Chapter 5 transform analysis of linear time-invariant system 5.1 the frequency response of LTI system 5.2 system function 5.3 frequency response for rational system function 5.4 relationship between magnitude and phase 5.5 all-pass system 5.6 minimum-phase system 5.7 linear system with generalized linear phase 5.1 the frequency response of LTI system H ( e j ) h[ n]e jn n H (e j ) | H (e j )|e jH ( e j ) magnitude-frequency characteristic: 1. | H (e j ) | magnitude response or gain 2. | H (e j ) |2 H (e j ) H (e j ) magnitude square function 3.20 log10 | H (e j ) |, unit : dB log magnitude 20 log 10 | H (e j ) | magnitude attenuation log magnitude linear magnitude transform curve from linear to log magnitude phase-frequency characteristic: 4.H (e j ) phase response 5. ARG[ H (e j )] principal phase 6. arg[ H (e j )] continuous phase d arg[ H (e j )] group delay 7.grd [ H (e )] d j Figure 5.7 EXAMPLE Figure 5.1 understand group delay 0.85 ,0.25 ,0.5 Figure 5.2 0.5 ,0.25 5.2 system function H ( z) h[n]z n n Characteristics of zeros and poles: (1)take origin and zeros and poles at infinite into consideration, the numbers of zeros and poles are the same. (2)for real coefficient, complex zeros and poles are conjugated, respectively. (3)if causal and stable, poles are all in the unit circle. (4)FIR:have no nonzero poles, called all-zeros type, steady IIR:have nonzero pole; if no nonzero zeros , called all-poles type EXAMPLE y[n] x[n] x[n 1] y[n 1] H ( z) 1 z 2 1 z 1 1 z 1 y[n] x[n] x[n 1] Difference about zeros and poles in FIR and IIR 5.3 frequency response for rational system function H ( z) H (e j ) 1.formular method H (e j ) H ( z ) |z e j 2. Geometrical method M H ( z ) Bz N M (z c ) k k 1 N (z d k ) k 1 M | H (e j ) || H ( z ) | z e j | B | j | e ck | k 1 N | e d j k | k 1 M arg[ H (e )] arg[ B] arg[ e j k 1 j N ck ] arg[ e j d k ] | ( N M ) k 1 EXAMPLE | H (e j ) || B | L1 , arg[ H (e j )] 1 2 3 L2 L3 magnitude response in w near zeros is minimum, there are zeros in unit circle, then the magnitude is 0; magnitude response in w near poles is maximum;zeros and poles counteracted each other and in origin does not influence the magnitude. EXAMPLE H ( z) 1 1 az 1 , | z || a | | H (e j ) | 1 /(1 a) 1 /(1 a ) a 0 2 ω arg{ H (e j )] ω 0 2 EXAMPLE y[n] x[n] x[n 4] 3.matlab method EXAMPLE B=1 A=[1,-0.5] figure(1) zplane(B,A) figure(2) freqz(B,A) figure(3) grpdelay(B,A,10) H ( z) 1 1 0.5 z 1 5.4 relationship between magnitude and phase | H (e j ) | H ( z ) | H (e j ) |2 H ( z ) H * (1 / z * ) | H (e j ) |2 e j z uniform 4 conjugate reciprocal zeros 4 conjugate reciprocal poles nonuniform H ( z) EXAMPLE Pole-zero plot for H ( z ) H * (1/ z* ) ,H(z): causal and stable, Confirm the poles and zeros z1 , z 2 , z3 z , z , z 1 2 6 p1 , p2 , p3 z 4 , z5 , z3 z 4 , z5 , z 6 Figure 5.20 5.5 all-pass system | H ap (e j ) | cons tan t Mr 1 dk Mc ( z 1 e * ) ( z 1 e ) k k (1 e z 1) (1 e * z 1) k k k 1 1 d z H ap ( z ) A k 1 z k 1 Zeros and poles are conjugate reciprocal For real coefficient, zeros are conjugated , poles are conjugated. EXAMPLE Y 3/4 N 4/3 Y Y Characteristics of causal and stable all-pass system: grd[ H ap (e j )] 0 arg[ H ap (e j )] 0, for 0 application: 1. compensate the phase distortion H ( z ) H ap ( z ) H ' ( z ),| H (e j ) || H ' (e j ) | 2. compensate the magnitude distortion together with minimum-phase system H ( z ).H min ( z ) H ap ( z ) 5.6 minimum-phase system inverse system: H ( z ) H i ( z ) 1, that is , H i ( z ) 1 / H ( z ) then , X ( z ) H ( z ) H i ( z ) X ( z ) or : h[n] * hi [n] [n], then , x[n] * h[n] * hi [n] x[n] the condition of h[n] * hi [n] [n] :, ROC of H ( z ) and H i ( z ) must have intersecti on explanation: (1)not all the systems have inverse system。 (2)inverse system may be nonuniform。 (3)the inverse system of causal and stable system may not be causal and stable。 the condition of both original and its inverse system causal and stable: zeros and poles are all in the unit circle,such system is called minimumphase system, corresponding h[n] is minimum-phase sequence。 poles are all in the unit circle, zeros are all outside the unit circle, such system is called maximum-phase system。 minimum-phase and all-pass decomposition: If H(z) is rational, then : H ( z ) H min ( z ) H ap ( z ) | H (e j ) || H min (e j ) | poles outside the unit circle all-pass system: counteracted zeros and poles, zeros and poles outside the circle minimum-phase system: conjugate reciprocal zeros and poles zeros outside the unit circle Application of minimum-phase and all-pass decomposition: Compensate for amplitude distortion Figure 5.25 H d ( z ) H min ( z ) H ap ( z ), H c ( z ) 1 H min ( z ) , total system : H ( z ) H d ( z ) H c ( z ) H ap ( z ) Minimum-phase system and some all-pass system in cascade can make up of another system having the same magnitude response, so there are infinite systems having the same magnitude response. H min ( z ) H ap ( z ) H ( z ) | H min (e j ) || H (e j ) | Properties of minimum-phase systems: (1)minimum phase-delay arg[ H min (e j )] arg[ H (e j )] arg[ H ap (e j )] 0,0 (2)minimum group-delay grd[ H min (e j )] grd[ H (e j )] grd[ H ap (e j )] 0 (3)minimum energy-delay(i.e. the partial energy is most concentrated around n=0) define : E[ n] n 2 | h [ m ] | partial energy , then m 0 Emin [ n] E[ n] | H min (e j ) || H (e j ) | n 0 n 0 | h[ n] |2 | hmin [ n] |2 then, Emin [] E[] but , Emin [ n] E[ n], n EXAMPLE 最小相位 minimum phase maximum phase Systems having the same magnitude response Figure 5.30 minimum phase Figure 5.31 Figure 5.32 5.7 linear system with generalized linear phase 5.7.1 definition 5.7.2 conditions of generalized linear phase system 5.7.3 causal generalized linear phase (FIR)system 5.7.1 definition Strict: H (e j ) | H (e j ) | e j | | arg[ H (e j )] (line ), grd[ H (e j )] (real ) Generalized: H (e j ) A(e j )e j j | | A(e j ) is a real function arg[ H (e j )] , grd[ H (e j )] Systems having constant group delay phase EXAMPLE ideal delay system hid [n] [n m] y[n] x[n] hid [n] x[n m] H id (e j ) e jm EXAMPLE | | differentiator:magnitude and phase are all linear H (e j ) j / T j / 2 e , 0, / 2, A(e j ) T T physical meaning: all components of input signal are delayed by the same amount in strict linear phase system ,then there is only magnitude distortion, no phase distortion. it is very important for image signal and high-fidelity audio signal to have no phase distortion. when B=0, for generalized linear phase, the phase in the whole band is not linear, but is linear in the pass band, because the phase +PI only occurs when magnitude is 0, and the magnitude in the pass band is not 0. EXAMPLE square wave with fundamental frequency 100 Hz