Chap 2. Discrete-Time Signals and Systems

Size: px

Start display at page:

Download "Chap 2. Discrete-Time Signals and Systems"

Lisa Powers
5 years ago
Views:

1 Digital Signal Processing Chap 2. Discrete-Time Signals and Systems Chang-Su Kim

2 Discrete-Time Signals CT Signal DT Signal

3 Representation Functional representation 1, n 1,3 x[ n] 4, n 2 0, otherwise Tabular representation n x[n] Sequence representation x n = {, 0,0,1,4,1,0,0, }

4 Elementary Sequences Unit sample sequence (impulse function, delta function) 0, n 0 δ n = ቊ 1, n = 0 Unit step sequence 1, n 0 u n = ቊ 0, n < 0 Exponential sequence x n = Aα n Sinusoidal sequence x n = A cos(ω 0 n + φ)

5 Properties of Impulse and Step Functions 1) [ n] u[ n] u[ n 1] 2) u[ n] [ k] [ n k] k0 3) x[ n] [ n] x[0] [ n] n k 4) x[ n] [ n n ] x[ n ] [ n n ] ) x[ n] x[ k] [ n k] k

6 Properties of Exponential and Sinusoidal Sequences Exponential n x[ n] a n j0n re n r (cos n j sin n) 0 0

7 Properties of e jω 0n and cos(ω 0 n) ω 0 + 2π = ω 0 They are periodic only if ω 0 N = 2πk cf) Note the differences from the CT case

8 Discrete-Time Systems y n = T x n Examples 1. Ideal delay y n = x[n 2] 2. Moving average y n = 1 (x n 1 + x n + x[n + 1]) 3

9 Memoryless Systems Memoryless Systems: The output y[n] at any instance n depends only on the input value at the current time n, i.e. y[n] is a function of x[n] Systems with Memory: The output y[n] at an instance n depends on the input values at past and/or future time instances as well as the current time instance Examples: A resistor: y[n] = R x[n] A unit delay system: y[n] = x[n 1] An accumulator: n y n = x[k] k=

10 Proving or Disproving Mathematical Statement For a system to possess a given property, the property must hold for every possible input signal to the system A counter example is sufficient to prove that a system does not possess a property To prove that the system has the property, we must prove that the property holds for every possible input signal

11 Linear Systems A system is linear if it satisfies two properties. Additivity: x 1 n + x 2 n y 1 [n] + y 2 [n] Homogeneity: cx 1 n cy 1 [n] The two properties can be combined into a single property (linearity). a 1 x 1 n + a 2 x 2 n a 1 y 1 [n] + a 2 y 2 [n] Examples y n = x 2 n y n = log x n y n = 2x n + 3 y n = σ k= x[k] n

12 Time-Invariant Systems A system is time-invariant if a delay (or a time-shift) in the input signal causes the same amount of delay in the output. x n n 0 y[n n 0 ] Examples: y n = x 2 n y n = sin x n y n = x 2n y n = σ k= x[k] n

13 Causal Systems Causality: A system is causal if the output at any time instance depends only on the input values at the current and/or past time instances. Examples: y[n] = x[n] x[n 1] y n = x n + 1 A memoryless system is always causal.

14 Stable Systems Stability: A system is stable if a bounded input yields a bounded output (BIBO). In other words, if x[n] < k 1 then y[n] < k 2. Examples: y n = x 2 n y n = sin x n y n = x 2n y n = σ k= x[k] n

15 Linear Time-Invariant Systems and Their Properties

Divide and Conquer Divide an input signal into a sum of shifted scaled versions of an elementary signal If you know the system output in response to the elementary signal, you

16 Divide and Conquer Divide an input signal into a sum of shifted scaled versions of an elementary signal If you know the system output in response to the elementary signal, you also know the output in response to the input signal Ex) An LTI system processes x 1 [n] to make y 1 [n]. The same system processes another input x 2 [n] to make y 2 [n]. Plot y 2 [n].

17 Representing Signals in Terms of Impulses Sifting property 0 x[ n] x[ k] [ n k] k x[ 2] [ n 2] x[ 1] [ n1] x[0] [ n] x[1] [ n1] x[2] [ n2] xn [ ] x[ 2] [ n2] + x[ 1] [ n1] + x[0] [ n] + x[1] [ n1] + x[2] [ n 2]

18 Impulse Response The response of a system H to the unit impulse [n] is called the impulse response, which is denoted by h[n] h[n] = H{[n]} [n] System H h[n]

19 Convolution Sum Let h[n] be the impulse response of an LTI system. Given h[n], we can compute the response y[n] of the system to any input signal x[n]. [n] h[n]

21 Convolution Sum Let h[n] be the impulse response of an LTI system. Given h[n], we can compute the response y[n] of the system to any input signal x[n].

