Digital Signal Processing N. G. Palan Lecturer, Department of Electronics and Telecommunication Cummins College of Engineering, Pune. Maharashtra, India [ a [lech Xr, rane Innovation ThroughoutTable of Contents 1 DSP, (E & TC) [Chapter 4 : Introduction to DSP 1-1 to 1-4] 1 12 13 14 15 Introduction 1.1.1 What is DSP 2, 14.2 System Block Diagram of Digital System Advantages of DSP . Applications of DSP Questions from University Papers .. [Chapter 2 :Signals and Systems 24 22 23 24 25 25 Introduction ... eoreer a Basic Concepts of Signals as Arrays of Values . 22.1 Continuous and Discrete Time Signals : 2.22 — Continuous Valued or Discrete Valued Signals . 223 Periodic and Non-periodic Signals... 224 — Even and.Odd Signals .. 22.5 — Energy and Power Signals 2.26 Deterministic and Random Signals’... 22.7 Multichannel and Multidimensional Signals 2.28 — Summary of Classification of Signals .. Representation of Discrete Time Signals... Standard Test Signals . 24.4 Delta or Unit Impulse Function 242 Unit Step Signal 2.43 Unit Ramp Signal 244 — Exponential Signal .. 245 Sinusoidal Waveform 2.48 — Some Other Important Signals. 2.47 Summary of Standard Signals Basic Operations on Discrete Time Signals 25.1 Time Shifting Operations 252 Time Scaling Operations 253 Amplitude Scaling Operation .. Disorete Time System 26.1 Symbols used to Represent Discrete Time Systems 262 Interconnection of Discrete Time Systems 26.3 Properties of Systems 264 Causal or Anticausal Systems (Causality Property)Table of Contents 2 DSP. (E & TC) 27 28 29 240 an 212 213 265 Stable or Unstable Systems (Stabilty Property) 266 — More Solved Problems on 0.T. Systems ~ Linear Shift Invariant (LSI) Systems .. 27.4 Representation of Discrete Time Signal as Weighted Impulses 27.2 Linear Convolution (Convolution Sum) ... . 273 Computation of Linear Convolution 27.4 — Properties of Linear Convolution Stabilty of LTI (or LSI) System Causality of LT! System Difference Equations . . 2.40.1. LTI Systems Characterized by Constant Coefficient Difference Equations .. 2.40.2 Solution of Linear Constant Coefficient Difference Equation Correlation ‘Analog to Digital (A to D) Conversion Process 2.42.1 Sampling Process ..... 2.422 Sampling Theorem 2.12.3 Frequency Spectrum of Sampled Signals 2.124 — Aliasing Effect or Foldover Error .... 2.42.5 Reconstruction Filter (low pass fiter) 2.126 — Quantization Process ... 2427 — Encoding .... . Questions from University Papers Chapter 3 : Z - transform 3.4 Introduction ... . . 3.1.1 Advantages of Z-transform .. 32 Z-Transform ree 33 _Ztransform of Standard Sequences 34 Properties of Z-transform .. 35 Inverse Z-Transform (IZT) .... 35.1 Power Series Expansion Method... 3.5.2 Partial Fraction Expansion (P.F.E.) Method 3.53 — IZT using Cauchy's Residue Theorem 36 System Function from Z Transform and Pole-Zero Plot 3.6.1 Pole-Zero Plot 3.62 System Transfer Function 3.6.3 Conditions for Causality and Stabilty in terms of Z- Transform . 37 Questions from University Papers [Chapter 4 : Digital Filter Structure 4-1 to 4-53 44 Introduction 42Table of Contents 42 43 44 45 48 47 48 49 481 — Transposed Structures Review of Analog Filters Basics of Digital Fitters ....... 43.1 Advantages of Digital Fiters 43.2 Disadvantages of Digital Fiters Describing Equation and System Transfer Function Digital Fier Structures Basic Structure for FIR System 481 Direct Form Realization of FIR System 482 — Cascade Form Realization of FIR Systems 463 Frequency Sampling Structure .. 464 Lattice Ladder Structure Basic Structures for lIR Systems .... 47.4 Direct Form Structure for lIR Systems 472 Cascade Combination of Second Order Section for IR System 47.3 Parallel Combination of Second Order Sections for lIR Systems Representation of Structures using Signal Flow Graphs Questions from University Papers (Chapter 5 : DTFT, DFT and FFT BA 52 53 54 55 56 87 58 59 5.10 Introduction Discrete Time Fourier Transform (OTFT) Fourier Transform of Standard Signals Properties. of DTFT Discrete Fourier Transform (DFT) 55.1 DFT of Standard Signals. 552 Cyclic Property of Twiddle Factor Properties of OFT Linear Fitting Using DFT Linear Filtering of Long Data Sequences 58.1 Overlap Save Method 5.8.2 Overlap Add Method Frequency Analysis of Signals Using DFT Fast Fourier Transform (FFT) Algorthms 5.401 Radie-2 FFT Algorithm 5.10.2 Radix-2 Decimation In Time (DIT) Algorithm 5.10.3° Comparison of Computational Complexity with Direct ‘Computation 5.10.4 Inplace Computation to Reduce Memory Size 5105 — Radix-2 Decimation in Frequency (DIF) FFT Algorithm 5.10.6 _ Bit Reversal Algorithm 5.10.7 Computation of Inverse DFT (IDFT) using FFT AlgorthnsTable of Contents © DSP. {E & TC) 5.11 5.12 5.13 514 5.15 5.16 517 5.10.8 Computation of Linear Filtering and Correlation using FFT Comp.tatign of DFT by Linear Filtering. Approach 5.11.1 Goéttzel Algorithm 5.11.2 Chirp-Z Algorithm Effect of Round-off Errors in FFT Algorithms Energy Density Spectrum of Aperiodic DT Signal Convergence Effect .. Relationship between Fourier Transform and Z-Transform Relationship between DFT and.ZT Questions from University Papers... Chapter 6 : Filter Design 61 62 63 64 65 66 67 Introduction _....... vs Inppulse Response of Ideal Low Pass Filler Design of.IR Filters : 63.1 Approximation of Derivatives 632 Impulse Invariance Method. . 63.3 Bilinear Transformation Method (Bl 634 — Matched Z Transform 635 Least Squares Filter Design : Induction to Analog Fiters for designing Digital Fiters 64.4 Butterworth Filter Approximation 64.2 Chebyshev Filter 643 Elliptic Filters Frequency Transformations Design of Linear Phase FIR Filters FIR Filter Design using Windows 67.1. _ Other Types of Window Functions 67.2 Gibb's Phenomenon .. 67.3 Advantages and Disadvantages of Window Method : Design of Linear Phase FIR Filters using Frequency Sampling, .....0.0:o-onn 6-70 Design of Optimum Equiripple Linear Phase FIR Filters FIR Differentiator : 7 Design of Hilbert Transforms... Comparision of Design Methods for Linear Phase FIR Fiters .. Errors due to Finite Word Length... Comparison between IIR and FIR Fitters... Questions from University Papers. [Chapter 7 : Hardware Architecture of DSP 7A to 7-21 7 72 Introduction ..... SL eLL9H92L22eoe mmTable of Contents 8 SP. (E & TC) y 72 Computer Architecture 73 Internal Architecture of ADSP 21XX Family 7.31 Arithmetic/Logic Unit (ALU) ... 7.32 Multiplier - Accurnutator . 733 Barrel Shifter... 7.34 Data Address Generators 738 Program Sequencer 7.36 Cache Memory TA Features of ADSP 21XX Family of Processors 78 Texas instruments TMS320C5X .. 7.6 Comparison of Microprocessor with DSP Processor 7.7 Questions from University Papers Chapter 8 : Analysis of Finite’ Wordlength Effects 8-1 to 8-9 a4 Introduction 82 Finite Wordlength, Effects ...... - 83 The Quantization Process and Errors ... 84 Analysis of Coefficients of Quantization Effects in FIR Filters ... 85 AD Conversion Noise Analysis .. 86 Analysis of Arithmetic Round Etfect Errors 87 Dynami Range Scaling 88 Low Sensitivity Digital Fitters. 89 Reduction of Product Round-off Errors 840 Limit Cycles in IR Fiters 8.11 Questions from University Papers . [Chapter 9 : Applications of DSP ot 92 93 94 95 96 97 98 99 9.10 911 9.12 Introduction Dualtone Multifrequency Signal Detection ‘Short term DFT Voice Privacy . Applications of D DSP in Image Processing Echo Cancellation ....... Applications of DSP in Speech .. Oversampling A/D Converter .. Oversampling D/A Converter .. Applications of DSP in Vibration Analysis ‘Subband Coding of Speech and Audio Signals Questions From University Papers * University Question Papers at to "a6 agahis table shows you the weightage to be given to this CHAPTER from Examination point of view Introduction to. DSP aa ummss Statistical Facts About This Chapter =-~_ Marks for || Marks for theory examples ee | Dec 2000 May 2001 Comments been covered. + Study rating : Marginally important. © This is an review chapter. All introductory concepts of DSP hasDSP. (E & TC) 12 Introduction to DSP 1.1 Introduction : ‘The world of science and engineering is filled with signals such as images from remote space probes, voltages generated by the heart and brain and countless other applications. 1.1.1. What is DSP ? Digital signal processing is used in a wide variety of applications. It is hard to get exact definition of DSP. Let us first look at the dictionary. meanings of these words : Digital : Operating by the use of discrete signals to represent data in the form of numbers. Sigiial : A variable parameter by which information is conveyed through an electronic circuit. Processing : To perform operations on data according to programmed instructions. This leads to‘a simple definition of DSP. Definition of DSP : DSP is defined as changing or analysing information which is measured as discrete sequences of numbers. 1.1.2 System : - System is associated with signals. . System is nothing but physical device that performs an operation on a signal. a —[ seen ou (Processing) Fig, 1.1 : Signal processing v 5 ‘As shown in Fig. 1.1, when we apply an input signal to the system, the-system will process the signal and we get the expected response, This phenomenon is called as signal processing, 1.2 Block Diagram of Digital System : Fig. 1.2 shows the basic elements of the digital system, S\ ag lg 10 Signal in digitise DsP reconstruct signal out - Fig. 1.2 : Basic elements of DSPDSP. (E & TC) 13 Introduction to DSP Most of the signals which are analog in nature are converted to digital form using A/D converter. Output of ADC is a digital signal. The digital signal processor performs signal processing operations like filtering, frequency analysis etc. The digital output signal is converted to analog signal by using D/A converter. There are two types of digital signal processings that are possible. They are : (1) Real time signal processing, @) — Non-real time signal processing. ‘When analog data is processed as it is generated, then, it is called as real time processing, Example : Processing of radar signal or speech signal, In nonsteal time processing, analog signal converted into digital form is stored in some storage media and is processed further. 1.3 Advantages of DSP : >, ASKED IN EXAM - MAY 2003!!! ] (1) Versatility : Digital systems can be reprogrammed for other applications (where programmable DSP chips are used). Moreover, digital systems can be ported to different hardware, @) Repeatability : Digital systems can be easily duplicated, These systems do not depend upon component tolerances and temperature G) Simplicity : It is easy to built any digital system as compared to an analog one. 1.4 Applications of DSP : MAY 2003 > asneo | Sr.No. | Application of DSP__| Fields where it i _| (1) _ | Image processing (a) Pattern recognition | (b) Animation | (©) Robotic vision (@_ Image enhancement (2) | Instrumentation and control | (a) Spectral analysis | (0) Noise reduction a (© Data compression (3) | Speech/Audio | @) Speech recognition | ©) Speech synthesis (©) _ Equalisation 4) | Biomedical (@) Scanners | >) ECG analysis (© _ Patient monitoringDSP. (E & TC) 1-4 Introduction to DSP Sr. No. | Application of DSP. Fields where it is used (5) | Telecommunication (@ Echo cancellation (©) Spread spectrum © Data communication © | Military (a) Sonar processing (b) Radar processing (© _ Secure communication 1.5 Questions from University Papers : Q.1 What are the advantages of ‘digital signal processing’ ? (May 2003, 4 Marks) ‘Ans. : Refer to section 1.3 for answer. Q.2_List application of DSP in Telecommunication and Biomedical engineering. May 2003, 4 Marks) ‘Ans, : Refer to section 1.4 for answer. gooThis table shows you the weightage to be given to this CHAPTER from Examination point of view Signals and Systems mms Statistical Facts About This Chapter mms Fvoe | nse [usages [mina Dec 2000 - 8 “8 May 2001 - 10 10 Dec 2001 6 37 43 May 2002 12 44 56 Dec 2002 12 8 20 May 2003 43 14 27 “Total 43 121 164 heen" : 77 || 2017 we 4 Comments : + More weightage is given to problems. ~ * Study rating : Extremely important. * Questions are based on theory as well as problemsDSP. (E & TC) 22 Signals and Systems 24 Introduction : In a communication system, the word ‘signal’ is very commonly used. Therefore we must know its exact meaning © Mathematically, signal is described as a function of one or more independent variables. © Basically it is a physical quantity. It varies with some dependent or independent variables * So the term signal is defined as “A physica: quantity which contains some information and which is function of one or more independent variables.” 