pulse Response representation for discrete-time LTI systems will discuss to construct any discrete-time signal x(n) in tert : ms of 2.2.1 Im In this section, initially we time shifted unit impulses. Consider a discrete-time signal x(n) as shown in Fig. 2.1. For simplicity, we have taken § non-zero samples only. Fig. 2.1 This siens 5 sae A his signal x(n) shown in Fig. 2.1 can be splitted as shown in Fig. 2.2 (a) to (e) bel . 2. elow. x(n) = r(-2)5(n+2)signals & Svstems 231 a(n) = (0)5(n) 2 wei 2 ob es 1 " © a(n) = 2(1)6(n—1) 1 So n eee cee 304 o 5(n) = 2(2)3(n—2) = 0s 249) Aviad gal © aiiim Fig.22 By observing Fig. 2.1 and Fig. 2.2 we get, x(n) = x(—2)6(n + 2) + x(1)6(n + 1) + fowin + x(1)5(n = 1) + x(2)6(n = 2) 2 x(n) = > x(K)5(n = k) fay Soin general, any discrete-time signal x(n) can be expressed as, * x(n) = . x(k)6(n — k) 1 = Inegn. (2.1), we ex ; Pressed x(n) as the weighted sum of time-shifted impulses. Now consider a discrete-time LT1 s: yystem as shown in Fig. 2.3. a(n) T {-}-— y(n) Fig. 2.3 Where x(n) is input to the system, ¥(1) is output of the system. T(} is a system operator.= 2. Time Domain Representations fg 232 ime Ons Fr TT 2. We have, y(n) = Tix(n)} Qa) Substituting eqn. (2.1) in eqn. (2.2) we get, y(n) = 1 > (k(n - v} Kare Using linearity property we get, + ya) =D) xdoT {én - 15} ny ite : In eqn. (2.3), T{s(n ~ k)} corresponds to the operation of the system performed on time shifted impulse 6(n — k). Using time-invariance property we get, -. T{6(n - k)} = h(n -k) (24) where h(n) is the impulse response of the system. Substituting eqn. (2.4) in eqn. (2.3) we get, yea) = J) x(oh(n - k) @3) K y(n) = x(n) + h(n) 26 From eqn. (2.5), we can say that the output of an LTI system is given by a weighted super Position of time-shifted impulse responses. The eqn. (2.5) is known as convolution sum. . This concept is better understood by the following example. Consider an LTI system having impulse response h(n) as shown in Fig, 2.4 below. h(n)signals & Systems “ pene Systeme Now say. We want to find the output y(n) of this system to an input x(n) shown in x(n) " Fig. 2.5 The output y(n) is obtained by adding the responses due to each individual sample of the input as shown in Fig, 2.6 n(n) ny t 1 1 f h(n) 05 annem 0.25 a o—+ 1 2101s 34 2012354 5 2(n) ya(n) 1 1 | h(n) | ie + 0.25, eet eae n 21 0 1 tf 210 12 13 a(n) , 05 h(n) yalt) + — 02% Seen 21012 3 4 27 012 r4(n) os h(n) + l — +1 +n 7 7 21012 3 42(n) = 2 (n)+22(n)x3(n)z4(n) i 05 n(n) n Sao wese Fig. 2.6 Pye: [ia sh] him: fe ys 4] latte) 4 end 3 r a1 ) | | er a 3 ora sa a3t 3 tal ; et @_P |e yo H | Ly tf th FG) a) oe eee | : ‘a 3 7 wy 2 aaa! 3 ha) yy) : 41 = |} se oO aan or 1 = [| ws | a 4 Sq gee qo BT TI ee = V exible 29 % 30 4 a1 3 6 6 I | | eee | ty [ a nrG , ee $ bt A» Lhe “4h Ae 34 qa Be 6 se ss He sy] ' 215 36 3255 30 ailY Evaluate the duce - time tonvelutian um given f, eto? yin) © Ula) 9 OL0 -3) Set"r Connidea — MWA) = LCN) HLM = LU9-3) The Aigale —%,C0? G Ayn) are plotted Sespechvely Abown boabow- Ho) — Ae ~ o1;ns nm % tn). III, 6412 BS EM We hove to obkain. yen) ~ UM) + Oln -3) oq) %&L) T= XL) %,0 -K) Ke -0 ber ut daa Ck)» Gt UL obkained by Aeplacing \ bbe behow « wu Ox wa w xk) ce nd ° t 2 3 AWD obk gin CK). Then Obkain %yln- wth nao, & Kating the miso image 4 C6) A thown 39 below tig x Cn-&) “6-5-4 3-3 -) 0 KOnc . © we take me meg image wraron we gee mY Conti ike St or Kah - ee ee Aeit Be oem eo" qd WL -K) With nso ok Ke 3: > Now we hae fo Multiply 2) & Bh ted ical out for the entire wk) , eo Auemokien musik be Casried K- axis te, —we 3 4 « @ ko. 1 \ xte) @ K-1 | a I aaa aaa “not K owe hun’) [+a +a ey yaeap ene ita Pee ae dint equal qed = n : ay qo =) nZo x) 0 n 1 232 n. bi 2 yin = ae) hw) K-10 4 xu) ie hun-k) Ane) x) = | Ye \ 1 [LI] 2 Kk “3-2-1 Oo K @ ko " hur \ ee 2 te A ; kK. mee He P ry i L wa +——.. 1+ o rhot) ij. 2 “372-1 Yu fe AN wu “sop eK 20. Ig tn) = 4 3p St n. “3-274 ° ' 2 3 qo: oO n || = x [3 uth) ->/3] a Zt. mn i=o Hence, the system is stable 2. Given y(n) = nx(n) System multiplies the input x(n) by n, therefore, h(n) =n Hence, the system is stable if |n| < co, EXAMPLE 3.3 Di i q 7 : . aes etermine, whether discrete LTI system given by h(n) = [AJ" u(n) is causal and Solution Causal: Given in) = [BP un)From Eq. (2.10), we have, u(n) = {i forn20 0 forn<0 Therefore, A(n)=0 forn<0 Hence, the system is causal. Stable: From Eq. (3.21), the system is stable if s [R(k)] <0 ¥ |r] = x Wate = Sat |= rp ices i mp | Therefore, the system is stable if || < 1, and the system is unstable if || = 1.43 CORRELATIONS OF DISCRETE TIME SIGNALS Mathematical operation of correlation closely resembles convolution. In both convolution and correlation two signals are involved. As discussed in Section 3.2 convolution produces an output by convolving the given input signal x(7) with the impulse response h(”) of an LTI system and output (7) is given by Eq. (3.11). . yeny= x) h(n k) (43) k==e0 In correlation there are two types + Cross-correlation + Auto-correlation 4.3.1 Cross-correlation of Discrete Time Signals Cross-correlation r,,(k) is used to measure the similarity between two sequences x(72) and y() Cross-correlation of two sequences x(n) and y(n) is defined by ry (k)= y x(n) y(n =k) k=04142-5 ad or, equivalently as Ty (k)= > x(n +k) y(n) 4.7) k=04142-- a : itive Hero, index & takes only integer value, In Eq, (4.6) the sequence x(n) is fixed and for posits values of k the sequence (1) is shifted right by k units in time: for negative values of k the sea