Signals and Systems Electronics and Telecommunications Faculty Communications Department Instructor: Lecturer Dr. Eng. Corina Nafornita 1 COURSE OBJECTIVES This course is frequently found in electrical engineering curricula, the concepts and techniques that form the core of the subject are of fundamental importance in all engineering disciplines. Our approach has been guided by the continuing developments in technologies for signal and system design and implementation, which made it increasingly important for a student to have equal familiarity with techniques suitable for analyzing and synthesizing both continuous-time and discretetime systems. 2 1 COURSE TOPICS Signals and systems: Continuous-Time and Discrete-Time Signals; Exponential and Sinusoidal Signals; Continuous-Time and Discrete-Time Systems; Basic System Properties. Linear time-invariant systems: Discrete-Time LTI Systems: The Convolution Sum; Continuous-Time LTI Systems: The Convolution Integral; Properties of Linear Time-Invariant Systems; Singularity Functions. Fourier Series Representation: The Response of LTI Systems to Complex Exponentials; Fourier Series Representation of Continuous-Time and Discrete-Time Periodic Signals. The Continuous-Time Fourier Transform: Representation of Aperiodic Signals: The Continuous-Time Fourier Transform; Properties of the Continuous-Time Fourier Transform, Systems Characterized by Linear ConstantCoefficient Differential Equations. The Discrete-Time Fourier Transform: Representation of Discrete-Time Aperiodic Signals: The Discrete-Time Fourier Transform; Properties of the Discrete-Time Fourier Transform; Duality; Systems Characterized by Linear Constant-Coefficient Difference Equations 3 TEXTBOOKS/REFERENCES 1. Corina Nafornita, Signals and Systems, vol. 1, Politehnica Publishing House, 2009, ISBN 978-606-554-013-2 (978-606-554-014-9 vol I), Table of contents http://shannon.etc.upt.ro/teaching/ss-pi/Signals_Systems_TOC.pdf 2. Alan V. Oppenheim, Alan S. Willsky with S. Hamid Nawab, Signals & Systems, Second Edition, Prentice Hall, Upper Saddle River, New Jersey, 1997. 3. Simon Haykin, Barry Van Veen, Signals and Systems, 2nd edition, John Wiley & Sons, 2003 WEBPAGE http://shannon.etc.upt.ro/teaching/ss-pi/ CONTACT corina.nafornita [at] gmail [dot] com 4 2 Signals Signal - A time-variable phenomenon that carries an information. Signal types: Biological, acoustical, chemical, optical, electronic,… 5 a) b) An electrocardiogram. A voice signal. 