real valued signals • Let us assume that x(t) is a real-valued signal. • In other words Im{x(t)} = 0. • Then Xn* is given by : * 1 − j 2πnf o t * Xn = x(t )e dt T T 1 − j 2πnf o t * = x(t )e dt T T 1 = x * (t )e j 2πnf ot dt T T 1 = x(t )e − j 2π ( − n) f o t dt = X − n T ∫ ∫( ) ∫ ∫ T Lecture 2 Trigonometric Fourier Series • Assuming x(t) is real, the real part of Xn is : 1 − j 2πnf o t Re{ X n } = Re x(t )e dt T T 1 = Re x(t )(cos 2πnf o t − j sin 2πnf o t )dt T T 1 1 = Re x(t ) cos 2πnf o tdt − j x(t ) sin 2πnf o tdt T T T T 1 = x(t ) cos 2πnf o tdt T ∫ ∫ ∫ ∫ ∫ T Lecture 2 Trigonometric Fourier Series (2) • Therefore the imaginary part of Xn is: Im{X n } = − 1 T ∫ x(t ) sin 2πnf tdt o T • The complex exponential Fourier series can be written as : ∞ x(t ) = ∑ X n e j 2πnf o t n = −∞ = X0 + ∞ ∑ (X − j 2πnf o t j 2πnf o t e + X e n −n ) n =1 Lecture 2 Trigonometric Fourier Series (3) • Since x(t) is real, X-n = Xn*, ∞ x(t ) = X 0 + ∑ {(Re{ X n =1 j Im{ X n })(cos(2πnf o t ) + j sin(2πnf o t ) ) + (Re{ X n } − j Im{ X n })(cos(2πnf o t ) − j sin(2πnf o t ) )} ∞ = X0 + n}+ ∑ (2 Re{X n } cos( 2πnf o t ) − 2 Im{ X n } sin( 2πnf o t ) n =1 ∞ = a0 + ∑a n =1 ) ∞ n cos(2πnf o t ) + ∑b n =1 n sin(2πnf o t ) a0 = X 0 = 1 T ∫ x(t )dt T 2 x(t ) cos 2πnf o tdt T T 2 bn = −2 Im{ X n } = x(t ) sin 2πnf o tdt T a n = 2 Re{ X n } = ∫ ∫ T Lecture 2 Example x(t) A … … 0.25 0.5 0.75 t -A ∞ x(t ) = ∑ n = −∞ n impaire ∞ 2 A j 4πnt 2 A j 4πnt e = ∑ − j e jπn πn n = −∞ n impaire X0 = 0, Re{Xn} = 0 and Im{Xn} = -2A/πn for odd values of n. Therefore bn = 4A/πn for odd values of n. Lecture 2 Example continued ∞ x(t ) = ∑ n =1 n impaire N Let x N (t ) = ∑ i =1 4A sin 4πnt = πn ∞ ∑ i =1 4A sin 4π (2i − 1)t π (2i − 1) 4A sin 4π (2i − 1)t π (2i − 1) represent the first N terms of x(t). Lecture 2 Lecture 2 Fourier Transform • The Fourier transform is a frequency-dependent function that is an extension of the Fourier series to non periodic functions. • It describes the spectral content of a signal – In other words it is the frequency domain representation of a signal. Lecture 2 Fourier Transform (2) • La fonction X(f) est la transformée de Fourier de x(t). • X(f) décrit le contenu spectral de x(t). ∞ • X(f) = F{x(t)} = ∫ x(t )e − j 2πft dt −∞ ∞ • x(t) = F-1{X(f)} = ∫ X ( f )e j 2πft df −∞ Lecture 2 Example • Find the Fourier Transform of x(t) = Π(t). • Solution • The Fourier Transform of x(t) is : ∞ 1/ 2 −∞ −1 / 2 X ( f ) = ∫ Π (t )e − j 2πft dt = 1 − j 2πft =− e j 2πf − j 2πft e dt ∫ 1/ 2 −1 / 2 1 =− (e − jπf − e jπf ) j 2πf e jπf − e − jπf sin πf = = = sinc( f ) j 2πf πf Lecture 2 Example 2 • Trouvez la transformée de Fourier de x(t) = δ(t). • Solution ∞ X(f ) = ∫ δ (t )e − j 2πft dt = e − j 2πft t =0 =1 −∞ Lecture 2 Fourier Transform Properties 1. Linearity • The Fourier Transform is a linear function. In other words, if X1(f) =F{x1(t)} et X2(f) = F{x2(t)}, then for x3(t) = ax1(t) + bx2(t), X3(f) = F{x3(t)}=aX1(f) + bX2(f). 2. Time Delay • If the Fourier Transform of x1(t) is X1(f) then the Fourier transform of x2(t) = x1(t-to) is X ( f ) = X ( f )e − j 2πfto 2 1 3. Time Scaling • If F{x(t)} = X(f), then F{x(at)} = (1/|a|)X(f/a). 4. Duality • If F{x(t)} = X(f), then F{X(t)} = x(-f). Lecture 2 Properties (2) 5. Frequency shift • If X(f) = F-{x(t)}, then X(f-fo) = F{x(t) e j 2πf o t } 6. Convolution • If z(t) = x(t)*y(t), then Z(f) = X(f)Y(f). 7. Multiplication • If z(t) = x(t)y(t), then Z(f) = X(f)*Y(f). 8. Derivative • F{ dx (t ) } = j2πfX(f) dt 9. Integration • t F x(λ )dλ = −∞ ∫ 1 j 2πf X ( f ) + 12 X (0)δ ( f ) Lecture 2 Transform Properties (3) 10.Complex conjugate • F{x*(t)} = X*(-f) Lecture 2