You are on page 1of 12
E412 A02 s¢c1 1SE47 IMPERIAL COLLEGE LONDON DEPARTMENT OF ELECTRICAL AND ELECTRONIC ENGINEERING EXAMINATIONS 2005 MSc and EEE/ISE PART IV: MEng and ACGI DIGITAL SIGNAL PROCESSING AND DIGITAL FILTERS Tuesday, 3 May 10:00 am Time allowed: 3:00 hours There are FIVE questions on this paper. Answer THREE questions. ected COPY All questions carry equal marks Cort Any special instructions for invigilators and information for candidates are on page 1. Examiners responsible First Marker(s) P.T. Stathaki ‘Second Marker(s): A.G. Constantinides University of London 2005 | (a) An M“-order, causal, real transfer function is given as Gg +027! ++ Oy 2™ Ay) aoe by 4B et by? M>0 (i) Derive the conditions on the numerator and denominator coefficients to yield [ay (e!))? =1, forall G1 (ii) Comment on the location of the zeros of a causal, real, stable, allpass transfer function within the z-plane with respect to the unit circle and the poles. Justify your answer. (1 (b) Consider the first-order, causal, real, stable, allpass transfer function 4,(2) (i) Determine the expression for 1~|4,(z)\° (5) (ii) Show that <1 for [> [4(@f=1 for [2|=1 >I for fe|<1 {5} {c) Discuss a possible application of allpass filters. {41 | Dygutal Signal Processing and Digital Filters}, Page 1 ofS * (a) Show that the transfer function ay Ie has a lowpass magnitude response. Determine the 3-dB cutoff frequency @, at which the gain response is 3 dB below the maximum value of 0 4B at @ = 0 (51 (b) Show that the transfer function A @) has a highpass magnitude response. Determine the 3-dB cut-off frequency @, at which the gain response is 3 dB below the maximum value of 0 dB at «@= 7. (1 {c) The lowpass transfer function H,(2) of equation (1) and the highpass transfer function H,(2) of equation (2) can be expressed in the form 1 Hy2)=51A@)- Al), Ha ‘4+ 402] where 4,(z) and 4,(z) are stable, allpass transfer functions. Determine A\(z) and Az) [10] Digital Signal Processing and Digital Filters] Page 2 0f5 (a) Show that a Type | linear-phase FIR transfer function H(z) of length 2M +1=7 can be expressed as ul), % merry] [HM DAE — m2 +l (41 (b) By using the relation where T(x) is the r~th order Chebychev polynomial in x, express H(z) of Part (a) in the form “a feartY Barf 243 ) ff He) Determine the relation between afr] and h(n) [8] (c) Develop a realization of H(z) based on equation (1) in the form of Figure 1, where F,(z"') and F(z") are causal structures. Determine the form of F,(2"') and F,(z™) (81 [The Chebychev polynomials satisfy the following recursive relationship: T,(x) = 2x7, (x) ~T, (3,722 Tye)=1, T()=x] : Lae) LAE) Fe) Fe") | ital Signal Processing and Digital Filters] Page 3 0f5 (a) ‘The bilinear transformation from the s~ plane to the 2 ~ plane is given by 1e2" (i) Explain using mathematical relations, where each p 5 — plane is mapped on the z ~ plane through the bilinear transformation. s=o+jQ on the {61 (ii) Prove that the relation between the continuous frequency © and the discrete frequency @ is non-linear. (4) (by A given real lowpass digital IR filter has a rational transfer function #7, (2) and a digital cutoff frequency @,. The transfer function is transformed by replacing z by F(é), to H ,(F(2)) of the same order as H(z) and another real lowpass rational form Hp (2) cutoff frequency @ (i) Give full account of the properties of F(2) in order to effect the transformation. BI (ii) Starting with a general expression for F(Z) find a specific expression for F(2) which has all its parameters derivable from the requirements of the problem. m Digital Signal Processing and Digital Filters] Page 4 of S < (a) (i) Explain the function of a down sampler with a down sampling factor M where Mis a positive intoger. Give the relation between the output sequence y[n] of a down sampler with a down sampling factor Mand its input sequence x[] (4) ii) Derive the relation between the spectrums of the input and the output of a down sampler. (b) Determine the condition under which a cascade of a factor of L. up sampler is interchangeable. | bea Signal Processing and Digital Filters] Page $ ofS Eee Tre «7 Ba tw Meri ae Fue Tees Zees Aor, cl | (a) 4) An M"-order causal real-coefficient allpass transfer function is of the form If we denote the denominator polynomial of the allpass function 4,,(z) as Dyy(2) with Dy(2)= Ved +t dy 2 dy then it follows that 4,,(2) can be written as Ay) From the above we see that Ay 2)Ay ay el) = Ay (2) (ii) The poles of a causal stable transfer function must lie inside the unit circle. As a result, all zeros of a causal stable allpass transfer function lie outside the unit circle in a mirror-image symmetry with its poles situated inside the unit circle. (b) Consider the first-order causal and stable allpass transfer function 4,(2) with its pole located at d, Gi) Determine the expression for | =|4,(2)|" ded le+d, 20 we see that (i) 2 0/4 (ef <1 “The rest are proved in a similar fashion te) A simple but often used application of an allpass filter is as a delay equalizer. Let G(=)be the transfer function of a digital filter that has been designed to meet a prescribed Page 4 ot 6 ‘magnitude response. The nonlinear phase response of this filter can be corrected by cascading it with an allpass filter section A(z) so that the overall cascade with transfer function G(z)A(2) as a constant group delay over the frequency domain of interest. Since the allpass filter has @ unity magnitude response, the magnitude response of the tll equal to |G(e’") , while the overall delay is given by the sum of the group cascade is delays of G(z) and A(z), The allpass is designed so that the overall group delay is approximately a constant in the frequency region of interest (a) For the transfer function we have H, (jo 2 ae aor =) (reese) tsin?@ _(l-a)* __reos@ 2) (-acosa)i +a'sintw 2 Ira? -2acoso The above has a lowpass magnitude response with a monotonically decreasing magnitude (+60: 1,0) aoe I cos)" +a? sin? 0 (=a)? ae 2 ape ue (1+ coszy ssi (4 2) a bat) (-acosz) ta?sinte (2) (ray +0 AH (J@)")_ =a)? (-sin (+a ~2a cose) ~ (I+ c08@)2ersin @ do 2 (+a? -2acoseo)? _ (lea) sino(-1-a? + 2ecos@-2a-2acoso) __(I-a)?__(1+@)* sino 2 (+a? ~2acosw)? 2 (+a? -2acose)? which is monotonically decreasing within the range « =(0,] ‘The 3-dB cut-off frequency «, at which the gain response is 3 dB below the maximum value of 0 dB at @=0 is found below (=a)? _treose, 1, 2 14a? -2acoso, 2 “Tea? ae? (by For the transfer function Leal a 2 5 @) we have |HVjay? (34) (cose)? +sin’@ __ (+a)? |-cosw (-acose)? +a? sin?a 2 acoso The above has a highpass magnitude response with 2 monotonically increasing magnitude since y (cose) +sin?@ (lta)? (1-ecose)? +a sin? cos@ 2acosa d(H (jo) (+a)? sina +a? ~ 2a cose) ~(1-cose)2asin do 2 (14a? —2acosa)* 2a+2acosw) (Ita) 2 (+a? ~2acose)* 2 (+a? is monotonically increasing within the range «@=[0,]. IB cut-off frequency «, at which the gain response is 3 dB below the maximum value of 0 dB at c= 0 is found below ) sino 2acosa)* [4 Yoy=1|4(joy]=1 «a H(z) > WO} + Alle! + A[2]z* + AB]e* + ALA]z * + AfS]z~* + Al6]2® {6 = HO}, ALS] = LL, ALA] = AL2] H(z) = WO\L + 2°) + Alle! + 28) + A(z? +24) + AB] => SAIS] + AfOK2! + 2°) + Ae? +27) + wLQWe + 21)} with 7, (x)= 2x7, 0X) = T,-2Qyr 22 T(x) =1 T(x) =x Ty(xy= 2x? 1 T(x) = 2x2? -I) aan! ~3e Thus Page Bet 6 For reasons of simplicity | call ce M()=2 al] = A{3]— 2411 a (09) *(AL3}x” + ALOW Sx! — 6x) + ALL 4x? — 2) + 2A[2]x) S[ML3] - 2LID)x” + (2A(2)— 6ALOpx + 4A[Lp? + SAO") who HEL Sam") {2} ~ 6h{O],a{2] = 4h]. al3] = BAO} {c) Develop a realization of H(z) based on equation (1) in the form of Figure 1, where F(z ') and Fy(2"') are causal structures, Determine the form of F(z!) and F,(2"") wo li Gi) tb) ib (i) ley is 14 jy 1-9, implies that a point on the imaginary axis in the s~ plane is mapped onto a point fon the unit circle in the 2—plane where |2|=1. In the general case, for v= 04+ jQy, For s= jQ, we have that which has a unity magnitude. This ston + 0) pron) +05 T= (a4 + jQy) (l-ay)' +Q5 A point in the left-half s ~ plane with @, <0 is mapped onto a point inside the unit cirele in the 2 ~ plane as |z| <1. Likewise, a point in the right-half plane with oy > 0 is mapped onto a point outside the unit circle in the 2 — plane as fot Prove that the relation between the continuous frequency © and the discrete Frequency « is non-linear F(Z) must be real, rational function 2, of order | and stable. Starting with a general expression for (2) as follows Page ket 6 For @=0, 3=e” =1 we require Hy(2) =H, (F(2))=1. Thus y= 228 -1xatbacrd etd For 6=2, £=e!" =-1 we require H(3)=H,(F(2))=0. Thus FO vd Oo Teed From the above conditions we see that b Therefore, we have F(2)==*" Therefore, (2) is an allpass filter. (a) (i) The down sampler with a down sampling factor M , where Mis a positive integer, develops an output sequence y[n] with a sampling rate that is (1/M1)® of that of the input sequence x[1}. The down sampling operation is, implemented by keeping every M "sample of the input sequence and removing M ~1 in-between samples, to generate the output sequence according to the relation, yin] = x[nit | ui) Y(z)= ¥x[Mn je-* cag be] = {sth = OEM AIM ™ (0. otherwise oe] ata] Ex IMn et = E rig ie = Xa 2) Xe] = el ]atn] ln] 1, n= 08M 22M, l= 19, otherwise Wy Setmdvtne" "we Jat" “XW it) +E Sav: (b) Determine the condition under which a cascade of a factor of M down sampler with factor of Lup sampler is interchangeable. abn vil) ene > tL iM Page 5.6 Figure | xt) vata] yale] + iM aq te > Figure2 Figure | we have V,(2) = X (24) an ao For Figure | we have Vi(=) = X(z") and ¥\(2) = 77 For Figure 2 we have iM ba Vyey= Exe way and Hi)= EX gt sade a7 EXC ) and (2) = 5 EXC ) The functions Wy‘ and W,,"* are the same only if the numbers £ and Mare relatively prime, Page 6+ 6

You might also like