‘'gacol the eat rae a type of signal processing pesfomed ‘on! continuous analog signals by some analog means. f Digital signal processing is referedto as processing of sights by digital systems, where signals are represented digitally as sequences of ssimples by performing various mathematical opefations. «The block diagram of a ie soa processing (DSP) system i 5 shown in gue, tbat Digital signal] DIA ‘Analog outpt proceséor Converter signal AD Converter: It convers the analog it inp signal to its equivalent nig value. Digital Signal Processor H Wis the heart of a DSP system.” 1. % -Itcan be a large digital computer or a: Latinas prograitined {6 perform required operations on the input signal. % Itean also be a digital signal processing hardware eonfigued io process the signal Werking: From figure it ean be observed that, he alg inp sal is applied oan A/D converter, which converts thé anal ‘npit to its digital Value sind i then fed to'a digital signal procéssot, The digital signal processor performs required! operations on ‘the iiput signal aad is fed to a D/A converter which converts the digital Signals to their equivalent analog values and produces a ‘continuous alog signal asthe outpit.¢ A : SPECTRUM ALLIN-ONE JOpRIAL FOR Om ENGINEERING STUDENT: Scanned with CamScanner OnProcessing? i ‘March-17, (R13), Qt{e), (Refer any Applications) Ans: Advantages f 1. Flexibility: Dsp 5, i ir ; e rystems have higher flexi ility in Sonfizuration than ASP. ‘systems since ,to reconfigure the ber geo we have change the program whereas to igur asialog system, we hay ete the atalog sytem, we ave to redesign the 2. Accuracy: DSP systems have superior control of ‘accuracy than analog systems, 3. Cost Effective: Digital reatizationo systems is much cheaper than the analog realization, 4 Ease of Data Storage: The storage of digital signals in magnetic tapes is very easy involving no loss or deterioration of signal fidelity. Hence, signals become transportable and can be processed off-line in remote laboratories. : ‘Time Sharing: A given processorcan be shared among ‘umber of signals by time sharing property of DSP. Thus reducing the cost of signal processing. led’ Algorithms: Implementation of highly sophisticated algorithms can ‘be done with the use of DSP. Disadvantages se states 1. "System Complezity: The use of devices like A/D and D/A converters and their respective filters in the digital Processing of an analog signal increases thie compl of the system. : ; . Low Speed: Most of ihe DSP's iised for signal Processing have very less speed, which is not! enough for processing of signals with large bandwidth. Power Cos ‘consisting of more than 4 lakh transistors results in power dissipation of approximately 1 watt. © <5!» Applications 1. . Communication: DSP firds applications in traismission channels fot elimination of noise and echo cancellation by adopting various filter techniques, Image Processing: DSP is applicable in various image processing techniques sich as image compression, image enhancement, image analysis and recognition. = Medicine: It is used in the field of medicine to perform ~ speetral analysis of ECG and EEG signals to detect the disorders in heart and brain. 3 3. 