6 3 Images. 7 Mathematical model function having as independent variable the time x ( t ) = 10 ⋅ sin 2π ⋅103 ⋅ t [V] 8 4 Discrete-time signals Sampling x(t) with step Ts=0,05 ms x̂ ( t ) = x ( nTs ) = 10 ⋅ sin 2 ⋅ π ⋅103 ⋅ 0 ,05 ⋅10−3 ⋅ n = = 10 ⋅ sin 0 ,1 ⋅ π ⋅ n [V] n∈] n=t/Ts – discrete time x [ n ] = x ( nTs ) n ∈ ] 9 Sampling. 10 5 Some important signals used in electrical engineering i) Sinusoidal signal x ( t ) = Acos ( ω0t + ϕ ) ; A, ω0 = 2πf 0 , T0 , ϕ The sinusoidal signal is periodic x ( t + T0 ) = x ( t ) , x ( t + nT0 ) = x ( t ) , ∀t and ∀n ∈ ] Acos ⎡⎣ω0 ( t + T0 ) + ϕ⎤⎦ = Acos ( ω0t + ϕ ) , ∀t cos [ ω0t + ϕ + ω0T0 ] = cos ( ω0t + ϕ ) , ∀t ω0T0 = 2π; T0 = 1 2π = f0 ω0 11 ii) Sinusoidal discrete-time signal x [ n] = A cos ( ω0Ts n + ϕ ) [ω0Ts ] = [ω0 ][Ts ] = rad ⋅ s = rad s f Ω0 = ω0Ts = 2π 0 - discrete frequency fs x [ n] = A cos ( Ω0 n + ϕ ) cos ⎡⎣( Ω0 + 2π ) n + ϕ⎤⎦ = cos ( Ω0 n + ϕ ) 12 6 Discrete frequency for x [ n] = cos Ω0n 13 “Confusion” due to sampling Ω0 = 0; xk ( t ) = Acos k 2π t; k = 0 ,1,... Ts 14 7 Ω0 = π; xk ( t ) = A cos ( 2k + 1) π t ; k = 0,1,... Ts 15 Periodicity of the d.t. sine wave, period N Acos ⎡⎣Ω0 ( n + N ) + ϕ⎤⎦ = Acos ( Ω0 n + ϕ ) , N =k Example ∀n, Ω0 N = k 2π 2π π ∈` ⇒ ∈ _ (rational number) Ω0 Ω0 Ω0 = 4π 7 2⋅7 7 π ⇒ = ⇒ N =k⋅ =k⋅ 7 Ω0 4 4 2 minimum k for which N is integer : k=2 ⇒ N=7 ⎛ 4π ⎞ x [ n ] = Acos ⎜ n + ϕ ⎟ ⎝ 7 ⎠ The signal x [ n] = Acos ( 2n + ϕ ) is not periodic. 16 8 iii) Continuous-time unit step signal ⎧1, t ≥ 0 σ (t ) = ⎨ ⎩0 , t < 0 This is only a model. It can not be generated in practice. 17 iv)Discrete-time unit step signal ⎧1, n ≥ 0 σ [ n ] = σ ( nTs ) = ⎨ ⎩0 , n < 0 18 9 v)Continuous-time unit impulse. Dirac impulse Δk → 0 ∞ ∫ f k ( t ) dt = 1 −∞ ⎧∞ , t = 0 lim f k ( t ) = ⎨ k →∞ ⎩ 0, t ≠ 0 Δ k →0 ⎧∞ , t = 0 δ (t ) = ⎨ ⎩ 0, t ≠ 0 ∞ ∫ δ ( t ) dt = 1 −∞ 19 A remarkable property ϕ ( t ) fk ( t ) ≅ ϕ ( 0 ) fk ( t ) lim ϕ ( t ) f k ( t ) = ϕ ( 0 ) lim f k ( t ) Δ k →0 Δ k →0 ϕ ( t ) δ ( t ) = ϕ ( 0) δ (t ) ∞ ∞ ∫ ϕ ( t ) δ ( t ) dt = ∫ ϕ ( 0 ) δ ( t ) dt = −∞ −∞ ∞ = ϕ ( 0 ) ∫ δ ( t ) dt =ϕ ( 0 ) ⋅1 = ϕ ( 0 ) −∞ The filtering property of the Dirac impulse ∞ ∫ ϕ ( t ) δ ( t ) dt =ϕ ( 0 ) −∞ 20 10 Unit impulse and unit step connection lim g k ( t ) = σ ( t ) Δ k →0 g'k ( t ) = f k ( t ) lim g'k ( t ) = lim f k ( t ) = δ ( t ) Δ k →0 Δ k →0 ' ⎛ ⎞ ⎜ lim g k ( t ) ⎟ = δ ( t ) ⎝ Δ k →0 ⎠ σ' ( t ) = δ ( t ) 21 σ' ( t ) = δ ( t ) t ⎧1, t > 0 ∫ δ ( τ )d τ = ⎨0, t < 0 ⎩ −∞ t ∫ δ ( τ )d τ = σ ( t ) −∞ 22 11 vi) Discrete-time unit impulse ⎧1, n = 0 δ [ n] = ⎨ ⎩0 , n ≠ 0 23 Discrete-time unit impulse and unit step connection n ∑ δ [ k ] = σ [ n] k =−∞ σ [ n ] − σ [ n − 1] = δ [ n ] n −1 n n −1 ∑ δ [ k ] = σ [ n − 1] ; ∑ δ [ k ] − ∑ δ [ k ] = σ [ n] - σ [ n − 1] k =−∞ n −1 k =−∞ n −1 k =−∞ ∑ δ [ k ] + δ [ n] − ∑ δ [ k ] = σ [ n ] - σ [ n − 1] k =−∞ k =−∞ 24 12 Other properties of the discretetime unit impulse x [ n ] δ [ n ] = x [ 0] δ [ n ] ∞ ∑ x [ k ] δ [ n − k ] = ... + x [ −2] δ [ n + 2] + x [ −1] δ [ n + 1] + x [ 0] δ [ n ] + k =−∞ + x [1] δ [ n − 1] + ... + x [ n − 1] δ ⎡⎣ n − ( n − 1) ⎤⎦ + x [ n ] δ [ n − n ] + x [ n + 1] δ ⎡⎣ n − ( n + 1) ⎤⎦ + ... ∞ x [ n] = ∑ x [ k ] δ [ n − k ] k =−∞ 25 26 13 vii) Continuous-time ramp signal ⎧t ⎪ ∫ d τ = t, t ≥ 0 r ( t ) = ∫ σ ( τ )d τ = ⎨0 ⎪ −∞ t<0 ⎩ 0, t ⎧ t, t ≥ 0 = tσ ( t ) r (t ) = ⎨ ⎩0 , t < 0 27 viii) Discrete-time ramp signal ⎧ n −1 ⎪ ∑ 1 = n, n ≥ 1 r [ n] = ∑ σ [ k ] = ⎨k = 0 k =−∞ ⎪ 0, n <1 ⎩ n −1 ⎧n, n ≥ 0 r [ n] = ⎨ = nσ [ n ] ⎩0, n < 0 28 14 ix) Continuous-time exponential signal x ( t ) = eat , a ∈ \, e ~ 2.7182 a > 0 ; lim eat = 0 ; lim eat = ∞ ; e0 = 1 t →−∞ a<0; lim e t →−∞ t →∞ at = ∞ ; lim eat = 0 ; e0 = 1 t →∞ 29 Causal exponential ⎧⎪eat , t ≥ 0 x ( t ) = eat σ ( t ) = ⎨ ;a<0 ⎪⎩ 0 , t < 0 30 15 x)Discrete-time exponential signal ( ) x [ n ] = ebnTs = ebTs n ; ebTs = a ⇒ x [ n ] = a n , a ∈ \ 0<a<1 a>1 Homework: sketch the signal for a<-1 -1<a<0 31 Discrete-time causal exponential ⎧⎪a n , n ≥ 0 x [ n] = a nσ [ n] = ⎨ ⎪⎩ 0 , n < 0 32 16 xi) Oscillation with exponential envelope in continuous-time x ( t ) = eat sin ω0t sin ω0tk = 1; tk = π 2π +k ; 2ω0 ω0 sin ω0tl = −1; tl = − π 2π ; +k 2ω0 ω0 x ( tk ) = e atk x ( tl ) = −e atl 33 Causal case ⎧⎪eat sin ω t, t ≥ 0 0 x ( t ) = eat sin ω0t σ ( t ) = ⎨ t<0 0, ⎪⎩ 34 17 xii) Oscillation with exponential envelope in discrete-time x [ n ] = a n cos Ω0 n Exercise: Draw the waveform of this signal for the case a>1. 35 1.3. Complex signals. Phasors. e jθ = cos θ + j sin θ ; e− jθ = cos θ − j sin θ e jθ + e − jθ cos θ = 2 ; e jθ − e− jθ sin θ = 2j cos θ = Re e jθ ; sin θ = Im e jθ { } { } 36 18 i) Relation between real sinusoidal signal and complex exponential { } j ω t +ϕ Re Ae ( 0 ) = A cos ( ω0t + ϕ ) ; {( ) } Re Ae jϕ e jω0t = A cos ( ω0t + ϕ ) ; A ∈ \+ A ∈ \+ e jω0t - oscillatory part Ae jϕ - complex amplitude A = Ae jϕ ∈ ^ { jω0t A cos ( ω0t + ϕ ) = Re Ae } jω0t - phasor that rotates with the angular velocity ω Ae 