2.2 Basic Concepts of Signals as Arrays of Values : ‘There are varicus types of signals. Every signal is having its own characteristic. The processing of signal mainly depends on the characteristics of that particular signal. So classification of signal ¥s necessary. Broadly the signals are classified as follows : © Continuous and discrete time signals Continuous valued and discrete valued signals Periodic and non-periodic signals Even and odd signals Energy and power signals Deterministic and random signals Multichannel and multidimensional signals. 2.2.1 Continuous and Discrete Time Signals : = Continuous signal : a askeD IN EXAM - MAY 2003 A signal of continvous amplitude or time is known as continnous signal or analog signal. This signal is having some value at every instant of time. Examples : Sinewave, cosinewave, triangular wave etc, Similarly certain electrical signals derived in proportion with physical quantity such as temperature, pressure, sound etc. are also examples of continuous signal. Some of the continuous signals are as shown in Fig, 2.1 x(t) : x(t) . 14 Continuous signal Continuous signal Fig. 2.1: Examples of continuous signals Mathematical expression : Mathematically a continuous signal (eg, sinewave) can be expressed as,DSP. (E & TC) 23 Signals and Systems x(t) = Asin(ot+0) Here A = Amplitude of signal © = Angular frequency = 2 af © = Phase shift Characteristics : ‘ © For every fix value of t, x(t) is periodic in nature, © If the frequency (3) is increased then the rate of oscillation also changes. Discrete time signal : >, ASKED IN EXAM - MAY 2003 II! } In this case the value of signal is specified only at specific time. So the signal represented at “discrete interval of time” is called as discrete time signal The discrete time signal i generated from continuous time signal by using the sampling operriion, We will study the sampling operation in detail, later in this chapter. This process is shown in Fig. 2.2. x), 4. + Continuous signal: (@) Train of sampling pulses (by 1_--Discrete signal © Fig. 2.2 : Discrete time signal ‘© Consider a continuous analog signal as shown in Fig. 2.2(a). This signal is continuous in nature from — 0 to +. © The sampling pulses are shown in Fig. 2.20). These are train of pulses. Here the samples are taken at +7, £27, £37, ... and T, is the sampling timeDSP.(E&TC) | 24 Signals and Systems © Fig. 2.2(c) shows discrete time signal. Observe that this signal takes the value, only where the sampling pulse is present. In between. the two sampling pulses the signal is absent. So this is called as discrete time signal. © In Fig. 2.2), on X-axis time () is plotted. On Y-axis the amplitude is plotted. So continuous time signal is represented by x(t). Observe Fig. 2.2(c). On X-axis index n is plotted. Here n is the number of corresponding sample, So discrete time signal is denoted by x(n). © For signal in Fig. 2.2(@), the expression is, x(t) = Acosot . and for signal shown in Fig. 2.2(¢), the expression is, x(n) = Acoson Characteristics : © Discrete time sinusoidal signals are identical when their frequencies are separated by integer multiple of 27. © If the frequency of discrete time sinusoidal is a rational number, then such signal is periodic in nature, © For the discrete time simusoidal, the highest oscillation is obtained when angular frequency o=tn 2.2.2 Continuous Valued or Discrete Valued Signals : Continuous valued signal : If the variation in the amplitude of signal is continuous thea, it is called as continuous valued signal. Such signal may be contimuous or discrete in nature. Such signals are as shown in Figs. 2.2(a) and 2.2(c). Discrete valued signal : If the variation in the amplitude of, signal is not continuous; but the signal has certain discrete amplitude levels then such signal is called as discrete valued signal. Such signal may be again continuous or discrete in nature as shown in Figs. 2.3(a) and 2.3(b). _ aqy,__Diseelg amp ovalsDSP. (E & TC) 25 Signals and Systems x(n) Fig. 2.300) : Discrete amplitude signal diserete in nature ‘As shown in Fig, 2.30), the signal is defined at all instants of time. So it is contimoys signa But it takes only certain discrete amplitude levels, The amplinde isnot continuously changing with time. So it is discrete amplitude signal confinuous i nature ‘As shown in Fig, 2.30), the signal is defined only at discrete intervals of time. So it is discrete signal. And this signal takes only certain discrete amplitude Teves So it is discrete amplitude signal discrete in nature. 2.2.3. Periodic and Non-periodic Signals : Periodic signal : [A signal which repeat itself after a fixed time period or interval is called as periodic signal ‘The periodicity of continuous ime signal can be defined mathematically a5, x(t) = X(14T) @ ‘This is called as condition of periodicity. Here Ty is called as fundamental period ‘That means afler this period the signal repeats itself. For the discrete time signal, the condition of periodicity is, x(n) =x(n+N) 2) Here number ‘N’ is the period of signal. ‘The smallest value of N for which the condition of periodicity exists is called as fundamental period. Periodic signals are shown in Figs. 2.4(a) and (b).DSP. (E & TC) 26 Signals and Systems Periodic signal xl) = Asin ot Period of repetition (T) Fig, 2.4(a) : Continuous time periodic signal Periodic signal x(n) =A sin an Petiod of repetition (N) pl on Fig, 2.4(b) : Discrete time periodic signal Non-periodic signal : A signal which does not repeat itself after a fixed time period or does not repeat at all ‘s called as non-periodic or aperiodic signal, Thus mathematical expression for non-periodic signal is, x(t) #x(t+T,) 8) and : (A) x(n) #x(n+N) Sometimes it is said that non-periodi¢ signal has a period T = oo as shown in Fig, 2.4(c). This is exponential signal having period, T = co, ‘Note that one eycie period ig, T= = Fig. 2.4(6) : Aperiodie signal heving period, T = Important condition for periodicity : * A discrete time sinusoidal signal is periodic only if its frequency (fy) is rational. That means frequency fy should be in tie form of ratio of two integers.DSP. (E & TC) 27 Signals and Systems Proof : For the discrete signal, the condition of periodicity is, x(n+N) = x(n) A) Let x(m) be the cosine wave. So it can be expressed as, x(n) = Acos(2xf,n+0) =) Here A = Amplitude and. 0 = Phase shift Now the equation of x (n-+N) can be obtained by replacing ‘n’ by ‘n+ N’ in Equation (2). x(n+N) = Acos[2af, (n+N)+8] GB) According to condition of periodicity (Equation (1)); we can equate Equations (2) and (3). Acos[2af,(n+N)+0] = Acos(2xf,n+0) Acos(2nf,n+2xf,N+8) = Acos(2af,n+0) ld) To satisfy this equation, 2afyN = Ink (3) where k is an integer ipek : 3 6) Here k and N both are integers. Thus discrete time (DT) signal is periodic if its frequency f, is rational. Solved Problems : Prob. 1: Determine whether or not the following signals are periodic. f periodic specify its i) x(n) = cos(n/8) cos(xn/8) = cos("*)- sin a xine 088) so 8] 3 Soln. : = a no x(n) = <=()os( fl ) a) ‘The standard equation of x(n) is, fundamental period, 4 3) i) The given signal is, x(n) = cos (o, n)¢os(@,n) -Q) Comparing Equations (1) and (2),\ DSP. (E & TC) 28 Signals and Systems But we have, o = 2nf 1 2nf, = vg and 2nf = w 8 1 1 f= pcm ine Here f, is the ratio of nom-integer values, So it is non-periodic. While fis the ratio of two integer values, So it is periodic. That means total signal x(n) is multiplication of non-periodic and periodic signals. So the resultant signal x (n) is non-periodic. ii) The given equation is, x(n) = co ( 3 }-sa(f oe mE 3) Observe this equation. The last term contains 5, which is the value of phase shift ©, But to check the periodicity condition we are interested in frequency and not the phase shift, The standard equation can be written as, x(n) = cos(o,n)-sin(o,n) +3 cos(o,n+6) wd) Comparing Equations (3) and (4), nn =p O78 x ant, = 5 1 ou {8 1% ™G-3 Her alleges viz ff, and f, are ratio of two integers, So the given signal is periodic. Fundamental period : Now we have, frequency (f) = f= 1 Here T,, T, and T; are called as periods of signal. So for the given signal, periods are 4, 8 and 16, Let fondamental frequency f= } then f, = fy f= yf, = 3 Ths fondamental period isT)=DSP. (E & TC) 29 Signals and Systems Prob, 2: . x 5 Given the signal x(t) = 2ens(oxt-J}r4anc 102). Js it periodic ? If so, find its period. : Soln. : ‘The given signal is, x(t) = 2oxs(éx1-§ 4s a0=t a) ‘The standard equation can be written as, x(t) = 2eos(ot-§ }easinco,0) (2) Comparing Equations (1) and (2), 0, = 6x ando, = 10% 2nf, = Gn and Inf, = 10% f= 3 andf=5 Since f, and f, are integers; the given signal is periodic in nature, Fundamental period : Here ©, = Let @ = Now T, = — Here , = Ans. Note ; The smallest value of frequency, for which the signal is periodic is called as fundamental frequency f, And the corresponding tine period (Ty=1/ fy) is called as fundamental period, And i the signal i periodic for period To then it's also periodic for 2T, 3To, 4Tp. ete 2.2.4 Even and Odd Signals : Even signals : [An even signal is also called as symmetrical signal. A continuous time (CT) signal x(t) is said 10 be symmettical or even if it satisfies the following condition :DSP. (E& TC) 240 Condition for symmetry : x(t) = x(—t) Signals and Systems .. for C.T. signal, Here x (~t) indicates that the signal is presemt for negative time period. That means x (~t) is the signal which is reflected about vertical (¥) axis. So even signals are symmetric about vertical axis or at t= 0, Examples : Figs. 2.5(a) and (b) are examples of C-T. even signals. Fig, .5(b) : Pulse, centered at origin x(t) = x(—t) A discrete time real valued signal is said to be symmetric (even) if, Condition of symmetry : x(n) = x(-n) Examples : Fig. 2.6(a) and (b) are examples of D.T. even signals. «= for D.T. signal an) . x(n) x(-n) A 324712 (@) D-T. even signal Fig 26 (b) D.T. odd signalDSP. (E & TC) 211 Signals and Systems Odd (Antisymmetric) signal : A continuous time (C:T.) signal x(t) is said to be antisynimetric or odd if it satisfies the following condition Condition of antisymmetric : x (t) |... for C7. signal Here x(—t) indicates that the signal is present for negative time period. While — x(t) indicates that the amplitude of signal negative. Thus antisymmetric signal is not symmetric about verticalaxis. Example : Figs. 2.7(a) and (b) are examples of C-T. odd signals. Fig. 2.7(b) : Pulse x(t) =—x(t) odd signal Note that amplitude of antisymmetric signal is always zero at origin. Similarly discrete time (DT. signal x (mn) is said to be antisymmetric or odd if it satisfies the following condition. Condition of antisymmetric : x(-n)=-x(n) |... For D-T. signal Example : Fig. 2.8 is example of D-T. odd signal.DSP. (E & TC) 212 Signals and Systems Even and odd parts : Any C-T. signal can be expressed as the summation of x(n) even part and odd part. x(t) = x (t)+x, (1) a) Herex, (t) = Even part ofx(t) and x,(t) = Odd part of x(t) Now even parts and odd parts can be expressed as, = FRM 4x-D] and x,(1) = Fix(ty-x(-t)] -@) Fig. 28 : D.T. odd signal Similarly for D.T. signal we can wiite, x(n) = x,(n) +x, () @ x(n) = Flx(ny4x(-n)) 6) and x,(n) = 4 [x(n)-x(-n)] 6) ‘Solved Problems : Prob. 1: Prove that any arbitrary signal can be expressed as the sum of two signals, one of which is EVEN and other is ODD, (Dee. 2001, DSP E&TO) Soln. : Let the given D.T, signal be x(n). I can be expressed a, x(n) = xn) 300) ~) Adding and subtracting x(n) = x(n) x(n) , 2-9) x(-n) eee Rearranging the terms, x(n) = [Eger] [29 The terms in the bracket satisfies the definition of even and odd signals. 