3. mption: A digital signal processor |, DIGITAL SIGNAL PROCESSING [JNT WARNING: ‘Xerox/Photocopying of this book is CRIMINAL act. 5. A 4 ead Points to Remember “3 Signal are broadly categorized info the following per 7] 2 DSP techniques are “Speech Processing: D: om speech analysis methods forautorat me Speaker verification and speaker identification, er Electronics: It also pro ceerimer electronic applications such as Digit, TV, electronic musi syithesizer, Karaoke’ yg 7 7m . educational toys 4 ‘Sélsmology: Iti also helpful in spectrum analy, seismic signal to estimate earthquake, volcanics nuclear explosions and earth movement 3 Military: DSP also has wide applications in such as radar signal processing, sonar signal navigati 2" Discnere Time SiGNALS ano Si ‘CONVERSION OF CONTINUOUS 7; PpiscRETE SIGNALS, NORMALize; nity! | Contiauous-time sign own as Continuoiis-time signals." ‘A continuouistime signal is also referred as analog 2s signal 3° = known as discrete time signals. ed by x(n). Anyone fund guys UABLEt ace Scanned with CamScanner OwVirite short notes on Discrete-Tim igni & Signals that are represented at discrete instants of time are called diserete time signals, @ A discrete time signal is function of af independent variable which is an integer. tis defined for every integer value n for ~00 0 (= Ovelsewhere The graphical representation of unit step sequence is as shown as figure (2). ‘ ECTRUM ALL-IN-ONE JOURNAL FOR ENGINEERING. STUDENT: Ss: Scanned with CamScanner OwB me “eta 2 ue .DIGITAL SIGNAL PROCESSING [JNTU-HYDERABAp, : y » 3.5 UnitRamp Sequence ~ a 5 aul z Unit ramp sequence is defined as, ~ey [nr for nz0 u)= 10 for n 1, thine ‘Sequence is called rising exponential sequence. ‘The graphical representation of exponential sequence for different values of parameter ‘a is as shown in figure. @. Qeacd 2 Ure i a>. * a=re ‘Then, x(1) can be represented as, X()= (rer es ape, SE) ont ein On) fata LEGAL stadia’ Scanned with CamScanner Ow| i | | _ i “Figure ie <1 r= zo) and xf) wil have fixed ampliidé of unity as shown in figure (6). a ee ae ; Faw (ter = 1 + 20, Show hate) ws MN ae z 7 Statement: 5a)= as u(a— 1) + Proof: The plots of unit step sequence, (nad its delayed version ie, (n—1) areas shown in figures (1) and (@)respestvey. = *, (0) w=) 2 3 ‘ Figure (toa) : Figure @kwln= 1) From figures (1) anid @ the sgaomes if aang by subtracting the sequence un) {from x(n) plotted as shown @). " Scanned with CamScanner Omw “The discrete signals ae’classified as, Energy and Power Signals Energy Signal - The eneray ofa discrete signal xis defined as, En Lhe @ Itcan: sometinies ‘be finite (of) infinite, “ILE of the signal is finite (Le,, 0 < £ < ee) then it is referred as energy signal. ‘Signals with finite energy have zero average power. ignals with infinite energy may have finite or infinite power: ‘> _ Ifboth energy and power signals are infinite then the signa is neither an energy signal nora power + signal, ‘©. Asignal gannotbe bh an energy signal ora pov signal. “All practical signals are energy signals. Example: (7) =de* i Power Signal =. | * Sa: The Poet iso gent Ps Mona. wa : Shon Adiscreté: geist iipatity ce of thé signal is finite and energy (E) is infinite. ° _Mf the average power of sgl i finite, then it is “referred as power signal," i © Signals with finite powier will Be have infinite Example: x(2)= Auln): Period and AperidleSigals relation holds is called the fundamental period. A signal is aperiodic (ot) non-periodic if |’ ant N) #x(n). Sek The sum of Mf periodié discrete time sequences with periods 1,,’¥,..N,,is always periodic with period N where N= LCM(N,, N,N): ‘An example of periddic and aperiodic sequence is |" as shown in figure (1). WARNING: XergxIPhotocopying ‘of this book CAMANAL Ane oi iyi ABLE fa LEGAL paced. Totrres— Sacre re Figure (1: Periodic and Aperiodie Sequence Symmetric (Even) and AntL-aymmetrc (Odd) Sing, % A Signal ais sid to be symmetic (ot) eve, x(n) = 3(-n), “The graphical representation of the symmeg, signal is as shown in figure (2). x(a) “432101234 7 Figure (2): Even (ar) Symmetrie Signal $A signal x(n) is said to be anti-symmetric oro signal, ifx(-n)=—x(n). Example: x(n) = Asision ~ % _ The graphical represestation of anti-symmeti signal is as shown in figure (3). a) ae Figure (3): Odd (or) Anti-symmetrie Signal + Every signal can’ be expressed as the sum of “even and odd components, _ fe, x(n)= x(n) +x(n) Scanned with CamScanner OwUNIT-1 (Introduction {0 Digital Signal Processing and Multl-rate Bgtal Bignal Prcosingy) Product of eequences, fiven x Odd Odd” 04d * Odd » Even Even # Even Even Causal and Non-eausal Signals 4 @Asignal i said to be causal, ifx(n) 0 for n <0, Example: x(n)" a7 u(n) © A signal is said to be non-causalifx(n)4.0 for n <0, Determine the signal, State x)= n( 2). Ans: ‘The given signal is, (™ onal The energy ofthe signal is given by, z- F por power or energy ‘The power ofthe signal is piven by, SSE p ais SE zin se Since, the energy of the signal “ais finite. Wis a power signal }PECTRI () -x(n)= cos n+ cos wn O23, “Detarmine viheitiar ornotaach Atha taining, algnals {8 porfodic, the signal ts periodic, apacity Its fundamental period, () kn) oe wy x(n) = con n+con Ans: (x(n) oreme “The given signal is, an) = er Onecomparingtheabrreenption vith °C the sraper frequency can be written 2, 0," om Since, fundamental frequency is a multiple of 7, siggal is periodic, Fundamental prod 1-20] 2 {| 4 . ‘The minimum value of m Sor which 1 is an inteyes is 3 3 Mazel , Fundamental period = 1 3 Given signal is; = xn) = cos nt 068 F (On comparing the above equation with A eovn, the angular frequency can be written 2s, Then, the ratio of two fredquenties f and fis calculated as, Scanned with CamScanner OwTolational “The minimum value of m for which N, is integer is 1. N,=6()~6 Tie fimdamental period of cos Penis s m ce of ]-$0 4 ‘The ininimum value of m for which N, ig integer is 3, 8 it x= 50) : =8 % ‘Fundamental period of the signal is, N=LCM of 6 and 8 =3x8 =24 4 i 1 ‘G24. Explain the conversion of continuous time signal to discrete-time signal. Ans: + ‘The process of converting continuous ‘known as sampling. : ps * Gt me apling period T asshown figure, Figure (1: Samples of sty %. The process of sampling is explained by using sampler as a switch is illustrated in igure (2). Switch x) —0 v—x(v) Figure (2: Sampler as a Switch + Aiserete time signal uniformly distributed in time'is |} ‘The conversion process is carried out by representing |: ‘the continuous time signal a discrete time instants that |; DIGITAL SIGNAL PROCESSING [JNTU. HYDERABAn, + + + ° is. WARNING: XeruPitpgol ikis CRNA A ay From figure ( observed tha ofan, And the output (L., discrete sample) is taken at end. Met, The output is obtained by closing the switch, for, 'T seconds. ; When the switch is closed, the output signat same amplitade'as that-of x(0) and when the opened, the output signal will have zero valu ‘Will have Swick . The continuous time signal have tobe sampled oar, ~ original signal ean be reconstructed fom these sng, otherwise the sampling process will be of n0 use, For reconstruction of original signal from the. _ +s, anecessary condition is required which can be obta by analysing the signals in frequency domain, Let x(#) be sampled to obtain discrete time signal fy using a sampling function, g() as shown in figure) 10+) — 00 t Figure (4): Sampling Function 30 = 200. ‘ -0) ‘Here, the sampling function is a continuous tia of pulses that occur at regular time intervals, Tao! functions same as'a Switch. a ‘The safnpling function, g(() shown in igure (4) ca0¥© 80 Scanned with CamScanner Ow“e a '& From equation (4), it can be observed that, the sampled “‘@'°) “Suppose in frequericy domiain, if x(t) has nd ‘sreiieticy conn ave arian Sean he AOA ‘AppiyingFouriertransform to equation (3), we get Spectrum of (7, iexin= [ Sexnererra i xine Lee fxperl-20%d Welnow that, J xed = KY—n/) = x= Zexg=n9 "The frequency spectrum of the sampled signal to, ‘eeonstruct the