0 37 ϕ=0 . jω0t for φ=0. The mobile extremity Evolution in time of the phasor Ae of the phasor describes a cylindrical helix of radius A. 38 19 The negative frequency 39 ii) Relation between real oscillation with exponential envelope and complex exponential x ( t ) = Aeσ0t cos ( ω0t + ϕ ) , A ( t ) = Ae jϕeσ0t { } { σ0 ∈ \ } Re A ( t ) e jω0t = Re Aeσ0t e jϕe jω0t = Aeσ0t cos ( ω0t + ϕ ) A ( t ) complex envelope of the signal x ( t ) The vector that rotates with the angular velocity ω0 describes a spiral. 40 20 iii) Sampling (discrete-time case) x [ n ] = Aeσ0Ts n cos ( ω0Ts n + ϕ ) = Aa n cos ( Ω0 n + ϕ ) j Ω n +ϕ ) Aa n e ( 0 associated phasor; A [ n ] = Aa n e jϕ complex envelope { } { j Ω n +ϕ ) x [ n ] = Re Aa n e ( 0 = Re A [ n ] e jΩ0 n } σ0 = 0 : complex envelope constant A [ n ] = Ae jϕ the vector rotating with angular velocity Ω0 - constant magnitude ϕ = 0 : A cos Ω0 n = A j Ω0 n A − j Ω 0 n e + e → "negative frequency" 2 2 41 1.4. Simple signal transformations i)Weighting- amplification or attenuation of signal 42 21 ii) Time shifting x ( t − t0 ) shifted to the right if t0 > 0 x [ n − n0 ] shifted to the right if n0 > 0 to the left if t0 < 0 to the left if n0 < 0 43 x ( t ) = x ( −t ) iii) Time reversal x [ n] = x [ −n] 44 22 iv) Time scaling compresses or dilates the signal by multiplying the time variable by a constant y ( t ) = x ( at ) , a ∈ \ 45 v) Discrete-time scaling ⎧ ⎡n⎤ , if k divides n ⎪x x( k ) [ n ] = ⎨ ⎢⎣ k ⎥⎦ ⎪ 0, otherwise ⎩ 46 23 vi) Simple transformations x ( t ) ⎯⎯ → 2 x ( −2t − 2 ) 47 Even and odd parts of a real signal x ( t ) = xe ( t ) + xo ( t ) ; xe ( t ) = xe ( −t ) = xe ( t ) x [ n ] = xe [ n ] + xo [ n ] ; x ( t ) + x ( −t ) x ( t ) − x ( −t ) ; xo ( t ) = 2 2 xo ( −t ) = − xo ( t ) ; xe [ n ] = x [ n] + x [ −n ] 2 ; xo [ n ] = x [ n ] − x [ −n] 2 48 24 Energy and Power ∞ Energy W = ∫ x (t ) 2 dt < ∞ for complex signal −∞ ∞ Discrete-time: W = ∑ x [n] 2 <∞ n =−∞ Continuous-time finite energy signals - square integrable functions (space L2 ) ∞ ∫ x (t ) −∞ 2 dt < ∞ ⇒ x ( t ) ∈ L2 Discrete-time finite energy signals - square summable functions (space l 2 ) ∞ ∑ n =−∞ x[n] < ∞ ⇒ x [ n ] ∈ l 2 2 49 1 – Causal decreasing exponential x ( t ) = e −t σ ( t ) ∞ W = ∫e 0 −2t −2t ∞ e dt = −2 0 1 − e−∞ 1 = = 2 2 50 25 2 – Causal oscillation with exponential envelope x ( t ) = e−t sin ω0tσ ( t ) ∞ ∞ 0 0 W = ∫ e−2t sin 2 ω0tdt = ∫ e−2t W= ω02 1 1 − = 4 4 1 + ω2 4 1 + ω02 0 ( ) ( 1 − cos 2ω0t 1∞ 1∞ dt = ∫ e−2t dt − ∫ e−2t cos 2ω0tdt 2 20 20 ) “fast” oscillations (ω0 large), approximation 1 W≅ 4 half of the energy for the case when there are no oscillations 51 3- Discrete-time causal exponential signal x [ n] = a n σ [ n] , a < 1 ∞ W =∑ a n =0 2n = 1 1− a 2 1 + a + a 2 + ... + a n + ... = 1 , 1− a a <1 52 26 4- Sine wave Periodic signal x ( t ) = A sin ω0t average energy computed over one period T0 T0 WT0 = ∫ A2 sin 2 ω0tdt = 0 A2 2 T0 ∫ (1 − cos 2ω t ) dt = 0 0 A2 T0 2 53 5 – Unit step signal x [ n] = σ [ n] ∞ N W = ∑ 12 = lim ∑ 1 = lim ( N + 1) = ∞ n=0 N →∞ n=0 N →∞ 54 27 Power Average power of the signal P : average flux of energy, ratio of signal energy and time interval when that energy was developed. infinite duration signals: 1 τ 2 x ( t ) dt ∫ τ→∞ 2τ −τ P = lim N 1 2 ∑ x [ n] N →∞ 2 N + 1 n =− N P = lim 55 Energy and average power, finite duration signals Discrete-time signals Continuous-time signals, support [t1,t2] t2 2 support {N1, N1+1,…,N2} N2 W = ∫ x ( t ) dt W = ∑ x [ n] 2 n = N1 t1 t 1 2 W 2 P= = x ( t ) dt ∫ t2 − t1 t2 − t1 t 1 P= N2 1 W 2 = ∑ x [ n] N 2 − N1 + 1 N 2 − N1 + 1 n = N1 56 28 Periodic signals average power is computed over one period: P= 1 2 x t dt ( ) ∫ T0 T0 P= 1 2 ∑ x [ n] N n∈ N 57 6-Sine wave, average power Sinusoidal Signal 1 τ 2 2 A2 τ 1 − cos 2ω0t P = lim dt = ∫ A sin ω0tdt = lim 2τ ∫ 2 τ→∞ 2τ −τ τ→∞ −τ τ ⎤ ⎡ 2 2 2 2 2 2 sin ω t A A 0 ⎥ = A − A lim sin 2ω0 τ = A = lim ⎢ ⋅ 2τ − ⋅ ⎥ 2 4τ 2ω0 4 τ→∞ 2ω0 τ 2 τ→∞ ⎢ 4τ −τ ⎥⎦ ⎢⎣ 58 29 1.7. Distributions function distribution operator 59 Example of Distribution: The Dirac Impulse δ (t ) : ϕ (t ) → ϕ ( 0) δ ( t − t0 ) : ϕ ( t ) → ϕ ( t0 ) δ(t) associates to any test function ϕ(t), its value from origin, ϕ(0) δ(t-t0) associates to any test function ϕ(t), its value from t0, ϕ(t0) f – distribution. The test function φ and a number (scalar product between f and φ) are associated ∞ ϕ ( t ) ⎯⎯ → ∫ ϕ ( t ) f ( t ) dt f −∞ f : ϕ ( t ) ⎯⎯ → f (t ) , ϕ (t ) = ∞ ∫ ϕ ( t ) f ( t ) dt 60 −∞ 30 The Derivative of a Distribution ∞ f ' ( t ) , ϕ ( t ) = − ∫ ϕ' ( t ) f ( t ) dt = f ( t ) , −ϕ' ( t ) −∞ δ' ( t ) , ϕ ( t ) = δ ( t ) , −ϕ' ( t ) = −ϕ' ( 0 ) δ' ϕ ( t ) →− ϕ' ( 0 ) ϕ (t ) the negative value of its first derivative computed in zero δ' ( t ) associates to the test function 61 Unit Step Distribution σ( t ) ∞ ϕ ( t ) ⎯⎯⎯ → ∫ ϕ ( t ) dt 0 σ ′( t ) ∞ ∞ ϕ ( t ) ⎯⎯⎯→ σ ( t ) , −ϕ′ ( t ) = − ∫ ϕ′ ( t ) dt = −ϕ ( t ) = ϕ ( 0 ) − ϕ ( ∞ ) 0 0 σ ′( t ) ϕ ( t ) ⎯⎯⎯→ ϕ ( 0 ) ⇒ σ′ ( t ) = δ ( t ) i) Functions are useful for modeling signals, ii) Distributions are useful for modeling some signals and processes like sampling, iii) Operators are useful for modeling signal processing systems. 