Signal Processing and Linear Systems-B.P.Lathi copy

13 we c an d etermine a a nd a2 b y clearing

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: = z5(z - 1) 6 Therefore 00 =L z f [rJz-(r-m) F[z] = (z _ 1)2 - z5(z - 1)2 - z5(z - 1) r =m 6z + 5 z5(z - 1)2 Z6 - - • r=m E xercise E l1.5 U sing o nly t he f act t hat u[kJ ¢ =? z :'1 a nd t he r ight-shift p roperty [Eq. (11.15)], find t he z -transforms o f t he s ignals i n F igs. 1 l.2 a nd 11.3. T he a nswers a re g iven i n E xample 11.2d a nd £::" m -l 00 = zm [ ~f[rJz-r - ~ f [rJz-r E xercise E m -l = zmF[zJ - zm L f [rJz-r r=O 'V C onvolution T he t ime c onvolution p roperty a nd t he f requency convolution p roperty s tate t hat i f h [kJ Fl[ZJ a nd h [kJ F2[ZJ, = f lkl = t hen ( time c onvolution) 5 (11.18) a nd ( frequency convolution) (11.19) o 2 F ig. 1 1.4 3 4 5 Signal for Example 11.4. P roof: T hese p roperties a pply t o c ausal as well a s n oncausal s equences. F or t his r eason, we s hall p rove t hem for t he m ore g eneral c ase o f n oncausal s equences, w here t he c onvolution s um r anges f rom - 00 t o 0 0. T o p rove t he t ime c onvolution, we have • E xample 1 1.4 F ind the z-transform of the signal f[kJ depicted in Fig. 11.4. T he signal f[kJ can be expressed a s a p roduct of k a nd a gate pulse u[kJ - u[k - 6J. T herefore f[kJ = k {u[kJ - u[k - 6]} = ku[kJ - ku[k - 6J 00 = L k =-oo 00 z -k L m =-oo h [mJh[k - mJ 684 11 Discrete-Time Systems Analysis Using t he Z - Transform Interchanging the order of summation, I::. 00 Z [h[k] * h[kJ] = 11.3 Z -Transform Solution of Linear Difference Equations 685 E xercise E ll.6 U sing E q. ( 11.21), derive P airs 7 a nd 8 i n T able 11.1 from P airs 2 a nd 3, respectively. 00 h Im] L h [k - m]z-k m =-oo k=-oo \l L 00 L = Multiplication by k (Scaling in t he z-Domain 00 hIm] L m =-oo If h[r]z-(r+m) J[k]u[k] r =-oo 00 m =-oo ¢ => F[z] t hen 00 r =-oo kJ[k]u[k] d ¢ => - z-F[z] dz (11.22) Proof: T o prove t he frequency convolution, we s tart w ith -z~F[z] = -z~ f f[k]z-k = - z dz 00 Z {h[k]h[k]} = L h [k]h[k]z-k k=-oo dz k=O f - kf[k]z-k-l k=O 00 = L kf[k]z-k = Z {kf[k]u[k]} k=O I::. E xercise E ll.7 U sing E q. (11.22), derive P airs 3 a nd 4 i n T able 11.1 from P air 2. Similarly, derive P airs 8 I nterchanging t he o rder of summation a nd i ntegration a nd 9 f rom P air 7. \l Initial and Final Value = 2!j f Ftlu]F2 [~] For a causal f[k], u- 1 J[O] = lim F[z] (11.23a) z~oo du T his result follows immediately from Eq. (11.9) We can also show t hat i f (z - l)F{z) has no poles outside the u nit circle, t hen LTID System Response I t is interesting t o a pply t he t ime convolution p roperty t o t he LTID inputoutput equation y[k] = f[k]*h[k]. I n Eq. (11.14), we have shown t hat h[k] ¢ => H[z]. Hence, according t o Eq. (11.18), i t follows t hat Y[z] = F[z]H[z] (11.20) lim f {N) = lim{z - l)F{z) N --+oo 1 1.3 z --+l (11.23b) Z- Transform Solution o f linear Difference Equations E arlier in t he c hapter, we derived this i mportant r esult using informal arguments. Multiplication by 'Yk If f[k]u[k] ¢ => F[z] 'Yk f[k]u[k] ¢ => F t hen [~] (11.21) T he time-shifting (left- or right-shift) pr...
View Full Document

This note was uploaded on 04/14/2013 for the course ENG 350 taught by Professor Bayliss during the Spring '13 term at Northwestern.

Ask a homework question - tutors are online