original signal is, xn=Xexg-m >. =: and is illustrated in figure (S) } i Figura (St: Spectrum of Sainpled Signal signal x0 contains spectrum of conitinious signal, “(0 and spectrum9f continuous signal converted into’ hhamonis of saimling frequency. Here, every frequency converted is multiplied by constant. To reconstruct the original signal we have to pass the| The filer used for reconstuctng the signal is ‘econstruction filter, This filer produces C,X{/) and x(()' atts oitput in frequency and time domains respectively observed that, to fecover X(/) front X,(); samp frequency, must be gréater than (oi) ial to twice the | maximum fequeney 2, 160 f2%ies > LNIT=1{introduction to Digital Signal Processing and Mult-rato Digital Signal Processing) 13 [err An: normalized frequen 1 oo» :Normalized/telaive frequency gives the relation between discrete time signal frequency. (f), continuous time signal frequency (F) and sampling frequency (F_) o- Bear (Here, o = 2nf) Range of discrete time frequencies is, og sted oe ai of ie frequencies is, (@)-xsos < F Se5 (61) 2f," then the continuous time signal may bé completely represented {nit samples and completely recovered back.” ‘or Bandpass Signals: A Banidpass signal x(¢) whose maxi- 3) sum: bandwidth is “2, can be completely represented and recovered from its samples, if sampling frequency is equal to tice the Benda, Scanned with CamScanner OwDIGITAL SIGNAL PROCESSING [JNTU-HYDERARA, ()__sine(100 Rt) +3 sinci(60 mt) Nygaist Rate inimum samp ing aed ‘The given signal is, jnimum sampling rate at which both sampling an a The miner pal (om Hs suroplcs ean be pecforsied 14 z = sinc(100 ni) +3 sinc'(60 mi a wie Its denoted x0 78 “Tout iny dorian ealled Nyquist rat. denoted by sue Be * (0) = sine(100 ni) be oy = 1008 Where, : : uM {Maximum frequency component, _ = a ote ‘1 : = f.=50Hz 4, Nyquist Interval fi: ‘ "Nyquist interval can be defined as the maxitnum time Consider, interval between the equally spaced samples of the signal during 20) =3 sinc}(60 m1) which sampling rate is equal to Nyquist rate: - t , Nyquist inter‘al is equal to reciprocal of Nyquist rate and is given by, ' } 2 Bhan esti aes Where, ji t : i 3 3c05120nr Sou:~Maxiraum frequeicy component, | 7200n?1? -7200n7/7 lyquist sampling Of = 120nt Nyquist sampling interval for. th ‘Signal: i 2 wht = 120m (a) sine{100 xt) + sinc(50 nt) fz 08 = 60 Hz ___, (0), sine(100 nt) + 3 sinc#(60 nt). +; The maximum frequency in x(t) is = f,= 60 Hz Ans: i Then, Nyquist sampling rate, ,=2f_ 2) slne(100 8) + sine(S0 mt) oc bs 38 til 2260) ‘The given signal is, ata i|sgeeo + 340) = sinc(100 mz) + sine(SOnt) Salah sin(l00m) ¢ sin(S0ns) [ BAAS ater som [soe Nyquist sapling inetval, From the above equation, pee 1 Stage me * Nyquist sampling rate = aft = 100m S eta Sf, =50Hz ania = Fp 0.00833, og = Son =) nfs = sO" > frie pons, “+ The maximum frequency inx(0is f= $0 Ha. Then, Nyquist singling rate, =p, Shi =2(50) Ha ng can be defined as a ‘Phenomenon in which hig = 00 He Seti frequency Sofiponents of the signal'appéars as low frequen“! : ‘ Siashhug e's somPonents during. sampling.,t is also, known.as unde ‘ Y toe | oN : Nyquist sampling berm ‘ : - ie The f ‘ate the conditions to eliminate aliasing 1 be of oT yee “ + Nyquist sampling rate’ ‘Too © Jods x(0 must be band-timited to 8 ‘Suitable frequency (/.