62 31 Systems Their mathematical model is the operator. d dt ∫ : x ( t ) → x' ( t ) t : x ( t ) → ∫ x ( τ )d τ -∞ 63 2.1. Systems Au > 10000 u2 = Au ; VS u u VS = 2 < 2 Au 10000 u2 < 5V ; VS < 500 μV ; VS = 0 V ii = S Rin ; Rin = 100 k Ω ; ii < 500 μV = 5 nA 100 kΩ 64 32 Digital system moving average filter (running averager). 65 Simulated analog system x ( nTs ) ≅ qx [ n] converter on 10 bits (1024 quantization levels) domain of the input voltage 10V 10V ≅ 10mV 1024 Max quantization error ± 5mV q≅ 66 33 Mathematical model y ( t ) = S { x ( t )} S or x ( t ) ⎯⎯ → y (t ) y [ n ] = S d { x [ n ]} d or x [ n ] ⎯⎯→ y [ n] S 67 2.2. Linear systems S {a1x1 ( t ) + a2 x2 ( t )} = a1S { x1 ( t )} + a2 S { x2 ( t )} Sd {a1x1 [ n ] + a2 x2 [ n ]} = a1Sd { x1 [ n ]} + a2 Sd { x2 [ n ]} Superposition principle 68 34 Homogeneity S d {ax [ n ]} = aS d { x [ n ]} S {ax ( t )} = aS { x ( t )} 69 Incremental linear systems Homogeneity, a = 0: S {0 x ( t )} = 0 S { x ( t )} = 0 Systems with increments of the output, proportional with the increments of the input, not homogeneous ⇒ linear system; at the output the zero-input response y0 is added. 70 35 Additivity The linear system response at the sum of two input signals equals the sum of responses at each signal. x1 ( t ) + x2 ( t ) ⎯⎯ → y1 ( t ) + y2 ( t ) 71 2.3. Time invariant systems S { x ( t )} = y ( t ) ⇒ S { x ( t − t0 )} = y ( t − t0 ) S d { x [ n ]} = y [ n ] ⇒ S d { x [ n − n0 ]} = y [ n − n0 ] 72 36 Stability a) Stable equilibrium: the impulse applied to the ball creates attenuated oscillations of its position. b) Neutrally stable equilibrium: the impulse applied to the ball modifies the equilibrium position. c) Unstable equilibrium: the impulse applied to the ball produces loss of equilibrium 73 Causality •Between the output and input of the system : relation of the type “cause-effect” •The effect does not appear before the cause. 74 37 75 2.6. Systems described by linear constant-coefficients differential equations and difference equations First order linear system. Second order linear system. Homework: Prove the linearity of these systems. 