2) n)ex,(n) 0)DSP. (E & TC) 243 Signals and Systems Put n= =n x(-n) = x,(-n)+x, (-n) But from the definition of even and odd signals we can write, x,(-n) = x(n) CM) Putting Equations (5) and (6) in Equation (4) we get, x(-n) = x (n)-x,(n) Adding Equations (3) and (7) we get, x(n)+x(-n) = 2x,(n) and x,(-n) = x(n)+x(-n)] ‘Now subtracting Equation (7) from Equation (3) we get, x(m)-x(=n) = 2x,(n) x(n)-x(-n 2 sng (a) (8) (6) (7) (8) (9) Prob. 2: Function x(t) is as shown in Fig, 2.9. Draw even and odd parts of x(t). Fig, 29 : Given function x(t) Solin. : Even and odd parts are calculated by using equations x, (t) = FW +x(-0) and (0-4 Ex)-x(-0) This calculation is as shown in Fig. 2.10.DSP. (E & TC) 124 Signals and systens tx) x-0} F by exo) - @ 1 a t CL. 4 4 1 1 7 © 0 1 Hex +x) poo) os _ fos, = 1 ' t + 1 1 05 (9) s (hy $ J ¥ Etat + 0) + a) = x0) Fig, 2.10DSP. (E & TC) 245 Signals and Systems 2.2.5 Energy and Power Signals : Power signals ; There are three power signals : (A) Instantaneous power B) Normalized power (© Average normalized power. (A) Instantaneous power : An instantaneous power across resistor R is given by, 2 Instantaneous power = @ Here ‘V’ is the voltage across resistor R. Let this voltage be denoted by x(t). Then Equation (1) becomes, 2 Instantaneous power = 21) -0) Similarly in terms of current we have, Instantaneous power = 7R 8) If wersay current is denoted by x(t) then we get, Instantaneous power = x7(t)R (4) (B) Normalized power : Everytime we may not know whether x(t) is a voltage signal or current signal. So to make the equation of power independent of nature of x(t); we will normalize it by putting R = 1. Thus Equations (2) and (4) becomes, Normalized instantaneous powet = x? (t) (5) (C) Average normalized power : Average normalized power is given by, T, 12 # OS woe 6) °-1/2 . P * Here magnitude of x(t) is written so this equation is also applicable, if x (t) is complex. Ty that ‘means it is for entire time period Ty. a Ty Based on above discussion we can define the power signal. 1 * Integration is taken from "to + © Ty take the average value; integration term is multiplied byDSP. (E & TC) “246 Signals and Systems Definition of Power Signal : A signal x(t) is said to be power signal if its normalized average power is non-zero and finite, Thus for the power signals the normalized average power, P is finite and non-zero. For the discrete time signals, average power P is given by, Fig, 2.11(a) In Equation (7), it is expected that N >> 1. ‘The power signal is as shown in Fig, 2.11(a). Energy signal : The total normalized energy for a “real” signal x(t) is given by, o B= fxtcyat (8) -c0 But if the signal x (t) is complex then Equation (8) is modified as, B= fix(tyPat (9) B= f ix(pPa (10) -« Definition of Energy Signal : A signal x (t) is said to be an energy signal if its normalized energy is non-zero and finite. Hence for the energy signals the total normalized energy (E) is non-zero and finite, ‘The energy of discrete time signal is denoted by E. It is given by, o YD ix(yh (ly n=-0 Fig 2.110) ‘The energy signal is as shown in Fig, 2.11(b).\ \ DSP. (E & TC) 247 Signals and Systems Note ‘Signal 7 Condition Energy signal O 1.233. O1: Let a= 3, Thus we have, x(n) aa Graphically such signal is represented as shown in Fig. 2.18(a). ofl) Q) i Fig, 2.18(@) : Rising exponential signal ‘Since the signal is exponentially growing; it is called as rising exponential signal. |DSP. (E & TC) 2-28 Case (i): When 01 Ford0 -1 fort<0 Fig. 2.21(a) : Signum function eeeDSP. (E & TC) 1.232 Signals and Systems ° This function is as shown in Fig. 2.21(a). This is an odd or antisymmetric function. x() = reot(t) 2, — Rectangular pulse : A unit rectangular pulse is centered about Y axis at t = 0. It is denoted by rect (1). Mathematically itis represented as, 1 for-1/2sts} 0 elsewhere A unit rectangular pulse is as shown in Fig. 2.210). rect () = | Fig. 2.21(b) : Unit rectangular pulse A rectangular pulse having amplitude ‘A’ and duration ‘T” is as shown in Fig, 2.21(0. It is denoted by A rect (7} Mathematically it is expressed as, | : Teel t Awa(#) = for->sts> i 0 elsewhere xt) = Ars Fig. 2.21(¢) : Rectangular pulse ~ 3. Sine function : ‘The sinc function is mathematically represented as, ~ sin(nx) sinc (x) = Here x is an independent variable. Sine function passes through zero at multiple values of x that means at x= +1, +2, +3 .... This is because sin(xx) = 0, For multiple values of x and the amplitude of “sinc” function decreases with increase in the magnitude of x. Fig. 2.21(@) 2.4.7 Summary of Standard Signals : Table 2.3 gives the summary of standard signals.DSP. (E & TC) ‘Type 2:33 D. T. Waveform C.T. Waveform Signals and. Systems Delta or unit 2) impulse 3) _Jiforn=0 80)={oforn 0 2. |, Unit step w0= {oforteo 1 for t20 12 3 4 = Jnforn20 4O)= Yoforn<0 x(n) = A sinon Table 2.3 contd.. x()=A sinot i | i |DSP. (E & TC) 234 Signals and Systems ‘Type D.T. Waveform C.T. Waveform Exponential x) When a> 1 TT t oe Growing “exponential |. When O Original sequence and x(n-K) —> Original sequence delayed by k samples. Here k is an integer. Bg. : Let the given signal be, x(n) = (2,34, 5} t ‘This sequence is as shown in Fig. 2.22). Suppose we want 10 delay this sequence by 2 samples. Then here k= 2. Ts delayed sequence will be denoted by x(n-2). This sequence willbe shifted towards Heht by 2 samples This delayed sequence is shown in Fig. 2.22(b. From Fig, 2.22(b), we can write the detayed sequence as, x (nk) = x(n=2)=(0, 0, 1, 23,4 5) ily We know that, arrow always indicate the n = 0 sample. Since delayed. sequent shifted towards right; so the first sample starts at n= ‘2, In the original sequence it is starting at n = 0. Thus to delay the sequence.D.S.P.(E & TC) 2:38 Signals and Systems 0, 1, 2, 3, 4, 5 Fig, 2.22(@) : x(m) = (1, 2), 4, 5} Fig. 2.22(b) : Delayed version x(n -k) t * Shift the diagram towards right by ‘k” samples. * Shift the arrow towards left by ‘k +* To adjust the length of sequence, add zeros, 2, Time Advance (T.A) : Time alvance operation means the signals are present before they are generated. In a real lime ‘operation, this is practically impossible. But if the signals are stored the memory of compoter before stating a particular operation then it is possible to have time advanced signal. Time advance operation is opposite to the time delay operation Consider the same ‘sequence x(n) shown in Fig. 2.22(a) x(n42) = (1 x(n) = {1,2,3, 4, 5} t If we want to advance this sequence by two samples then it is denoted by x(m+2), in this case the diagram is shifted towards eft by two samples. This sequence is shown in Fig. 2.22(c), Fig. 2.22(6) : Advanced version, x (n+k) From Fig. 2.22(c), we can write advanced sequence as, x(ntk) = x(n+2)= {1,2,3, 4,5) t 3. Folding (FD) : Folding is also called as reflection. So folding means taking the mirror image of signal. That means the signal is folded about time origin n = 0, Here independent variable ‘n’ is replaced by - n, ‘Thus if x(n) represents input signal then x (—n) represents folded input signal Consider the same input signal x(n) as shown in Fig. 2.22(a) x(n) = {1, 2,3, 4, 5} Then folded signal x (—m) is as shown in Fig. 2.22(d),DSP. (E & TC) 237 So the sequence x(-m) = 643,20) + “Thus to obtain folded version of sequence, ‘eTake the mirror image of sequence about n = 0. ‘s Write the sequence in reverse order. «Keep the position of arrow as it is. In the sequence * sequence x (~m) arrow is still at 1 4. Folding and Delay : This operation is a combination of folding and delaying operation. (a) For drawing diagrams = «To obtain delayed sequence, (B) For writing the sequence : Signals and Systems (n) amow is at 1 and in the Now let us take a quick view at the modifications used for different ‘operations. shift the original diagram towards right by “X’ samples. Te obtain advanced sequence, shift the original diagram towards left by ‘“k° samples. © ro obtain the folded version, take the mirror image of the «The delayed version of x(n) is denoted by x(n~K) Negative sign always indicates delay operation of x (7+ © The advanced version of x ( n) is denoted by x(n+k) Positive sign always indicates advanced operation of x «To obtain the folded version of x(n); replace n by — ~ Now folding and delay operation means = « Fint fold the sequence x(n); that means obtain x(-n) © Then delay the folded sequence by k samples: (n) and x (=m) x(=n) denotes the misror image of x(n). One major difference between x We know that in case of aninror imag sequence then its delayed version is Delay x(n) > x (= Folding LV Delay x(-n)-——> x But stick to the basic concepts Consider the same original seq diagram at n = 0. (») J left and right sides are reversed. Now if x(n) is original enoted by x(k), The folded sequence is denoted Wy (en), Since itis a mior image the delayed version of folded signal is denoted by x(~2+k). k) = (n-k) = x(-0+k) Delay means shift the diagram towards right by k samples. cence x(n) = (1, 2,3 4, Sas shown in Fig 2.224). The t folded sequence x (~ m) is shown in Fig. 2.22(@). Suppose We want to delay this folded sequence x(—n) by ‘2 samples then it will be denoted by x (~1-+2). This sequence is as shown in Fig. 2.22(). oremDSP. (E & Tc) 2.38 Signals and Systems “xen = 6, 4,3, 2, 1) Fig, 2.22(d) : Folded version x (—n) Fig, 2.22(6) : Delay of folded sequence x(-m42) From Fig. 2.22(¢), the sequence x(-n+2) can be written as, X(-n#2) = (5,4,3,2, 1} t 5. Folding and Advance : The advanced version of original sequence x(n) is denoted by x(n+k), The folded version of x(n) is denoted by x(~n). Since folding means minor image of Sequence; the advanced version of folded sequence is denoted by x(~n-—k is Advance 4 X(1)--——> x(n+k) Folding J Advance X(-n) ——~ x[-(n+k)] =x(-n-k) Consider the same folded sequence x(~n) shown in Fig. 2 22(d). Suppose we want to advance this sequence by “2” samples, then the advanced version is denoted by x(-m-2). Such a sequence is as shown in Fig, 2.22(0. 