Digital Signal Processing Lecture 5

Transcription

1 Remote Sensing Laboratory Dept. of Information Engineering and Computer Science University of Trento Via Sommarive, 14, I Povo, Trento, Italy Digital Signal Processing Lecture 5 Begüm Demir demir@disi.unitn.it Web page:

2 Convolution

3 Convolution

4 Convolution

5 Convolution

6 Convolution

7 Convolution No Overlap for n < 0

8 Convolution Partial Overlap

9 Convolution Partial Overlap

10 Convolution Partial Overlap

11 Convolution Full Overlap

12 Convolution Full Overlap

13 Convolution Full Overlap

14 Convolution Partial Overlap

15 Convolution Partial Overlap

16 Convolution Partial Overlap

17 Convolution No Overlap for n > 0

18 Building Blocks for DT Systems Unit delay = Memory => store at one sampling interval and read at the next one

19 Examples Signal flow graphs provide compact representation

20 Interconnection of LTI Systems

21 System Realization For causal LTI systems, h[n] = 0 for n < 0. Finite impulse response (FIR): Infinite impulse response (IIR): M 1 yn [ ] hkxn [ ] [ k] k 0 yn [ ] hkxn [ ] [ k] k 0 The convolution summation formula expresses the output of the linear timeinvariant system explicitly and only in terms of the input signal. When n is increasing, memory requirements also increases with time. How would one realize these systems?

22 System Realization Moving Average if y[n] depends only on the present and past inputs, such a system is called nonrecursive.

23 System Realization Accumulator This is an example of a recursive system. In the recursive systems y[n] depends not only on the present and past inputs, but also available past output values.

24

25 Linear Constant-Coefficient Difference Equations (LCCDE) Discrete-time systems described by difference equations express the output of the system not only in terms of the present and past values of the input, but also in terms of the already available past output values: ayn [ ] ayn [ 1]... a yn [ N] bxn [ ] bxn [ 1]... b xn [ M] 0 1 N 0 1 M N k 0 a k y[ n k] M k 0 b k x[ n k] M N 1 yn [ ] bxn k [ k] ayn k [ k] 0 k 0 k 1 B. Demir

26 Linear Constant-Coefficient Difference Equations If the output signal does not depend on the past values of output (N=0), it is defined as: M yn [ ] bxn [ k] k 0 k M k k 0 hn [ ] b [ n k] bk, 0 n M 0, otherwise The length of impulse response is M+1. B. Demir

27 Example-1 y[ n] y[ n 1] x[ n] y[ 1] 0 n 0 y[0] y[ 1] x[0] 1 n 1 y[1] y[0] x[1] 3 n 2 y[2] y[1] x[2] 4 n 3 y[3] y[2] x[3] 4 n 4 x[ n] 0, y[ n] 4 n 0 x[ n] 0, y[ 1] 0, y[ n] 0 B. Demir

28 Example-2 B. Demir

29 Example-2-Cont B. Demir

30 Example-2-Cont B. Demir

31 Example-2-Cont B. Demir

32 Linear Constant-Coefficient Difference Equations Given LCCDE as the I/O relationship describing LTI system, the objective is to determine an explicit expression for the output y[n]. Basically, the goal is to determine y[n], n 0, of the system given a specific input x[n], n 0, and set of initial conditions. The direct solution method assumes that the total solution is the sum of two parts: B. Demir

33 Linear Constant-Coefficient Difference Equations Homogeneous Solution: N k 0 ayn [ k] 0 k It is assumed that the solution of this eq. is in the form of n yn [ ] and the eq. is described as polynomial eq. a... a 0 n n 1 n N 1 N N N 1 a1... an 0 characteristic poly. B. Demir

34 Linear Constant-Coefficient Difference Equations The polynomial has N roots (λ 1, λ 2,, λ N ). The roots can be real or complex valued. Complex-valued roots occur as complex conjugate pairs. Some of N roots may be identical. If the roots are distinct: y [ n] C C C n n n h N N where C 1, C 2,, C N are weighting coefficients. These coefficients are determined from the initial conditions. If λ 1 is a root of multiplicity m, then eq. becomes y [ n] C C n C n C n h n n 2 n 3 n Cn C C m 1 n n n m 1 m 1 m 1 N N B. Demir

35 Linear Constant-Coefficient Difference Equations Particular Solution: N a y [ n k] b x[ n k] M k p k k 0 k 0 To solve this eq., it is assumed for y p [n], a form that depends on the form of the input x[n]. If x[n] is given as an exponential, it is assumed that the particular solution is also exponential. If x[n] is sinusoidal, the particular solution is also sinusoidal. Thus, the assumed form for the particular solution takes the basic form of the signal x[n]. B. Demir

36 Example-1 B. Demir

37 Example-1 B. Demir

38 Example-1 Cont

39 Example-1 Cont

40 Linear Constant-Coefficient Difference Equations The total solution can be also defined as the sum of two parts: yn [ ] y [ n] y [ n] zi zs y zi [n]= zero input response y zs =zero state response N k 0 ay[ n k] 0 k zi N n zi[ ] j j j 1 y n C y [ n] y [n] y [n] zs h p Assume that all the initial conditions are zero B. Demir 40

41 Example yn [ ] 3 yn [ 1] 4 yn [ 2] 0 y[-1]=5 ve y[-2]=0 yzi n [ n] ( 1) (4) n 2 B. Demir

42 Linear Constant-Coefficient Difference Equations h[n] is the zero state response of LTI causal systems since h[n]=0 when n<0. B. Demir

43 Example B. Demir