76 38 General form of the linear constantcoefficients differential equation that describes an Nth order system d k y (t ) M d k x (t ) = , aN ≠ 0 (at least) b ∑ k dt k dt k k =0 k =0 The initial conditions should be null if the system is linear: N ∑ ak y ( t0 ) = dy ( t ) d 2 y (t ) d N −1 y ( t ) = = ... = =0 dt t =t dt 2 t =t dt N −1 t =t 0 0 0 if the input signal is applied at the moment of time t0 x ( t ) ≡ 0 for t < t0 77 Digital case - first order system dy ( t ) + y ( t ) = x ( t ) - equivalent digital system? dt dy ( t ) + y ( nTs ) = x ( nTs ) RC dt t =nT RC s dy ( t ) y ( nTs ) − y ( nTs − Ts ) y [ n ] − y [ n − 1] ≅ = dt t =nT Ts Ts s ⎛ RC ⎞ RC + 1⎟ y [ n ] − y [ n − 1] = x [ n ] ⎜ T T s ⎝ s ⎠ linear constant-coefficients difference equation, obtained by the approximation of the differential equation 78 39 • the slope of the secant line is a good approximation for the slope of the tangent line for a small sampling step - Ts 79 Digital case - second order system LC d 2 y (t ) dt d 2 y (t ) dt = 2 t = nTs + RC 2 t = nTs d ⎛ dy ( t ) ⎞ ⎜ ⎟ dt ⎝ dt ⎠ t = nTs dy ( t ) + y ( nTs ) = x ( nTs ) dt t = nT s dy ( t ) dy ( t ) − dt t = nT dt t = nT −T s s s ≅ Ts y [ n ] − y [ n − 1] y [ n − 1] − y [ n − 2] − y [ n ] − 2 y [ n − 1] + y [ n − 2] Ts Ts = = Ts Ts2 ⎛ LC RC ⎞ ⎛ 2 LC RC ⎞ LC + 1⎟ y [ n ] − ⎜ + ⎜ 2 + ⎟ y [ n − 1] + 2 y [ n − 2] = x [ n ] 2 ⎜T ⎟ ⎜ T Ts Ts ⎟⎠ Ts ⎝ s ⎠ ⎝ s linear constant-coefficients difference equation, obtained by the approximation of the differential equation 80 40 General form of the linear constantcoefficients difference equation that describes an Nth order system N M ∑ A y [n − k ] = ∑ B x [n − k ], k k =0 k AN ≠ 0 (at least) k =0 81 2.7. Some examples of systems i) Proportional ideal system y ( t ) = ax ( t ) , a ∈ \ y [ n] = ax [ n] , a ∈ \ memoryless system: the output signal at each time depends only on the input signal at the same value of the time, and it doesn’t depend on previous values. 82 41 ii) Ideal differentiator system y (t ) = dx ( t ) dt y [ n] = 1 ( x [ n] − x [ n − 1]) Ts system that implements the approximation of the derivative for the digital case83 iii) Ideal integrator system t y ( t ) = ∫ x ( τ )d τ −∞ n n −1 k =−∞ k =−∞ y [ n] = ∑ x [ k ] = ∑ x [ k ] + x [ n] y [ n ] = y [ n − 1] + x [ n ] Continuous time systems with memory : Discrete time: adder • digital adder • digital differentiator 84 42 2.8. Examples 1.Linear analog system with time-variable parameters d 2 y (t ) dy ( t ) 2 + 2t + t y (t ) = x (t ) 2 dt dt a) Linearity Additivity → y1 ( t ) ⇒ x1 ( t ) ⎯⎯ d 2 y1 ( t ) dt x2 ( t ) ⎯⎯ → y2 ( t ) ⇒ d 2 2 2 + 2t d 2 y2 ( t ) dt 