2.22 : Advance of folded s equence, x(—n-2) Remember the basic rule, Advancing the sequence means shifting the diagram towards leh ty ‘samples. From Fig. 22(0, we can write the sequence x (~ n= 2) 25, x(-n-2) = £5,4,3,2,1,0,0) 6| DSP. (E & TC) 2:39 Signals and Systems } Prob. 1: Prove that folding and time delaying or advancing of a signal are not commutative | operations. [Dec. 2001, DSP (E & TO] § Soln. : In this case we have to prove, Delaying (Folding) # Folding (Delaying) ) Let the input sequence be, ‘ x(a) = (410 : ‘ i This sequence is represented in Fig. 2.23(a) (8) Original sequence x(n) = (1,1,1,1) an t So by one sample xen) ={1, 1,4, 1} OP (ot) =40.11,1, 1} (b) Folded sequence x(-n) (@ Delayed sequence x(n-1) Delay by one | ae | Folding x(t) =41, 1,1, 1 mu x(-n—t) = (1,1,4, 1,0) : | 4 | | E (6) Delay (folded) sequence (© Folding (Delay) operation Fig. 2.23,DSP. (E & TC) 2-40 Signals and Systems Now consider LHS. Step I: First fold the sequence x(n). The folded sequence x (~n) is shown in Fig. 2.23(b). Step I: Delay this sequence by 1 sample. This gives delaying of folding operation. This sequence is shown in Fig. 2.23(c). Step TIL :Now consider the RH.S. First delay the sequence x (n) by one sample, This sequence is as shown in Fig, 2.23(4). Step IV : Now take the folded version of signal x(n—1). This gives folding of delaying operation. Such sequence is as shown in Fig, 2.23(€). Compare Fig, 2.23(c) and Fig, 2.23(€). Since these two diagrams are not same so; Delaying Folding) # Folding (Delaying) Hence proved. 2.5.2 Time Scaling Operations : As the name indicates, time scaling operations are related to the change in time scale. There are two types of time scaling operations, © Down scaling (Compression) * Up scaling (Expansion) 1. Down Scaling or Compression : Consider the same sequence x(n) = {1, 2, 3, 4, 5}. It is shown in Fig. 2.22(a). Let this t x(n) be input to some device which produces output y(n). Also say, y(n) = x(2n) Now from given sequence x (1) we can write, x(0)=1 x(1)=2 x(2 x(3)=4 x(4) x(5)=0.. This gives the value of x(n) for different values of n. Now we have, y(n) = x(2n) pCO x (0) aad y(1) = x(2)=3 y(2) = x(4)=5 = y(3) = x(6)=0 y(n) = ean see ee y(n) = x(2n) os Fig, 2.24(a) : Downsampling This sequence y(n) = x (2n) is represented as shown in Fig. 2.24(@).DSP. (E & TC) 2441 Signals and Systems Compare this diagram with original signal x(n) shown in Fig. 2.22(a). Since the scale is compressed this is called as down scaling operation. i 2. Up Scaling or Expansion : Consider same input sequence x(n) = {1, 2, 3, 4, 5} is applied to certain device which 4 t | produces output y (n) = (3) “Thus in this case, } q 0 i y(0) = x($}=x(0)=1 | i b ya) = {5 > No sample ' i v2) = x63 =x(1)=2 I y(3) = x{$]=x(15) > No sample | y(4) = x(f}=x@3 y(5) = x(3}=x(25) > No sample y(6) = x($}=xG3)=4 y(7) =x 7 = x (3.5) > No sample ! y(8) =x : =x(4)=5 = {1, 0,2, 0, 3, 0, 4, 0, 5} t Thus y(n) = x{4 This sequence is shown in Fig. 2.24(b). wisDSP. (E& TC) 2.42 Signals and Systems 1234567 8 Fig, 2.24(b) : Upsampling Compare this diagram with Fig. 2.22(a). Since the scale is expanded, this is called as upscaling operation. 2.5.3. Amplitude Scaling Operation : As the name indicates, in case of amplitude scaling operations, amplitude of signal is changed. Different amplitude ‘scaling operations are as follows © Upscaling (Amplification) © Downscaling (Attenuation) * Addition © Multiplication _ 1. Upscaling or Amplification : Consider same input signal x(n) shown in Fig. 2.22(a), x(n) = (1,2, 3,4, 5} t y(n) = 2x(n) “> Here x(0)=1, x(1)=2, -x(2)=3, xQ)=4 x(4)=5. Now let, y(n) = 2x(n) ‘This indicates multiplication of every amplitude by 2. y(n) = m(ad= 45, 8, 10} This sequence is shown in Fig. 2.25(a) Fig, 2.25(a) : Upscaling 2. — Downscaling or Attenuation : Downscaling or attenuation means reducing the amplitude of signal. Consider same sequence x(n) = {1, 2, 3, 4, 5} Let y(n) ‘This operation indicates the division of every amplitude by 2. y(n) = {0.5, 1, 1.5, 2, 2.5} t ‘This sequence is shown in Fig. 2.25(b).DSP. (E & TC) 2.43 Signals and Systems Fig, 2.25(b) : Downsealing 3. Addition : Consider two sequences, x,(n) = 1,101, 5 and x(n) = {2,2,0, 2, 2} + iv Let y(n) = x,(n)+x,(n) y(n) = (3,3, 0, 3, 3} t | ‘Addition operation is show in Fig, 2.26(a). | > show i | Fig. 2.26(a) : Addition of two sequence 1 4 |DSP. (E & TC) 244 Signals and Systems ( ) yst 4. Multiplication : Consider x, (n) = {1,1,0,1, 1} and x, (0) = {2,2,0, 2, 2} t t Let y(n) = x,(n)xx,(n) y(n) = {2, 2,0, 2, 2} fi Multiplication operation is as shown in Fig, 2.26(b). Fig. 2.26(b) : Multiplication of two sequences Prob. 1. - A discrete time signal is given by, x(n) © [1,4,4,4,2} t ‘Sketch the following signals : (@)x(n-2) ()x(n41) (e)x(3-n) (@)x(nyu(n-1) (e)x(n=1)-8(n-1) (f) Even samples of x(n) (g) Odd samples: of x(n)D.S.P. (E & TC) 245 Signals and Systems ‘Soln. : The given sequence is, x(n) ={1,1,1,1,2] t We know that arrow indicates the position of sample at n = 0. Thus the value of x(n) for different values of ‘n’ are as follows : x(-1)=1 x(0)=1, x(1)=1, x(2)= Land x(3)=2 This sequence is shown in Fig. 2.27(2). Fig, 2.27(a) : Given sequence x(n) (a) x(n-2): x(n-2)=(0,4,11.1.2) Here negative sign indicates delay operation. Delay operation means shift the diagram towards right. So we have to shift given sequence x(n) towards right by 2 samples. This sequence is shown in Fig. 2.27(b). (by x(n44): eR Here positive sign indicates advance operation, Advance operation means shift the diagram towards left. So we have to shift given sequence x(n) towards left by 1 sample, This sequence is shown in Fig. 2.27(c). (ce) x(3-n): This sequence can be written as x(~n+3). © Here ‘—n’ = folding operation, : «Remember that in case of folded signal, positive sign => delay operation. Delay means shift the diagram towards right. ‘ «So fold the sequence x(n) to obtain x (Mm) and then shift this diagram towards right by 3 a samples. ‘This sequence is shown in Fig. 2.27(d).DSP. (E & TC) 2.46 Signals and Systems = Delay ying 3-2-1 fot 0 7 29 4 Fig. 227(d) : Folding and delay operation, x (3 “ny (4) x(n)u(n—1): * This operation indicates multiplication of x(n) and u(n-1). * x(n) is shown in Fig, 2.27(a), + u(n) means unit step and u(n-1) => unit step delayed by 1 sample } * Draw u(n—1) and then take the multiplication, x(n) u(n—1). The multiplication takes place sample to sample, * This operation is shown in Fig. 2.27(¢) Fig, 2.27(@) : Multiplication of x(n) and u(n-1)DSP. (E & TC) 247 Signals and Systems (e) x(n-1)8(n-1): ‘© This operation indicates multiplication of x(n~1) and 8(n-1). © x(n-1) = delay of x(n) by 1 sample. © 8(n) is unit impulse and 6 (n- 1) is delayed unit impulse, © The multiplication takes place simple to sample. © This operation is shown in Fig. 2.27(f) x(n) = (141.4,2) cletaleniae Fig 2.27(f) : Multiplication of x(n—1) and 5(n-1) (f) Even samples of x(n): It means we have to find out x{2n). Given sequence is, x(n) = {1,1,1,1,2] Here x(-1)=1, x(Q)=1, x(1)=1, x(2)=1, x(3)=2 We want x (2n), Putting different values of n we get aon =-1, x(2n) = -x(=2) = 0 aon = 0, x(2n) x(0) |After this every value of x(2n) will be zero. This is because the given sequence is upto x (3 ) 2 x(2a) = (0,1, 10,0...) This sequence is shown in Fig. 2.27(g). Fig, 2.27(g) : Even samples of x(n) (9) Odd samples of x(n): It means we have to find out x (2n+ 1). We want x(2n+ 1). Putting different values of n we get, aon ==}, x(2ntl) = x(-241)= x(-1 aon 0, x(2nt1) = x(041) = x(1) aon 1, x(2+1) = x(2#1) = x(3) = 2 aon = 2, x(Qntl) = x(441) = x(5) = 0 After this every value of x (2n+1) will be zero. © x(2n41)=(1,1,2,0 DSP. (E & TC) \ 2.48 Signals and Systems aon = 1, x(2x1) = x(2) = ie t aon = 2, x(2x2) = x(4) = 0 “7 ' ¥ } f i | p I: | ‘This sequence is shown in Fig. 2.27(h). . Fig, 2.27(h) : Odd samples of x(n) Prob. 2: gkeich a discrete time signal x(n) = 27" for -2 y(n) OR y(n)=TEx(n)] Here T is called as transformation operation. Example : ‘A Giker 1g @ good example of a system. A signal containing noise is applied at the input of filter, This is an input signal to the system (filter). The filter cancels or attenuates ‘noise signal. This is the processing of signal.'A noise-free signal obtained at the output of filter is called as response of system, 2.6.1 Symbols used to Represent Discrete Time Systems : ‘The operation of discrete time system can be described by drawing a block schematic, ‘The different building blocks are used to form a complete block schematic. These building blocks are as fellows (1) Adder (2) Constant multiplier (3) Signal multiplier (4) Unit delay (5) Unit advance (1) Adder : ‘As the name indicates, adder block is used to perform addition of two input sequences. Symbol of an adder is as shown in Fig. 2.29(b). xo) -yln) = x4(n) + x2(n) x") Fig, 2.29(b) : An adderDSP. (E & TC) 2:54 Signals and Systems Here x, (m) and x(n) are two input sequences. Adder block adds x, (n) and x, (n). The corresponding output is y(n). * y(n) = x, (n)+x,(n) Sometimes instead of addition sign (*+"), sigma (5) sign is used to represent the addition, ‘The subtraction operation that “ineans, y(n) = x,(n)-x,(n) is represented as shown in Fig, 2.29(c). ~ X4(0) Qe ¥(n) = xy(n) + [-x2n)} + y(n) = x(n) —xp(n) x2in) Fig, 2.29(¢) : Subtraction operation In case of addition (or subtraction) operation, it is not required to store any one of the inputs in the memory. So this is memoryless operation. (2) Constant multiplier : This is used to multiply input sequence by some constant @). That means this operation is used to change the amplitude of input sample, This is also memoryless operation. Tt is as shown in Fig. 2,29(d). a x(n) ——+>—— y(n) = ax(n) Fig, 2.29(d) : Constant multiplier (3) Signal multiplier : As the name indicates, this block is used to multiply two input sequences. Here multiplication takes place sample to sample basis. Again this is memoryless operation. It is shown in Fig. 2.29(e). o- y(n) = x,(n) . x2(n) Fig. 2.29(¢) : Signal multiplier X4(n). x40) (4) Unit delay : This block is used to delay input sequence by one sample. It is represented as shown in Fig. 2.29). So input sequence x(n) is delayed by one sample and the ouput is y(n) = x(n~1). This indicates that yy, fz} ye) =x(n-1) the calculation of output at a particular instant, needs the previous input. Here x(n) is present input and x (n~ 1) is previous input, Fig, 2.29() : Unit delay ry Bu op 24 a Q) GB) (1)DSP. (E & TC) 2-55 Signals and Systems For example, let us say we want to calculate output at n = 0. Then putting n = 0 in the equation of y(n) we get, y(0) = x(0-1) -y(0) = ¥(-1) So to calculate output at n = 0 ie. y(0) we do not require input at n= 0, ic. x(0), But we want x( 1). Here x(—1) is previous input compared to x(0). So in this operation every time itis required to store previous input in the memory. Thus this block requires the memory; it is not memoryless operation. If we want to delay input sequence by two samples then two unit delay blocks should be connected in series as shown in Fig. 2.29(g). : wo EOE} y(n) = x(n-2) _Fig, 2.29(@) : Delay by two samples (5) Unit advance = ‘As the name indicates; this block is used to _ advance input sequence by one sample. This block is xw+—_[z} > y(n) = x(044) representéd as shown in Fig. 2.29(h). Fig, 2.29(h) : Unit advance Equation of output is, y(n) = x(n+1) To calculate the output at n= 0; put n= 0 in the equation of y(n) ~ y(0) = x(1) ‘That means to calculate present output y (0), we are not making use of present input x(0)- But we want input x (1). Here x (1) is future input corresponding to x (0 ). Thus in this operation, to calculate output y (1), a future input sample is required. This is not practically possible. So this operation is not realizable in practice 2.6.2 Interconnection of Discrete Time Systems = In case of complicated operations; it is necessary to interconnect different subsystems. There are two types of interconnections as follows : (1) Cascade or seties interconnection (2) Parallel interconnection @) Feedback