- =0.01 seconds = 10 msec," * ‘S ] ? se Sampling frequency's must be greaier than twice # i band: limiting frequency (7) T= 10. wee fe) ; le, 5p i WARNING: Xerox/Phetcopyng ofthis bookie CRIMINAL at Any me on ys LADLE OA ped Scanned with CamScanner Owunit {introduction to Digitar Sina Processing and Ms rate Digital Signal Processing) Points to Remember ae basically categorized into two types namely, 229. “Define discrete-time system.and discuss Its classification... Ans: Diserete-time System. * « s-Thesystem which accepts the disereté ime signal as input and transforms it into adisrete time output it said tobe Discrete Time (DT) systems 4 99 ot i y(n) output. Figure (1: Block diagram of a DT system _The relationship betveen input xn) 'and output xo) of DT sytem of a DT system is given by) = Hen)” ‘A system is sai tobe linear if tatlaes he fergie! ‘A superposition principle is a combination of two properties viz, Addiivity ind Nomogenciy.” ‘AdGitvty property sates that; ifthe responde of @ DT'system to x0) ‘and x(n) i y,(n) and y(n) respectively then, the response ofthe sytem to (x(n) +x(n) mist be {60 +34). oe "A relixed system is said to be linear if'and only if, Haj x(n) + 2,20] = a, Meo) ce me(0)) Where, x(n), 20) athe arbitrary input sequences and a,, a, are the arbitrary constants. ‘A relaxed system is said to non. linear if it does not satisfy. the superpositionprinciple.. | * ¥O)= a HERO) + HOC) Inthe above represeitation, if (a) ~3/Cn), then the inown as tnear other it is nonlinear. ‘Time Variant and Time Invariant Systems ips: ofthe dee tine system eb aching wich tim then the systems called time invariant. Ifthe input- utp relationshi Let, x(n) be the input and y(n) be ‘the output of the syster. Attn) AL. y(n), then x(n—K) 2 y(n =) forall possible va {SPECTRUM ALL-IN-ONE JOURNAL Fm. ENGINEERING 81 otk, Scanned with CamScanner OwBEG). eee es ai inpu-ovtpot relationships of discrete time system. ‘Fnage wit time then the system is called time variant ityte, #98 ~ B) even fr one value off then the system is sid to be time-varianYshift variant * 5, Causal arid Non-caitsal Systems s ‘ «If the output of the system at any time depends only «on present input-value (i.e. »{n)] and past input values | [ies, (x(n), x(7 - 1), x(- 2) «.}] but does not depend cori future input values (i.e. x( + 1),x(1 + 2) «J then. the system is knowaras causal system + ‘The mathematical expression for causal system is given bys : A en) = Fle(m), x(01 ~ 1), (7 = 2), x(1- 3), Y(n- 1), 0-2), n-3) . ‘{n)—Present output - x(0)'—Present input. 90s. x(n 1), (0-2); x(n —3),.- are past inputs: Mn—1), (1-2), (n"~3),. are past outputs. “4 iftheourput ofthe system stany time depends onpresent, 4 4 Asjstemis called boinded-input bouided itp stable’ |" ite boned th enone are bounded. © Asysiimiscaledinstabl for some bounded is inputs, the corresponding outputs are infinite: S. Static and Dyaamle Systems = The system is said to be state 0 Ghutpt of system depen only on If the system dutput deperds on. ‘values, then the system is said tobe system. + Bea PRA areata coe a0 Sper he techniques forthe analysis of Trear 3 systems, Ans: 5 nese an ted * following inethods, e ee ae “Direct solution 6finput-outpt equation 2. Decomposition of input signal into sum Bs emer! \ Gi) signals. IGITAL SIGNAL PROCESSING [JNTU-HYDERABAp, Dirett Solution of input-output Equation/Ditferey, For a general system, the input-output equation j represented as, “o{n) = PUAN 1), 901-2) MO=N), x xC~ Ny, Ray Gl Here, F(_) represents the function of the quantity ‘enclosed in the brackets. For an LTI system, the input-output relation hk ‘represented as,"”* oye Dayle ns Leate DQ) et Here, a, and b, are the constant parameters independ of x(n) and y(n) defines the system Equation Qi: is called: a difference equation. Decomposition of Input Signal into Sum of Elementary Signals In this method, we will follow the below steps. Decompose the input Signal into elementary signals, Determine the response of the system to the, selected elementary signal. ‘Add the responses of the system to elementary signals using linearity property sos the obtain the total response of the