linear phase filter: lowpass filter with cut-off frequency 400Hz nonlinear phase filter: lowpass filter with cut-off frequency 400Hz Generalized linear phase in the pass band is strict linear phase Generalized linear phase in the pass band is strict linear phase 5.7.2 conditions of generalized linear phase system 0 or (1)2 M (int eger ) h[2 n] h[n] / 2 or 3 / 2 (2) 2 M (int eger ) h[2 n] h[n] Or: (1)if : h[ M n] h[ n], n ,..., M is a int eger , then : 0 or M / 2 ( 2)if : h[ M n] h[ n], n ,..., M is a int eger , then : / 2 or 3 / 2 M / 2 M:even M:odd Figure 5.35 M:not integer EXAMPLE M:not integer EXAMPLE determine whether these system is linear phase,generalized or strict?a and ß=? (1) (2) 3 2 2 (3) (4) 3 2 1 1 5.7.3 causal generalized linear phase (FIR)system h[n] h[ M n],0 n M h[n] 0, for n 0 or n M (1)typeI : h[n] h[ M n], M : even (2)typeII : h[n] h[ M n], M : odd (3)typeIII : h[n] h[ M n], M : even (4)typeIV : h[n] h[ M n], M : odd Magnitude and phase characteristics of the 4 types: M ( ) / 2 2 A(e j ) M M h(n) cos n 2 2 n 0 M M /2 typeI : A(e ) h[n] cos n a[k ] cos( k ) k 0 n 0 2 where : a[0] h[ M / 2], a[k ] 2h[ M / 2 k ], k 1,2...M / 2 j M M M /2 typeII : A(e ) h[n] cos n b[k ] cos( (k 1 / 2)) k 0 n 0 2 where : b[k ] 2h[( M 1) / 2 k ], k 1,2...( M 1) / 2 j M M M /2 typeIII : A(e ) h[n] cos n c[k ] sin( k ) k 0 n 0 2 where : c[k ] 2h[ M / 2 k ], k 1,2...M / 2 j M M M /2 typeIV : A(e ) h[n] cos n d [k ] sin( (k 1 / 2)) k 0 n 0 2 where : d [k ] 2h[( M 1) / 2 k ], k 1,2...( M 1) / 2 j M I | H (e j ) | ARG{H (e j )} grd{H (e j )} II III | H (e j ) | ARG{H (e j )} grd{H (e j )} IV Characteristic of zeros: commonness 1/z1* z3 z5 z1 z2 1/z2 z4 z1* 1/z1 z3* Characteristic of every type: Figure 5.41 characteristic of magnitude get from characteristic of zeros: type I: H (e j ) 0 type II: type III: H (e j 0 ) 0, H (e j ) 0 type IV: H (e j 0 ) 0 Application of 4 types of linear phase system: M is even M is odd low high band pass band stop low high band pass band stop h[n] is even (I) Y Y h[n] is odd (III) N N Y Y Y N Y N Y N (II) N Y Y N (IV) summary 5.1 the frequency response of LTI system : H ( e j ) 5.2 system function 5.3 frequency response for rational system function: j H ( z ) H ( e ) 5.4 relationship between magnitude and phase : 5.5 all-pass system | H (e j ) | 2 确定 H ( z ) 5.6 minimum-phase system 5.7 linear system with generalized linear phase (FIR) 5.7.1 definition: 5.7.2 conditions : h[n] is symmetrical H (e j ) A(e j )e j , | | 5.7.3 causal generalized linear phase system 1.condition 2.classification 3.characteristics of magnitude and phase , filters in point respectively 4.analyse of characteristic of magnitude from the zeros of system function requirement: concept of magnitude and phase response, group delay; transformation among system function, phase response and difference equation; concept of all-pass, minimum-phase and linear phase system and characteristic of zeros and poles; minimum-phase and all-pass decomposition; conditions of linear phase system , restriction of using as filters key and difficulty: linear phase system exercises 5.17 complementarity:minimum-phase and all-pass decomposition 5.21 5.45 5.53 the first experiment Get subjects from the experiment instruction book or downloading from network: problem 1(D) problem 11 problem 13(C) problem 22(A) problem 24(A)(C)