interconnection (1) Cascade or series interconnection : In this case two systems are connected in series. That means output of one system. is connected to the input of other system. This is shown in Fig. 2.30(@)DSP. (E & TC) 256 Signals-and Systems Series Connection Fig, 2.30(a) : Cascade or series.interconnection Here T, and T, represents impulse responses of systems*1 and 2 respectively. Thus output of system-1 is, y(n) = TyEx(n)] This is applied as input to system 2. Thus total output is, y(n) = Ty] “ y(n) = mT[% ()] (2) Parallel connection : Two systems can be connected in parallel as shown in Fig. 2.300). Here same input, x(n) is applied to two systems. - Thus output of system-I is, y(n) = TEx(a)] and output of system-2 is, y,(n) = T,[x(n)]. Finally these two outputs are added by using adder block. Thus the total output is, y(n) = y,(n)+y,(n) ~ y(n) = TEx(n)] + Tx(a)] Fig, 2.30 (b) : Parallel interconnection (3) Feedback interconnection : As the name indicates, a feedback connection is done from ‘output fo the input. This interconnection is as shown in Fig. 2:30(c). EO Alon 1DSP. (E & TC) 287 Signals and Systems Output y(n) = Tyfetn) +Taly(n)]} Yvin) Tyo) Fig, 2.30(¢) : Feedback interconnection Hore output y(n) is applied to system-2 (7). ls output is T, Ly (nm) }. This signal added with input x (n),'So the output of adder is x(1)-+T)[y (m)]. This signal is applied to system-1 (7, ). Thus the total output is, y(n) = TLD) ‘Summary of symbols used for discrete time systems : Table 2.4 shows different symbols used to tepresent discrete time systems. i: Table 2.4 sr] Name Symbol ‘Significance ‘Adder panes eee It is memoryless operation x4(0) C—rvnesin ~) Constant a It is memoryless operation multiplier x(t) ——>—— yr) = axin) Signal x4) —— It is memoryless operation multiplier o- a) = 400) 209) Xn) ——— ~ - - = Unit delay It requires memory to x0} {z} ylr) =x(0-1) store the previous input Unit . : "This operation is not advance xi--+—[z > ytn) =x(01) realizable in practiceDSP. (E & TC) 258 Signals and Systems 2.6.3 Properties of Systems : Properties of systems are with respect fo inpit and output signal. For thevsimplicity we will Consider the system having single input and single output, But the same explanation is valid for the system having multiple inputs and multiple outputs: Important properties of systems are as follows : 1) Dynamic or static nature ‘Time variance Linearity Causality Stability __ Now based on the properties, the systems can oe as follows : Classification of system a a Statio ‘Time varient Linear Causal Stable or or or or or Dynamic ‘Time invarient Nonlinear Noncausal Unstable Static or Dynamic Systems (Dynamicity Property) : Static systems : Definition : It is a system in which output at any instant of time depends on input sample at the same time, Example : @ i) y(n) = 5x(n) - Here 5 is constant which multiplies input x(n). But output at n instant that means y (n) depends on input at the same (n"") time instant x (1). So this is static system. y(n) = x(n) +5x(n)+10 Here ‘also output at ni instant, y (in) depends on the inpui at né™ instant, So this is static system, . Significance : Observe the input-output relations of stitic system. Output does not depend on delayed [x(n-k)] or advanced [x(n+k)] input signals, It depends only on present (n!™) input signal. If output depends on delayed input signals then such signals should be stored in memory to calculate the output at n'™ instant, This is not required in static systems. Thus for static systems, memory is not required. So static systems are memoryless systems, ii) Dynamic systems : Definition ; It is a system in which output at any instant of time depends on input sample at the same time as well as at other times.(8 TC) 2.59 Signals and Systems Here other times means, other than the present time instant. It may be past time or future time, Note that if x (1) represents input signal at present instant then, (@ x(n—k); that means delayed input signal is called as past signal. (b) x(n+k); that means advanced input signal is called as future signal. Thus in dynamic systems, output depends on present input as well as past or future inputs. Examples : @ y(n) =x(n)+5x(n-1) Here output at n'* instant depends on input at ni™ instant, x(n) as well as (m1) instant x(n-1) is previous (past) sample. So the system is dynamic. (i) y(n) = 3x(n+2)4x(n) Here x(n+2) indicates advanced version of input sample that means it is future sample; so this is dynamic system. Significance Observe input-output relations of dynamic system. Since output depends on past or future input sample; we need a memory to store such samples, Thus dynamic system has a memory. Note : Just by observing input-output relations; we can check whether the system is static or dynamic. if output y(n) contains delayed or advanced terms of input then the system is dynamic, otherwise it is static system. - Prob. 1: Check whether the following systems are static or dynamic (1) y(n) =nx(n), (2) y(n)=nx(n)+bx2(n), (3) y(n) = 3x(n) +5, (4) y(n) = x(n), (5) y(n) = x(n) cos aqn, (6) y(n) =3|x(n)] 2 (7) y(n) =x(n)-3x(n~2) (8) y(n)= > x(n-k) k=0 (9) y(n) =x(-n) (10) y(n) = Ae) Soln. : Q y(n)=nx(n) It is static system because output depends on present input only. 2 y(n)=nx(n)+bx"(n) Note that x2(n) means, square the amplitude of input signal. So it is not related to time shifting. (It is not x(n). Here ‘b’ is some constant. Since the equation does not contain any advanced or delayed input term; it is static system,DSP. (E & TC) ~ 2-60 Signals and Systems 3) 4) 5) 9 8) %» 10) 2) time, y(n) =3x(n)+5 It is static system because output, y (n) depends only on present input x (1), y(n) = x(a) It is dynamic system, because at nt instant, system-expects future inputs (n?). For example, at n= Ly(1)=x(1?)=x(1) at n= 2,y(2)=x(2)=x(4) ‘Thus at n = 2, system requires future input, x (4). So it is dynamic system. y(n) = x(n)cos myn It is static system because output depends on present input only. y(n) =3]x(n)| It is static system because output depends on present input only. y(n) = x(n)-3x(n-2) It is dynamic system because output requires delayed input 2 y(n)= x(n-k) k=0 First we will expand this summation. y(n) = x(n)+x(n-1)+x(n-2) Since output contains delayed input terms; it is dynamic system. y(m)=x(-n) We know that x (—n) means folded input signal. Now we will put different values of n at n= 0,y(0)=x(-0)=x(0) at n= Ly(1)=x(-1) at n= 2y(2)=x(-2) Here x(~1), x(-2)... etc. indicates past input sample. So the system is dynamic. y(n = Ae) This is static system because past or future input samples are not present. Time variant or time invariant systems (Time variance property) : Definition : A system is time invariant if its input-output characteristics do not change with fiDSP. (E & TC) 261 Signals and Systems Explanation : A time invariant system means its input-output characteristics are not changing with time shifting. Let us consider we are applying input signal x(n) to the system as shown in Fig. 2.31(@). It produces output y(n). Now delay input by ‘k’ samples, That means input becomes x(n=k). Apply this delayed input to the same system as shown in Fig. 231(b). Let us say the system now gives output y (nk). Then this system is called as time or shift invariant system. x(n) + yin) x(n-4) 1+ ying @ () Fig, 2.31 : Time invariant system Observe that initially the output is y(n) with input x(n). When input is delayed by ‘k’ samples then the output is also delayed by same ‘k” samples. Thus the input-output characteristics are not changed. So the system is time invariant, Theorem : A system is time invariant if and only if T T x(n) ——> y(n) implies that x(n-k) —> y(n-k) How to determine whether the system is time invariant or not ? To check whether the system is time invariant (TIV) or time variant (TV) follow the following steps : Step 1: Delay the input x(n) by ‘k’ units that means x(n-k). Denote the corresponding output by y(n, ®). T That means x(n-k) > y(n,k) Step I: In the given equation of system y(n) replace ‘n’ by ‘n-k’ throughout, This output is y(n-k). Step Il: If y (n, k) = y(n-k) then system is time invariant (TIV) and if y (n, k) # y(n-k) then system is time variant (TV). Solved Problems : Prob. 1; Determine whether the following systems is time invariant or not. y(n) =x(n)-x(n-1) ‘De 2002, DSP (E&TC)] Soin. : Step I: Delay the input by ‘k’ samples and denote corresponding output by y (n, k) y (nk) = x(n-k)-x(n-k-1) wl)DSP: (E & TC) 2.62 Signals and Systems Step IL : Replace ‘n’ by ‘n—k’ throughout the given equation, y(n=k) = x(n-k)-x(n-k=1) (2) Step III : Compare Equations (1) and (2) Here y(n, k) = y(n-k), Thus the system is time invariant (TTV)- Prob, 2: Determine whether the following systems is time invariant or not TEx(n)] = 8 Soln. : We’ know that T[x(n)]-means x(n) is passed through the system to produce output y(n). Thus given equation'can be written as, T y(n) = e() that means x(n) —— y(n) = eX") Step 1: Delay input by k samples and denote this output by y (a, K) ¥ (ak) = ok Ca") Al) Step IT: Replace ‘n’ by ‘n—X’ throughout the given equation, y(n-k) = (nk) (2) Step III : Compare Equations (1) and (2) Since y (n, k) = y (n—k); the system is time invariant (TIV), Prob. 3: Determine whether the following systems is time invariant or not. T[x(n)] = ax(n)46 Soln. : The given equation can be written as, y(n) = ax(n)+6 Step 1: Delay inpnt by k samples and denote this output by y (n, k) y (ak) = ax(n-k)+6 (ly Step Il: Replace n by n—k throughout the equation, y(n-k) = ax(n-k)+6 -Q) Step III : Compare Equations (1) and (2) Here y (ak) = y(n—k). The system is time invariant (TIV). Prob. 4: Determine whether the following systems is time invariant or not. y(n) = cosx(n) [ec, 2001 DSP E&TO] Soln. : . Step 1: Delay the input by ‘k’ units and denote this output by y (n, k) y (ak) = cosx(n-k) fl) Step I: Replace ‘n’ by ‘n—K’ throughout the equation, y(n-k) = cosx(n-k) 0) Ste Ste Pre Sol Stey pres Stey Stey Pro Soll Step Step StepD.SP.(E.