22 Convolution Sum Let h[n] be the impulse response of an LTI system. Given h[n], we can compute the response y[n] of the system to any input signal x[n]. x[ n] x[ k] [ n k] k y[ n] H[ x[ n]] H x[ k] [ n k] k k k k H x[ k] [ n k] x[ k] H [ n k] x[ k] h[ n k]

23 Convolution Sum Notation for convolution sum y[ n] x[ n]* h[ n] x[ k] h[ n k] k The characteristic of an LTI system is completely determined by its impulse response. [n] x[n] LTI system h[n] x[n] h[n]

24 Convolution Sum To compute the convolution sum y[ n] x[ n]* h[ n] x[ k] h[ n k] k Step 1 Plot x and h vs k since the convolution sum is on k. Step 2 Flip h[k] around the vertical axis to obtain h[ k]. Step 3 Shift h[ k] by n to obtain h[n k]. Step 4 Multiply to obtain x[k]h[n k]. Step 5 Sum on k to compute σ x[k]h[n k]. Step 6 Change n and repeat Steps 3-6.

25 Example Consider an LTI system that has an impulse response h[n] = u[n] What is the response when an input signal is given by where 0 < a < 1? x[n] = a n u[n] For n0, yn [ ] n k0 k 1 1 n1 Therefore, n1 1 y[ n] u[ n] 1

26 Example Consider an LTI system that has an impulse response h[n] = u[n] u[n N] What is the response when an input signal is given by where 0 < a < 1? x[n] = a n u[n]

27 Properties of Convolution Identity property x[n] δ[n] = x[n] Shifting property x[n] δ[n k] = x[n k]

28 Properties of Convolution Commutative property x[n] h[n] = h[n] x[n]

29 Properties of Convolution Associative property {x[n] h 1 [n]} h 2 [n] = x[n] {h 1 [n] h 2 [n]}

30 Properties of Convolution Distributive property x[n] [h 1 [n] + h 2 [n]] = x[n] h 1 [n] + x[n] h 2 [n]

31 Causality of LTI Systems A system is causal if its output depends only on the past and present values of the input signal. Consider the following for a causal LTI system: y[ n] x[ k] h[ n k] k Because of causality h[n-k] must be zero for k > n. In other words, h[n] = 0 for n < 0.

32 Causality of LTI Systems So the convolution sum for a causal LTI system becomes n y[ n] x[ k] h[ n k] h[ k] x[ n k] k k0 So, if a given system is causal, one can infer that its impulse response is zero for negative time values, and use the above simpler convolution formulas.

33 Stability of LTI Systems A system is stable if a bounded input yields a bounded output (BIBO). In other words, if x[n] < M x then y[n] < M y. Note that y[ n] x[ n k] h[ k] x[ n k] h[ k] M h[ k] x k k k Therefore, a system is stable if k hk [ ]

34 Examples System Impulse response Causal Stable y n = x[n n d ] y n = 1 M 1 + M n y n = k= M 2 k= M 1 x[k] x[n k] y n = x n + 1 x[n] y n = x n x[n 1]

35 Examples

36 Constant-Coefficient Difference Equations (CCDE)

37 Discrete-Time Systems Block diagram representation constant multiplier adder unit delay

38 Recursive Systems If an impulse response has a finite duration, the system is an FIR system. Otherwise, an IIR system. An FIR system can be implemented directly using a finite number of adders, multipliers and delays. e.g.) Implement the system with h[n]

39 Recursive Systems Can you directly implement the cumulative averaging system? n 1 y( n) x( k), n 0,1,... n 1 It can be implemented in a recursive manner with a feedback loop k 0 Past output values are used to compute a current output value

40 Constant-Coefficient Difference Equations (CCDE) N a y[ n k] b x[ n k] M k k0 k0 The equation defines a recursive system, which processes an input x[n] to make the output y[n] N is the order of the equation or the corresponding system k