2 dy1 ( t ) 2 + t y1 ( t ) = x1 ( t ) dt + 2t dy2 ( t ) 2 + t y2 ( t ) = x2 ( t ) dt d 2 ⎣⎡ y1 ( t ) + y2 ( t ) ⎦⎤ + 2t dt ⎣⎡ y1 ( t ) + y2 ( t ) ⎦⎤ + t ⎣⎡ y1 ( t ) + y2 ( t ) ⎦⎤ = x1 ( t ) + x2 ( t ) dt ⇒ x1 ( t ) + x2 ( t ) ⎯⎯ → y1 ( t ) + y2 ( t ) 85 Homogeneity x ( t ) ⎯⎯ → y (t ) ⇒ d 2 y (t ) dt + 2t 2 dy ( t ) 2 + t y (t ) = x (t ) dt d2 d ⎡⎣ ay ( t ) ⎤⎦ + 2t ⎡⎣ ay ( t ) ⎤⎦ + t 2 ⎡⎣ ay ( t ) ⎤⎦ = ax ( t ) ⇒ ax ( t ) ⎯⎯ → ay ( t ) dt dt 2 b) Time shift invariance x ( t ) ⎯⎯ → y (t ) x ( t − t0 ) ⎯⎯ → y3 ( t ) d 2 y (t ) dt 2 + 2t d 2 y3 ( t ) time invariant system: dt 2 dy ( t ) dt + t 2 y (t ) = x (t ) + 2 ( t − t0 ) d 2 y ( t − t0 ) 2 dt + 2t dy3 ( t ) dt dy ( t − t0 ) dt 2 + ( t − t0 ) y3 ( t ) = x ( t − t0 ) + t 2 y ( t − t0 ) = x ( t − t0 ) y3 ( t ) ≠ y ( t − t0 ) ⇒ linear system, but not time-invariant 86 43 ii) The influence of null initial conditions on linearity of analog systems dy ( t ) + 2 y (t ) = x (t ) dt ⎧ K cos ω0t, t ≥ 0 x ( t ) = K cos ω0t σ ( t ) = ⎨ 0, t<0 ⎩ 87 Particular solution - steady state dy f ( t ) dt + 2 y f ( t ) = K cos ω0t , t ≥ 0 y f (t ) = K 4 + ω02 cos ( ω0t − θ ) , t ≥ 0 Homogeneous solution – transient state dytr ( t ) + 2 ytr ( t ) = 0 t ∈ \ dt ytr ( t ) = Be −2t , t ≥ 0 and ytr ( t ) = Ce−2t , t < 0 88 44 Final Solution ⎧⎪ y ( t ) + y f ( t ) , t ≥ 0 y ( t ) = ⎨ tr t<0 ⎪⎩ ytr ( t ) , K ⎧ −2 t ⎡ cos ( ω0t − θ ) − e −2t cos θ ⎤⎦ , t ≥ 0 ⎪ y0 e + 2 ⎣ y (t ) = ⎨ 4 + ω0 ⎪ y e −2 t , t<0 ⎩ 0 y ( t ) = y0 e −2t + K 2 0 4+ω ⎡⎣cos ( ω0t − θ ) − e −2t cos θ ⎤⎦ σ ( t ) t ∈ \ linear system K = 0 : x ( t ) = 0 ⎯⎯⎯⎯⎯ → y ( t ) K =0 = y0e−2t ≡ 0 ⇒ y0 = 0 null initial value for linear system! 89 iii) Influence of the null initial conditions on the linearity of digital systems ⎧ K cos Ω0 n , n ≥ 0 y [ n ] − 0 ,5 y [ n − 1] = x [ n ] ⇒ x [ n ] = K cos Ω0 nσ [ n ] = ⎨ 0, n<0 ⎩ Particular solution - steady state { } y f [ n ] − 0.5 y f [ n − 1] = K cos Ω0 n = Re Ke jΩ0 n , n ≥ 0 y f [ n ] = A cos ( Ω0 n − θ ) A= 0.5sin Ω0 K e − jθ ; θ = arctg 1 − 0.5cos Ω0 1.25 − cos Ω0 Homogeneous solution – transient state ytr [ n ] − 0,5 ytr [ n − 1] = 0 , n ∈ ` ytr [ n ] = B ( 0,5 ) n , n ≥ 0 and ytr [ n ] = C ( 0,5 ) n , n<0 90 45 Final solution K ⎧ n ⎪ B ( 0.5 ) + 1.25 − cos Ω cos ( Ω0 n − θ ) , n ≥ 0 0 y [ n] = ⎨ ⎪ n n<0 ⎩C ( 0.5) , linear system ⇔ null initial condition: y [ −1] = 0 y [ n] = K cos ( Ω0 n − θ ) σ [ n ] 1.25 − cos Ω0 N th order system, null initial condition is y [ −1] = y [ −2] = ... = y [ − N ] = 0 91 46