& TC), 2-63 Signals and Systems Step IIT : Compare Equations (1) and (2) Since y (n, k) = y(n—k); the system is time invariant (TIV). Prob; §: Determine whether the following systems is time invariant or not y(n) =|x(n)] [Dec. 2001, DSP (E&TC)] Soln. Step I: Delay input by °k” samples and denote this output by y (n, k) y@% = [x(n-k)] wlll) Step IL: Replace ‘n’ by ‘n—k’ throughout the equation y(n-k) = [x(n-k)| (2) Step HII : Compare Equations (1) and (2) Sine y(n, k) = y(n-k); the system is time invariant (TIV). Prob. 6 : Determine whether the following systetns is time invariant or not y(n)=x(n)+nx(n-1) Soln. Step : Delay input by ‘k’ units and denote this output by y (n, ) y(n,k) = x(n-k)+nx(n-k~1) ) Note that only input x(n) is delayed. That means in the given equation wherever x(n) is present, replace it by x(n—k). You should not replace “‘n’ present in the second term by n-k. ‘That means in this step do not write the equation as, yk) = x€n-k)+(n-k)x(n-k-1) This is not valid since we are delaying input, x(n) and not only n, Step Il: Replace ‘n’ by ‘n-k’ throughout the equation. y(m-k) = x(n-k)+(n-k)x(n=k=1) Q) Step IIT : Compare Equations (1) and (2) Since y (nk) # y(n—k); the system is time variant (TV). Prob. 7: Determine whether the following systems is time invariant or not. y(ny=x( i) [Dec. 2002, DSP (E&TC)] Soln. : 7 Step I: Delay input and denote this output by y (n, k) y@,® = x[(n-k)'] () Step Il: Replace n by n—k throughout the equation y(n-k) = x[(n-k)] @) Step HII : Compare Equations (1) and (2) Since y (n,k) = y(n-k); the system is time invariant (TIV).DSP. (E & TC) 2:64 Signals and Systems Prob. 8: Determine whether the following systems is time invariant or not. y(n) =x(2n) Soln. : Step I: Delay input by k units and denote this output by y (n, k) yk = x[2(n-k)] ww) Step If: Replace ‘n’ by ‘n—k’ throughout the equation y(n-k) = x[2(n-k)] (2) Step III : Compare Equations (1) and (2) Since y(n, k) = y(n-k); the system is time invariant (TIV). a 9: Determine whether the following systems is time invariant or not. y(n) =y(n-4)+x(n-4) Soln. : Note that the term y (n~4) denotes the delayed output. It also indicates that the feedback {s taken from the output y(n) and it is delayed by 4 units. Step'I: Delay input by k units and denote this output by y (n, 1) y(k) > y(n-4)+x(n-k-4) fl) Step Tl: Replace ‘n’ by ‘n—k’ throughout the equations y(n-k) = y(n-k-4)+x(n-k-4) QQ) Step IIL : Compare Equations (1) and (2) Since y (nk) # y(n-k); the system is time variant (TV). Prob. 10: Determine whether the foflowing systems is time invariant or not. v y(n) =x(-9) Soln. : Step I: Delay input by ‘K’ units and denote this output by y (a, k) y@k = x(-n-k) (a) Note that only ‘n’ should not be replaced by nk. “This is because we want to delay the whole input'x(n) by ‘K’ units. Thus x (~n) should be delayed by k samples, Step IL: Replace ‘n’ by ‘n-k’ throughout the equation y(n-k) = x[-(n-k)] =x(-n+k) 0) Step HI : Compare Equations (1) and (2) Since y (n,K) # y(n-k); the system time variant (TV) system. Prob. 11 : Show the following systems are time varying. () y(my=hx(n) Gi) y(n) = n(n) [May 2003, DSP (E&TC)} Soln. : aenetenees () Given, y(n) = nx(n) Step I: Delay input by k units and denote this output by y (n,k). y(a,k) = nx(n-k) Gi Ss St St 3)DSP. (E & TC) 2-65 Signals and Systems Step IV: Replace n by nk throughout the equation. y(n-k) = (n-k)x(n-k) Step TIL : Since y (n,k) #¥ (n-k); the system is time varient Gi) Given y (n) = 0x7 (0) Step 1: Delay input by K units and denote this output by ¥ (ak). y (mk) = atx(a-k)P Step I: Replace n by n—k throughout the equation. y(n-k) = n-kx (n-k) Step JIM : Since y (n,k) #y (nk) the system is time varient. 3) Linear or nonlinear systems (Linearity property) = A tinear system is a system which follows superposition principle. Consider a system as shown in Fig, 2.32(a), whose response is Its input is x (") ‘and it produces output y(n). x(0); -y(n) = T x(n) Fig, 2.32(a) : A system with response ’” Let input x, (1) produces output y, (m) and input x5 (nm) produces output y, (1). Consider two sibirary constanis-a, and a,, Then simply wwe can say tht input a, x, (1) produces output ayy, (1) and input a, %, (1) produces output ay, (M). Now we can write the theorem for linear systems a5 follows. ‘Theorem = ‘A ystem is said to be linear if whe combined response of &Xy (n) and a, x, (1) is equal to the addition of individual responses. ‘That means, Tla,x, (n)+a,x,(2)] = a TL, (n)} +a TI (9) wl) Equation (1) is called as superposition theorem ‘The graphical representation of L.H.S. and RUS. of Equation (1) is as shown in Fig, 2.32) and Fig, 2 32(6) respectively. y(n) =T fayey(n) + axe) xg) 2 Fig, 2.32(b) : LHS. of equation (1)DSP. (E.& TC) Signals and Systems Vn) = ayT fxy(n)] + agT Expin)) Fig, 232(0) : RHS, of equation (1) Important characteristic : An important characteristics of linear system is that when input is zero then it produces zero output. But if zero input gives some value at the output (non-zero) then it is non-linear system. If this condition is satisfied then apply the superposition theorem to check whether the system is linear oF not. How to decide whether the ‘system is linear or not’? Step 1: Apply zero input and check the output If output is zero then system is linear If this step is satisfied then follow the remaining steps, Step I Apply individual input to the system and note down Corresponding outputs. Then add all Outputs. Denote this addition by y'(m). This is RH.S. of Equation (1) as shown in _ Fig. 2:32(6). Step OT: Coniine all inputs, apply it tothe system and find out oiitput y"(n). This is LHS, of Equation (1) as shown in Fig. 2.32(b) Step TV rey’) = y" (nm) then the system is linear otherwise it is non-linear system, Note : (1) For the simplicity here we have assumed arbitrary constants a, = a, = 1 2 (2) if step | is satisfied then check remaining steps. But if step | is not satisfied then at this stage only we can say system is non-linear, Solved Problems : Q. Determine whether the following systems are linear or not y(n) = nx(n) [May 2003, DSP E&TO)] Step 1: When input x(n) is zero then output y(n) = n0 = 0, ‘Thus the system is linear. Now frst step is satisfied so we will check remaining steps for linearity. Step II: Let there are two inputs x, (n) and x, (mn). T ‘The given equation of output is y(n) = nx(n) that means x(n) —-—> nx(n). So the Funetion of system is to multiply input x(n) by ‘n’. Now for two inputs X,(m) and x, (n) the Ste Ste Pr So Ste Ste Ste2S8P. (E & TC) 287 Signals and Systems! orresponding outputs are y, (mn) and y, (m). T x, (1) —= y, (n) = nx, (n) T and x,(n) ——> y,(n) = nx, (n) Now add these output to get y'(n) y(n) = y, (2) +y, (n) = mx, (n) +x, (n) y (a) = n[x,(n)+x,(n)] () Step III : Now add inputs x, (n) and x, (m) and then pass it through the system. T (x, (2) +x, (2)] > y" (a) = Ts, (2) +x, (0) ] = 2, (0) +, (1)] We know that the function of system is to multiply input by ‘n’. Here [x, (n)+x, (m)] acts as one input to the system. So the corresponding output is, y(n) = nlx, (n)#x,(0)] 0) Step IV : Compare Equations (1) and (2) Since y'(n) = y"(n); the system is linear. Prob. 2: x(n) TIx(n) Soln. : We can write the given equation as, y(n) = eC") Step 1: ¢ input x(n) is zero then, output y(n) = e° = 1. Thus the system is non-linear. But we ‘will also check the system for superposition principle. Step Il: Here the function of system is to take exponential of x(n). Now let there ate two inputs, x, (nm) and x, (n) a x,(n) = y, (a) = 8") T and x, (n) —+ y,(n) =e ‘Add these output to get y' (n) y(n) = ch 40) ) Step II :First add input x, (m) and x, (m) and then apply it to the system. We know that theD.S.P. (E & TC) 2-68 Signals and Systems. system produces exponential of input. T [x, (n)#x,(n)] —> y"(n) = of¥(n)#%,(0)] y'(ay = ol (44 (91 y(n) = bh). g Q) Step IV : Compare Equations (1) and (2) Since y(n) # y" (n); the system is non-linear. Prob. Soin, : The given equation can be written as, y(n) = ax(n)+6 T{x(n)} = ax(n) +6. Step I: If input x(n) is zero then, y(n) = 046 6 y(n) = 6 Step I Consider two inputs x, (m) and x, (n). We will apply these two inputs separately to the system. T x,(n) —>y,(n) = ax, (1) +6 fr and x,(n) ——-> y,(m) = ax, (0) +6 ‘Add these outputs to get y’ (n) Note that the given equation contains the coefficients ‘a’ and 6. So to check the linearity we will add the outputs as follows y(n) = ayy, (n)+a,y_(0) 1 y(n) = a, [ax,(n)+6]+a,[ax,(n)+6] wal) Here ‘a,’ and ‘a,’ are arbitrary constants used to check the linearity. Step DI : We will add two inputs and then we will apply this signal to the system. Again we will use arbitrary constants a, and a, T [a,x, (m) +a, x, (m)] ——> ala, x, (m) +a) x, (m)] +6 y(n) = alas, (n) +ax,(n)] +6 @) Note that in this case the function of system is to multiply input by constant ‘a’ and then add Pr Sc Sti Sti St StDSP. (E & TC) 269 Signals and Systems 6, The term a, x, (n) +a, x, (m) acts as combined input signal. Thus the output y” (11) is obtained as given by Equation (2). Step IV : Compare Equations (1) and (2) Since y'(n) # y"(n); the system is non-linear. Note : Whenever the given equation contains some constants and the equation contains addition of two or more terms; then to check the linearity, use arbitrary constants a, and @,. In o! er cases we are assuming a, = a) = 1 Prob. 4: Soln. : Step 1: If input x(n) is zero then, y(n) = cos (0) y(n) =1 So the system is non-linear . It is not necessary to apply remaining steps but we will check the syste for superposition theorem. Step If: Consider two inputs x, (n) and x, (1). Apply these two inputs separately to the system. Observe that the function of system is to take ‘cos’ of input signal. T x,(n) > y, (1) = 008% (2) y(n) = cosx(n) [May 2001, Dec. 2001, DSP (E&TC)) T and x,(n) —>y,(n) = cos, (n) Now add two outputs to get y(n) y(n) = y,(n)+y,(0) y(n) = cosx, (n)+c0sx, (n) (l) Step THT : Add two inputs and pass it through the system. T [x (1) +x, (2) > 008 x) (0) #1) y'(n) = cos[x, (m)+x,(n)] 2) Use trignometric identity, cos[A+B] = cos AcosB+sin A sinB “Thus Equation (2) becomes, y(n) = cosx, (nm) cos x, (n) +sin x, (1) sinx, (n) -) Step IV : Compare Equations (1) and (2) Since y'(n) # y" (mn); the system is non-tinear.DSP: (E & TC) 2-70 Signals and Systems Prob. 5: y(n) =|x(n)} [May 2001, Dec. 2001, DSP (E&TO)| Soin. : Step 1: Apply zero input to the system that means x(n) = 0. y(n) = [x(n)]=0 ‘Thus step 1 is satisfied so we will check the remaining steps. Step II: Consider two inputs x, (n) and x, (1) and apply it to the system. T x (2) —>y,(n)= |x, (n)] T and x(n) —>y,(n)=]x,(n)| Add these two outputs to get y' (n) y(n) = y,(n)+y,(n) ¥ (a) = [x (n)]+1x,(a)] A) ‘Step III : Add two inputs and apply it to the system. Observe that the function of system is to take ‘magnitude of input signal, T [x (n) +x, (2)] > |x, (2) +x, (n)| Yn) = |x (2) +x, (0) a) Step IV : Compare Equations (1) and (2) Since y'(n) # y"(n); the system is nonlinear, Prob. 6 : Explain the following system with respect fo linearity property y(n) =x(n)+nx(n41) Soin, : Step 1: Apply input x(n) = 0 to the system. The output y(n) = 0 s0 we will check remaining conditions, Step II: Consider two inputs x, (2) and x, (n). Apply two inputs separately to the system. T x(n) > y, (1) = x (n) +x, (+1) T : and x(n) —- y,(n) =x, (n)+nx,(n+1) Add these two outputs to get y' (n)agen am DSP. (E &'TC) 271, Signals and Systems Y(a)= y(n) 4y, (n) y(n) = %(n) tnx, (141) +x) (m) +x, (41) y(n) = x, (1) +x, (m) +n[x, (n+ 1) +x, (n+1)] (1) Step IIT : Combine two inputs and apply it to the system, T Tx, (1) +x, (n)] —> [x (n) +x, (n)] +n [x (n+1) +x, (n+1)] y'(n) = [4 (0) +x, (0) ] +0 [x, (141) +x, (04 1)] ..(2) Step IV : Compare Equations (1) and (2) Since y'(n) = y"(n); the system is linear, Prob. 7: Explain the following system with respect to linearity property y(n) = x(2a) [Dec. 2002, DSP (E&TO)} ‘When input is zero, output y(n) is also zero. So we will check remaining conditions, Step I: Consider two separate inputs x, (m) and x, (1). Apply it to the system. T X(m) —> y,(n) = x, (20) T and x(n) —>y, (n) = x, Qn) Add y, (1) and y, (n) to get y'(n) ¥(m) = x (2n)+x, Qn) @ Step TIT : Combine two inputs and apply it to the system T x, (1) +, (n)] > x, (2n) +x, (0) y"(a) = x,(2n) +x, (20) +-Q) Step IV : Compare Equations (1) and (2) Since y'(n) = y"(n); the system is linear, Prob. 8: Find whether the following system is linear y(n) =x(-n42) [Dec. 2002, DSP (E&TO)] Solin. : Step 1: Apply zero input to the system. It produces zero output. So we will check the remaining conditions.