41 Constant-Coefficient Difference Equations (CCDE) N a y[ n k] b x[ n k] M k k0 k0 k Example 1: Accumulator y n = σn k= x[k]

42 Constant-Coefficient Difference Equations (CCDE) N a y[ n k] b x[ n k] k k0 k0 Example 1: MA System y n = 1 σ M 2 M 2 +1 k=0 M k x[n k]

43 Constant-Coefficient Difference Equations (CCDE) N a y[ n k] b x[ n k] k k0 k0 Suppose that x[n] is given, and we want to get y[n] for n 0. Which information do we need further? M k

44 Constant-Coefficient Difference Initial rest condition Equations (CCDE) N a y[ n k] b x[ n k] k k0 k0 If x[n] starts at n = n 0, i.e., x[n] = 0 when n < n 0, then y[n] = 0 when n < n 0. Alternatively, the initial values are y[n 0 1] = y[n 0 2] = = y[n 0 N] = 0. M k If we assume the initial rest condition, then the system described by the equation is LTI.

45 Constant-Coefficient Difference Equations (CCDE) More details will be studied later, especially in Chap 6.

46 Frequency Domain Representation of Discrete- Time Signals and Systems

47 Eigenfunctions for LTI Systems δ[n] x[n] e jωn LTI h[n] y n = x n h n H(ej ω )e jωn e jωn is an eigenfunction of LTI systems Its eigenvalue is given by the Fourier transform of impulse response, H(e jω ), which is called frequency response H e jω = k= h k e jωk

48 Eigenfunctions for LTI Systems x n = α k e jω kn k LTI y n = α k H(e jω k)e jω kn k

49 Frequency Response Ex) Determine the output sequence of the system with impulse response h n = 1 u[n] 2n when the input is a complex exponential x[n] = Ae jπ 2 n

50 Frequency Response Ex) Determine the magnitude and phase of H(e jω ) for the three-point moving average (MA) system y[n] = 1 {x n x n + x n 1 }. 3

51 Sinusoidal Input Assuming that h[n] is real, we have the input-output relationship A cos(ω 0 n + φ) LTI A H e jω 0 cos(ω 0 n + φ + θ(ω 0 )) H e jω = H e jω ejθ ω 1. The amplitude is multiplied by H e jω 2. The output has a phase lag relative to the input by an amount θ ω = H(e jω )

52 Ideal Filters H e jω = H e j(ω+2πr)

53 Representation of Sequences by Fourier Transforms

54 Discrete-Time Fourier Transform 1 j jn x[ n] X ( e ) e d 2 2 j X ( e ) x[ n] e n jn DTFT can be derived from DTFS (discrete-time Fourier series) Frequency response is the DTFT of impulse response The existence of X e jω A sufficient condition: x n is absolutely summable We avoid rigorous conditions/proofs and use well-known Fourier transform pairs

55 Fourier Transform Pairs

56 Symmetry Property

57 Symmetry Property Figure 2.22 Frequency response for a system with impulse response h[n] = a n u[n]. a > 0; a = 0.75 (solid curve) and a = 0.5 (dashed curve). (a) Real part. (b) Imaginary part. (c) Magnitude. (d) Phase.

58 Fourier Transform Theorems

59 Convolution Theorem y n = x n h n Y e jω = X e jω H e jω

60 Convolution Theorem: Another Perspective x n = α k e jω kn k LTI y n = α k H(e jω k)e jω kn k x n = 1 2π න X(e jω )e jωn dω 2π y n = 1 2π න X(e jω )H(e jω )e jωn dω 2π

61 Examples x n = a n u n 5. What is X(e jω )?

62 Examples X e jω = 1 (1 ae jω )(1 ae jω ). What is x n?

63 Examples X e jω = 1 (1 ae jω )(1 be jω ). What is x n?

64 Examples Determine the impulse response h[n] of a highpass filter with frequency response H e jω = e jωn d, ω c < ω < π, 0, ω < ω c.

65 Examples Determine the frequency response and the impulse response of a system described by a CCDE y n 1 2 y n 1 = x n 1 x n 1. 4

Digital Signal Processing Lecture 3 - Discrete-Time Systems

Digital Signal Processing Lecture 3 - Discrete-Time Systems Digital Signal Processing - Discrete-Time Systems Electrical Engineering and Computer Science University of Tennessee, Knoxville August 25, 2015 Overview 1 2 3 4 5 6 7 8 Introduction Three components of