system to the given input signal. ‘The detailed explanation of this method is discussed been. ‘Consider that an inpitt signal, x(n) has to’be resolved -, into a weighted sum of elementary signal feaspooch | (0). Then, x(n) = = Ln) 0 Horé, C, represents the set of amplitudes in te decomposition of signal x(n). Ifthe response of the system to the elementary sign! * (0) is y,(0) then, ‘ d= HE - me Ifthe system is relaxed and the respoinse to Cx, is4 (n): then, the total response of the input js, “= HEM), ; tit ting.equation. (3) in (5), we get, -2 won Des} = Dee exe) n= Tene, ‘ tal pote of tie sxe foie givennytso® “O (4 From equation ()) Scanned with CamScanner OwUNIT-1 (litreduction to Digial Signal Processing and Mult-rate Digital Signal Processing) 32. Define an LTIsystem and show that the output Gai, Explain the resolution of a discrete time signal Into impulses. ; : 4 Let an arbitrary signal, x(n) is to be resolved into a sum cof unit sample sequences, Consider the elementary signal x,(n) to be 340)" 8(n—#) Where, =) k—Delay of unit sample sequence. ‘Anixbitrary signal x(n) may have non-zero-values in an ” jnfinite duration. So to manage x(n), the unit impulses should also be infinite. fxn) is fmultiplied with 8(n ~&) with delay " another + sequence with value x(f) is obtained as shown in figure. ° fx(n) is multiplied with (n—m) with delay i’, another. Sequence with value x(m) will be obtained. i x00 (nm) = 30m) Bam) 2.8 Ifthe multiplication process is repeated over the delays ~ up signal x(0) signal y(n) * Se ¢ ‘e SS¢7°1SPECTRUM ALLIN-ONE JOURNAL FOM ENGINEERING STUDENTS! ‘Figure (1k: LT! System ‘We know that, the afbitrary signal x(n) is expressed as Dx®5e-H im sum of weighted impulses ic.x(n) - So, the response of system to x(n) will be the sum of weighted outputs. ie. n) = HOt) ‘HI Exws0-o} - ol) Equation (1) follows superposition property which gives. the response ofa linear system for any arbitrary sequence a(n) Itis applied to relaxed linear time variant system. For an LTL system, the response to the unit sample sequence (n) is given by impulse response h(n) as Shown in figure (2). he a(n) = HI6(@)) ~@ *Sx(a)= 8a) >}: > yo) = HE (@)} Figure 2k Response tothe Unit Sample Sequence __ The'response of the system to the delayed unit sample sequence 8(n—h)is given by,” Hn B)= HIB] : For’ time-invafiant system’h(n, k) = (n=) @) "Using the above relations, equation (1) can be written 3s, vn) = Exa nn -kh a a= x(n) + KO) Equation (4) gives the response; (n) ofthe LT system ‘5 4 finetion of the input signal, x(a) and the impulse “response h(n) called a convolution sum. Scanned with CamScanner OwTsI system, 33. Show that an fi its unit stop response: 5 Linear coavotation is defined 28, se Thayer “x(a) = u(n) the output will be, (2) = () Fora unt step input ie -_ sto) = zie sh. “We now ‘that a {see otherwise nek 2 obo Bene “Then; equation (2) ca be writen'as, sy= Pa) i s. The LSI system can be completely described oh its unit step response. Points to Remedi MA) by, tothe right for pos 1 the Jef For negative values of, onvoluti the sin seawenton jf te two signals Given to signals are»: x(n)= u(n) Ha) =a win) — + Convolution, (n)=2.: ‘The sraphical representation of wo, sens is'asstiown’ in figure (). 0) Boxe -% _From figure (I), For n=0. 0) = Samco me 40) (0) + W(A) x(-1) + AQ) aC D+. at (1) +0(0) +a2(0) + *+y(0) =1 Foran 1 x= Srane- a. “+ HO) x01) + A(1) x(0) + (2) x(-1) +. =. #10) #a(l) +a) +. WYaltaes | : Forn=2 3Q)= Pa) x(2=h) 2. #0) x2) + H{1) (1) + 12) 3(0) #43) DF = 22+ 10) + a(1) +1) +000) +... y2)=1t+ata Forn#3 50)= Di 20-5 5 ‘w+ ACO) x(3) + A(1)x(2) + h(2) x(1) + A) HO) + H(A) XA) hes 0) =ltatatta! Forn>0,4(n) =1ta+atart. an t0 facd LEGAL aieaodinne 1 Scanned with ‘Gameeanner Om