\ DSP. (E & TC) 272 Signals and Systems Step TL Consider two inputs x, (n) and x,(n). Apply these inputs separately to the system. x,(n) —> y, (n) =x (-042) .T and x(n) —>y, (0) = (-042) ‘Add these two outputs to get y(n) y(n) = y,(n)+y, (0) y(n) = x,(-n+2)+x,(-n+2) (I) Step III : Add two inputs and apply it to the system, T x, (1) +x, (2) > x (-042) +x (-942) y(n) = x,(-042)+x,(-m4+2) 2) Step TV = since y'(n) = y" (1m); the system is linear, Prob. 9: y(n) =x(nt) Soin. Step 1: For zero input output is zero. Step II: Consider there are two inputs x,(n) and x,(n). Apply these inputs separately to the system. T x(n) 9 y, (n) =X, (0) T and x(n) —>y, (2) = (0) ‘Add these outputs to get y' (1) ~ y(n) = y,(n)+y,(n) 2 y(n) = x (0) +x, (07) wl) Step III : Add two inputs and apply it to the system. uy bx, (2) +, (2) 9 Fx, (7) 4407) y(n) = x, (1) +x, (9) w=) \2 St Pr Ss st StDSP. (E & TC) 273 Signals and Systems Note that the function of system it to Square n™ sample, It is not squaring of input that means x(n), Step IV : Compare Equations (1) and (2) Since y(n) = y’ (n); the system is linear. Prob. 10: y(ny=nx?(n) [MAY 2003, DSP (E&TC)] Soln. : Step I: When input x(n) is zero; output is zero. Step I: Consider two inputs x, (1) and x, (1). Apply each input separately to the system. T x(n) y(n) = 0x (0) a 2 and x,(n) ——>y, (0) = 0x, (0) ‘Add these outputs to get y' (1) , 2.2 y(n) = af} (ay+x3(0)] A) Step III : Combine two inputs and apply it to the system. T [x (n) +x, (n)] — ax, (2) 4,00) P y'(n) = al, (a)+x, (2)? +) Step IV : Compare Equations (1) and (2) Since y'(m) # y"(m); the system is non-linear. 2.6.4 Causal or Anticausal Systems (Causality Property) : A) Causal systems : Definition : {A system is said to be causal system if output at any instant of time depends only on present and past inputs, But the output does not depend on future inputs. Examples: "The output of system depends on present and past inputs that means output, y(n) is @ function of x(n), x(n-1),x(m~2) .. ete, some examples of causal systems are > @ y(m)=x(n)+x(n-1) @ y(n) =3x(n) practically reliable. That means such system can be implemented practically. Generally al real time systems are causal systems; because in real time applications only present and past samples ae present.DSP. (E.& TC). 274 Signals and Systems Since, future samples are not present; causal system is memoryless system. B) Anticausal or non-causal systems : Definition : A. system is said to be anticausal system if its output depends’ not only-on Present and past inputs but also on future inputs. Examples : In this case, output y (1) is function of x(n), x(n-1),x(n—2) .... etc. as well as it.is function of x(m+1),x(n-+2) ... etc, some examples of non-causal systems are : @ y(n) =x(n)+x(n41) Gi) y(n) = Bx(n+2) Gil) y(n) = x(n) + ax(n41) Significance : Since non-causal system contains future samples; a non-causal system is practically not realizable, That means in practical cases it is not possible to implement a nor-cavsal system, © But if the signals are stored in the memory and at a later time they are used by a system then such signals are treated -as advanced or future signal. Because such signals are’ already present, before the system has started its operation. In such cases it is possible to implement a non-causal system, * Some practical examples of non-causal systems are as follows : (@® Population growth (ii) Weather forecasting (iii) Planning commission etc. Solved Problems : Prob. 1: Determine if the systems described by following equations are causal or non-causal. () y(m)=x(m)+x(n-2) i) Thx¢ny] =e") (ii) TEx(n)] = ax(n)+6 (i) y(n) = cosx(n) yon) =Ix(n)] (i) y(n) =x(n)+mx(n-4) (vii) y(n (n)+mx(n+1), (vil) y(n) =x(2n) (&) y(n) =x(=n+2) (%) y(n) = x(n?) Soln. : @. y(n)=x(n)+x(n-2) It is a causal system because output depends only on present input i.e. x(m) and past input ie. x(n-2), @ Thx(nyp=er™ ‘The given equation is, y(ny = ol") This is causal system since output depends on present input, x(n). Gi) T[x(m)]} = ax(ny +6 ‘The given equation is, @D.SP. (E & TC) 26 Signals and Systems y(n) = ax(n)+6 This is causal system because output depends on present input x(n). There is no future input signal, Gv) y(n) = cosx(n) [Dee. 2001, DSP E&TC)] This is a causal system because the function of system is to obtain cosine value of present input, @) y(a)=1x(n)| [May 2001, Dee, 2001, DSP (E&TO)] This is a causal system because output depends on present input. (i) y(a)=x(n)tnx(n-1) This is a causal system because output depends on present and past input but not on the future input. il) y(n) =x(n)+nx(n+1). Here output depends on future input ie. x (n+1). So this is non-causal system. (vill) y(n) =x(2n) This is non-causal system because output expects future input: This can be verified by putting different values of n. at n= 0 => y(0)=x(0) Here present output excepts present input only. atn=1=> y(1)=x(2) Here present output i.e. at n= 1, expects future value of input ie. x(2) (y(n) =x(-n+2) [May 2002, NSP (K&TO)] This is non-causal system, This is because at n= - 1 we get, y(-1) = x[-(-1) +2 [1+2]=x(3) Thus present output at n = ~ 1, expects future input ie. x (3). @® y(ayex(a?) This is non-causal system. This is because at n= 2 we get, y(2) = x(P)=x(4). ‘Thus present ontput at n= 2, expects future input x (4). 2.6.5 Stable or Unstable Systems (Stability Property) : To define stability of a system we will use the term ‘BEBO’. It stands for Bounded Input Bounded Output, The meaning of word ‘bounded’ is some finite vah ‘bounded input means input signal is having some finite value. ie. input signal is not infinite. Similarly bounded output ‘means, the output signal attains some finite value i.e. the output is not reaching to infinite level, Definition of stable system : An initially relaxed system is BIBO stable if and only if everyDSP. (E & TC) 2-78 Signals and Systems ‘bounded input produces bounded output. Here a relaxed system means when input to the system is zero then the output of system is also zero. Mathematical Representation : Let us consider some finite number M, whose value is less than infinity. That means M, y(2) = DY u(k)=u(0)+u(1)+u(2)= 1414153 BE2U k=0 S si step 1 Forn=5=>y(5) = D u(k)=u(0)+u(1)+u(2)+u(3)+u(4)+u(5) k=0 (i) ; = 1l4lel+l4i4l Step 1 = 6 Step T 10 Forn=10=>y(10) = >> u(k)DSP. (E & TC) 279 Signals and Systems = u(0)+u(1)+u(2)+u(3)+u(4)+u(5)#u(6) +u(7)+u(8)+u(9)+u(10) =u It is clear that as n —> <0, y(n) —> ©, even if input u(k) is bounded. Thus the system is unstable. (2) y(n) = sgn[x(n)] ‘The signum (sgn) function is defined as, 1 forn20 senlx(n)] SE forn<0 Itis shown in Fig, 234 sgn[x(n)} )— Dynamicity : Observe the given equation. The output contains only present input x(n). Thus the given system is static. (i) Time invariance : Step I: Delay the input x(n) by ‘k’ samples and apply it to the system. The corresponding output is y(n,k). y(n,k) = sen[x(n-k)] Al) Step Il: Replace ‘n’ by n—k throughout the system and denote the corresponding output by y(n-k), y(n-k) = Agn[x(n-k)] 2) Step III : Compare Equations (1) and (2). Since y(n,k) = y(n—k); the system is time invarient. (ii) Linearity : Step 1: For zero input, and zero output. Step IL: Consider that there are two inputs x, (m) and x, (1). Apply these input separately to the system,DSP. (E & TC) . 2-80 Signals and Systems = x(n) —> y,(n) = sen [x,(n)] T and x,(n) — y,(n)=sgn[x,(n)] ‘Add these two outputs to get y(n) } ¥ (a) = y,(n)+y,(n) ¥ (n) = sgn[x, (n)}+sen[x, (2)] @ Step IIL : Combine two inputs x, (n) and x, (n) and then apply it to the system. a2 2 Lx, (2) +x, (n)] > sen [x, (1) +x (0)] - y(n) = sgn[x, (n)+x,(n)] Q) Step IV : Compare Equations (1) and (2). In Equation (1), sotyconn = {1 Riney samaent cote | teeny While in Equation (2), : sen [x (n) +x ()] = i. ee Thus Equations (1) and (2) are not same. So given system is non-linear. (e) Causality : Observe the given equation of y(n). The output depends only on present input; it does not contain any future input term, So given system is causal. = () Stability : * From Fig. 2.34; we can say that the maximum value of y(n) = £1 for any value of n, This is defintely tess than infinity. Thus for bounded input system produces bounded output. Thus it is a stable system. (3) y(n) = Trunc [x(n] Here ‘Trane’ stands for truncation, So if x(n) = 3.4354 then, Trune [x(n)] = 3 @ — Dynamicity = ‘Truncation operation contains only present samples, not the past sample. So memory is not required. Thus it is a static system. (i) Time Invarience : If imput is delayed by some factor then truncation is not affected. Because this operation is related to amplitude change and not the change in time period. So this is time invarient system. ov discr syste impo+E & TC) 281 Signals and Systems Linearity : Consider two inputs x, (m) and x, (1). Let x,(n) = 3.4567 and x, (n) = 4.3456 Now y,(n) = Trunc [x, (n)] = Trune [3.4567] = 3 and y,(n) = Trune [x,(n)] = Trine [4.3456] = 4 y(n) = y,(n)ty,(n)=344=7 4 ow Trunc [x, (n) +x, (n)] = Trane [3.4567 + 4.3456] y(n) y(n) = 8 Since y(n) # y" (1); the system is non-linear, ‘Trane [7.8023] ¥) Causality : ‘The trancation operation depends only on present input, It does not contain any ture sample, So the system is causal, ) Stability : If input x(n) is bounded (less than infinity) then the result of truncation is also yunded. So it is BIBO stable system. Note : The properties of rounding operation are similar to truncation operation. So on the same lines we can show that rounding operation is static, non-linear, time invarient, causal and stable. 2.7- Linear Shift Invariant (LSI) Systems : => ASKED IN EXAM - MAY 2003 11! J [As the name indicates LSI systems are linear and shift (or time) invariant systems. So these systems are also called as linear time invariant (LT!) systems. Generally, in practical cases, majority discrete time systems are linear and time invarient in nature, In this section we will discuss LTT systems in detail. 2.7.1 Representation of Discrete Time Signal as Weighted Impulses : In this sub-section we will show that LTI system can be completely characterized by its impulse response. We will be using following notations related to discrete time systems. x(n) => Inpot sequence h(n) => Impulse response of system y(n) => Output sequence Consider input sequence, x(n) = {1,2,1,2, 1) wn) +D.S.P.(E & TC) 2-82 Signals and Systems This sequence is represented as shown in Fig, 2.35. The different values of x(m) are as follows : x(-2)=1, x(-1)=2,x(0)=1,x(1) =2,x(2) = pet fore Fig. 2.35(a) : Input sequence x(n) Now consider unit impulse 6 (1). It is expressed as, = [1 forn=o 5 = {Soi “®) ‘That means its value is 1 at n = 0 and its value is zero for all other values of ‘n’. Such unit impulse is as shown in Fig.'2.35(b) : 0 i Fig, 2.35(b) : Unit impulse, 5(n ) If we want to delay this unit impulse by ‘2° samples then it is denoted by 3(n-2) 1 forn=2 And 8(n-2) = 8) 0 otherwise This sequence is shown in Fig. 2.35(c). {B(n-2) ={0, 0, 1): Fig, 2.35(0) : Impulse sequence 8 (n-2) Similarly if we want to advance this unit impulse by ‘2’ samples then it is denoted by 3(n+2). 1 forn=2 6(n+2) = (4) 0 otherwiseDSP. (E & TC) 283 Signals and Systems This sequence is shown in Fig: 2.35(d) ‘&(nv2) = (1, 0,0}. 7-2 To til Fig, 2.35(d) : Impulse sequence 5(n +2) In general, if § (1) is delayed by ‘k’ samples then we can write, 1 forn=k~ S(n-k) = 6) 0 otherwise Decomposition of input sequence x(n) The given input sequence is, x(n) = 2,420 t Its range is from ~2 to #2. Now decomposition means obtaining a particular sample from given sequence x(n), Suppose we want to obtain sample at n = ~2 that means x(-2). Then take the multiplication of x(m) and 8(n4+2), We know that the value of 8 (n-+2) is one only-at n = 2, while for all other values of n, it is zero. Thus if we take multiplication of x ( n)and6(n+2) then we will get the sample x(~2). Because all ther terms become zero, after multiplication. This is shown in Fig, 2.35(¢). ~ 4 5. = Multiplication Bin42) = (1,0, 9, 0,0} Fig, 2.35(6) : Decomposed input signal sample at m = -2DSP. (E & TC) 284 Signals and Systems Similarly other decomposed input signal samples are obtained. Fig. 2.350) + Fig, 2.35(g) : Decomposed sample at n = 0DSP. (E & TC) 2-85 Signals and Systems &) <= Muttiplication | 8(n-1) = (0, 0,0, 1, 0} preteen nen e ene e nce secu eaten Fig. 2.35(h) : Decomposed sample at n= +1 @ = Multiplication 8(n-2) = {0, 0, 0, 0, 1} t ‘ i i O12 Fig. 2.35() : Decomposed sample at n = +2 ‘Now the original sequence x(n) can be obtained simply by adding all decomposed samples as shown in Fig. 2.36DSP. (E & TC) 2:86 Signals and Systems 2 Fig. 2.36 : x(n) = Addition of all decomposed samples ‘Mathematically we can write, x(n) = x(n)-8(n+2)+x(n)-8(n+1)+x(m)-8(n)+x(m)-3(n-1)+x(n)-6(n-2) (6) Observe every decomposed sample. We get a decomposed sample only at n = k. For example sar we x( ou de St de St StDSP. (E & TC) 287 Signals and Systems sample x (~2) is obtained by multiplication of x(n) and 5(n+2) as shown in Fig. 2.35(e). This sample is present at n = k = 2, Thus for simplicity replacing x(n) by x(k) in RHS. of Equation (6) we get, x(n) = x(k)8(n+2) +x (k)5 (n+ 1) +x(k)5 (0) +x(k) 3 (n-1) +x(k)8(m=2) 7) In terms of summation we can write, 2 x(n) = Y x(k)8(n-k) (8) k=-2 In Equation (7), the amplitude of every ‘8 term is 1. We are multiplying every unit impulse (advanced or delayed) by corresponding amplitude of input sample. This is called as scaling of unit samples. Thus Equations (7) and (8) indicates that input sequence x(n) is represented as sum of weighted (or sealed) unit impulses. In Equation (8) the limits of summation are from -2 to +2; this is because input sequence x(n) is present from n = —2 to n= +2. In general we can write Equation (8) as, © x(n)= YY x(k)8(n-k) k=-0 0) 2.7.2. Linear Convolution (Convolution Sum) : Proof of linear convolution Or Proof of “LTT system is completely characterized by unit impulse response h(n)" : Consider relaxed LSI (LT) system, A relaxed system means if input x(n) is zero then output y(n) is zero. Let us say unit impulse 8(m) is applied to this system then its output is denoted by h(n). “h(n)” is called as impulse response of system. Step I: T 8(n) —> y(n)=h(n) Since the system is time invariant, if we delay input by ‘k’ samples then output should be delayed by same amount. Step Il T 3(n-k) —-> y(n) =h(n-k) ‘Now multiplying both sides by x(k) we get, Step IH: T x(k)8(n-k) ———> y(n) =x(k)A(n-k)DSP. (E & TC) 2-88 Signals and Systems Since the system is linear, we can apply superposition theorem. So taking summation at both sides we get, Step IV: ~ T rs D x(k)8(n-k) —> y(n)= Y x(k)h(n-k) k=-0 ‘Thus we have, 0 y(my= Yo x(kyh(n-k) k=-0 (10) This summation is called as convolution sum. In this case output y (1) is obtained by taking convolution of x(n) and h(n). This is called as linear convolution and it is denoted by symbol o x(n)#h(n)= Y) x(k)h(n-k) In this case we can say that the behaviour of LTI system is completely characterized by its impulse response. 2.7.3 Computation of Linear Convolution : ‘The different methods used for the computation of linear convolution are as follows : 1) Graphical method 2) Using mathematical equation of convolution 3) Tabulation method 4) Maltiptication method 1) Graphical method : ‘The linear convolution of two sequence is given by, © y(n) yy x(k)h(n-k) wn) a) We will have to calculate output for different values of ‘n’. For n= 0 we get, © y(0) = Yo x(k)h(-k) 0) k ~~ DSP. (E-& TC) 2-89 Signals'and Systems Note that h (—k)'indicates folding of h (k). For n= 1 we get, © ya) = Y x(kyh(-k) «.) k=-0 “Here the term h(1-k) can be written as h (~k+1). Thus Equation (3) becomes, yl) = Y xGeh(-k+1) @) k=-0 Here h(—k+1) indicates shifting of folded signal h(~k). It indicates that h(—k) is delayed by ‘1° sample, Similarly for other values of ‘n’ output y (1) is calculated. ‘Thus different operations involved in the calculation of linear convolution are as follows : i) Folding operation : It indicates folding of sequence h(k). ii) Shifting operation : It indicates time shifting of h(—K) eg. h(~k+1). iii) Multiplication : It indicates multiplication of x(k) and h(n~k). iv) Summation : It indicates addition of all product terms obtained because of multiplication of x(k) and h(n-k) Solved Problems : Prob. 1: Obtain linear convolution of following sequences x(n) = (1, 2, 1, 2} and h(n) = (1, 1, 1} Soln. : If the sequences are given in terms of ‘n’ then obtain x(k) and h(k) by replacing ‘n’ by ®. x(k) = (1, 2,1, 2} and h(k) = 40 ‘Arrow is not mentioned in any sequences. So by default it is at first position. x(k) = €1,2,1,2) and = h(k) = (11,3 t t Now we have equation of convolution, : y(n) = DD x(k)a(n-k) ww) k=-0 ‘Now we have to decide range of ‘n’ and *k’.DSP. (E & TC) 2.90 Signals and Systems Range of ‘n’ : Basically to calculate y(n) we should know from which value of ‘n’. We should start (lower range of n) and what is ending value of ‘n’ (higher range of n). We will use following notations to decide range of y(n). y, = Lowest range of y(n) Y, = Highest range of y(n) x, = Lowest range of x(k) %, = Highest range of x(k) = Lowest range of h(k) 4h, = Highest range of hk) Now use following formulae to calculate range of y(n) Y= % th and y, =, +h, “From the given sequences we have, 420 30h, ¥, = X+h,=0+0=0 h=2 andy, = xth=3+2=5 Thus range of y (mn) is from y (0) to y(5). Range of ‘k': The value of k in the summation sign will be always same as the sequence x(k), Thus in this case the range of is from k = 0 to 3. So Equation (1) becomes, 3 y(n) = Y x(kyh(n-k) @) k=0 Now we will calculate output y (n) by putting differem values of n from 0 to 5, Calculation of y(0) : Putting n = 0 in Equation (2) we get, vers Stey Stey Ste SteDSP. (E & TC) 201 Signals and Systems Es é y(0) = ¥ x(k)(O-ky= YP x(kyh(-k) @) k=0 k=0 Now we have to perform calculations using graphical method. Here h(~k) indicates folded version of h (k). Step 1: Sketch x(k). Sketch of x(k) is as shown in Fig. 2.37(a) Fig, 2.37(a) : Sequence x(k) Step Il: Sketch h(k) Sketch of h (Kk) is shown in Fig. 2.370). WW) = (14,1) Fig, 2.37(b) : Sequence h (k) Step III : Fold h(k) to obtain h (~k), Sketch of h (~k) is shown in Fig. 2.37(c) Fig, 237(0) : Folded sequence h (-k) Step IV : Obtain multiplication of x(k) and h(—k). This multiplication takes place on sample to sample basis as shown in Fig. 2.37(@).DSP. (E & TC) 2:92 7 Signals ard Systems «= Multiplication t Product of x(k) and h(—k) Step V : Take the summation of all product terms. In this case as result of multiplication we have only one sample at n= 0. o y(O)=1 Calculation of y(1) : Putting n= 1 in Equation (2) we get, 3 y(n) = Y x(k) -k) A) k=0 h(1-k) can be written as h(—k +1). Thus Equation (4) becomes, 3 y(n) =D x(kyh(-k+1) 6) k=0 Step 1: Now h(—k+1) indicates delay of h(-k) by ‘1’ sample. So shift the sequence in Fig, 2.37(c) towards right by ‘1’ sample, This sequence is shown in Fig. 2.38(a). DSI Step Ste Ca StD.S.P. (E"& TC) 2.93 Signals and Systems Step IL: Obtain multiplication of x(k) and h(—k+1). It is shown in Fig. 2:38(. © e Matipication )=(0,1,1,1,0, Fig. 2.38(b) : Product of x(k) and h(-k+1) Step TI :Now according to Equation (5) we have to add all product terms shown in Fig. 2.38(b) to obtain y(n). » y(1) = (1x1)+(2x1) 5142 . y()=3 Calculation of y(2) : Putting n= 2 in Equation (2) we get, 3 y(2) = x(k)h(2-k) -O k=0 ‘The term h(2-k) is same as h(-k+2). -_ 3 » y(2) =D x(k)h(-k+2) AD) k=0 Step 1: Hore h(—k-+2) indicates delay of h(-k) by ‘2’ samples. This is obtained by shifting sequence in Fig. 237(c) towards" right by ‘2’ samples. This sequence is shown in Fig. 239(a).DSP. (E & TC) 2.94 Signals and Systems a p(k) = (11 AP . | Stey 11s Fig. 2.39(a) : Sequence h (~k+2) Step I: Obtain muttiplication of x(k) and h(-k+2). It is shown in Fig. 2.390). Ste 123 Fig, 2.39(b) : Product of x(k) and h(-k +2) Step. III : Add all product terms to obtain y (2). y(2) = (ix1)+(1x2)+(ixl) 214241 y(2)=4 Calculation of y (3): Putting n = 3 in Equation (2) we get, 3 y(3) = LD x(kyh(3-k) -@) e ee k=0 ‘The term h(3-k) is same as h(-k +3) 3 St y(3) = Y x(kyh(-k+3) -O) k=01 a DSP. (E & TC) 295 Signals and Systems Step 1: Here h(-:k--3) indicates delay of h(—k) by 3 samples, It is shown in Fig. 2.40(a). Fig. 2.40(a) : Sequence h (~k +3) Step 11: Obtain multiplication of x(k) and h (~k +3), This sequence is shown in Fig. 2.40(b). @ © Multiplication “F n(-ke3) = (0, 1,1. 1) - Beene a Co opt MB 2 op 123 Fig. 2.40(b) : Product of x(k) and h(-k+3) Step III : Add all product terms to obtain y (3). y(3) = (0x1) 4(1x2) +(x 1) 4(2x1) = 0F2 +142 y(3)=5 Calculation of y (4) : Putting n= 4 in Equation (2) we get, 4 y(4) =D x(kyh(4-k) k=0 (10)DSP. (E & TC) 2-96 Signals and Systems The term h(4—K) is same as h(-k +4) 4 y(4) = Y x(kyn(-k+4) 411) k=0 Step 1: Here h(-k+4) indicates delay of h(-k) by ‘4” samples. It is shown in Fig. 2.41(a): b+) =(0,0,1,4,1} t Fig, 2.41(a).: Sequence h(—-k+4) Step I: Obtain multiplication of x(k) and h(-k+4), This sequence is shown in Fig. 2.410). Fig, 2.41(b) : Product of x(k) and h(-k +4) Step III : Add all product terms to obtain y (4). y(4) = (1x0)+(2x0)4(1x1)+(2x1)+(0x1) = 040414240 6 y(4)=3 Calculation of y(5): Putting n= 5 in Equation (2) we get,D.S.P. (E & TC) “ 297 Signals and Systems. 3 ; y(5) =D, x(k)h(5-k) (12) k=0 ‘The term h(5—k) is same as h(-k +5) 3 ‘ y(5) = x(kyh(-k+5) (13) k=0 Step I: Here h(—k-+5) indicates delay of h(—k) by ‘5° samples. Its shown in Fig. 2.42(a) h-k#5) Sena cs Fig. 2.42(a) : Sequence h(—-k+5) Step I: Obtain multiplication of x(k) and h(-k-+5). This sequence is shown in Fig, 2.42(b). 12348 ©: & Matipication Fig, 2.42(b) : Product of x(k) and h(-k+5) ‘Step HII : Add all product terms to obtain y (5). y(5) = (1x0)+(2x0)+(1x0)+(2«1)+(0x1)+(0%1)DSP. (E & TC) 2-98 Signals and Systems The result of convolution, y (1) is represented graphically as shown in Fig. 2.43. 1,8, 4, 5,3, 2) o 12345 Fig, 2.43 : Sketch of y(n) = { 1, 3, 4, 5, 3,2} lg Prob, 2: For the following signals, determine ‘and sketch convolution y (n) graphically x(n) = $n Osns6 = 0 otherwise and h(n) = 4 -25n82 = 0 otherwise _ [May 2002, DSP (E&TO)] Soin. : First we will write sequences x(n) and h(n) by putting different values of n. Sequence x(n): Here range of x (1) is from n= 0 ton=6 1 and x(n) = 30 Forn=0 3 x(0)=2x0=0 Forn=1 2 x(1)=dx1= Forn=2 => x(2)=5x2= Forn=3 = x(3)=4%3= Forn=4 =>x(4)=1x4= 3 1 Forn=5 = x(5)=3x5= - a1y6- Forn=6 => x(6)= 4x6 = WIA BIW BIE BIE BIR wiE ‘Thus x(n) can be written as, T loDSP. (E & TC) 2-99 Signalg and Systems x(n) = {2x¢0),x(1) (2), x(3),X(4),%(5),%(6)} le2esg4oeo (o4-53-25-5] ~@) t ’ x(n) Now h(n) is given as, h(n) = 1 for-2 x(k)h(-k) 9) k=0 Step I: h(—K) is folded version of h(k). It is shown in Fig, 2.44(0). Step Il The product of x(k) and h(—k) is shown in Fig, 2.46,y DSP. (E & TC) Fig, 2.46 : Product of x(k) and h(—k) Step Ill :y (0) is obtained by adding all product terms. = 1 21)-042 y(0) = ov(ge}+(Fet) ote y(0)=1 Calculation of y(1) : Putting n= 1 in Equation (6) we get, 6 yl) = & x(kyh(1-k) (10) k=0 Step 1: h(1—k) is same as h(-k+1). It indicates delay of h(—k) by “1? sample. It is obtained by shifting h (—_k) towards right by ‘1’ sample as shown in Fig. 2.47(a).SP. (E & TC) 2404 Signals and Systems Fig, 2.47(a) : Sequence h (-k+1) tep IL: The product x (k)-h(~k-+1) is shown in Fig. 2.470). 129456 é ® e Muttplcation : Fig. 2:47(b) : Product of x(k) and h(-k+1) 19 St Cc Ste Sti GGDSP. (E & TC} 2-105 Signals and Systems Step IMl ty (1) is obtained by adding all product terms. 2 yQ)= (0x1) +(0x0)4 Ex 1) 4Ex Hed) = 54945" 3 A yQ)=2 Calculation of y(2): . Putting n= 2 in Equation (6) we get, 6 yQ2) =D x(k)h(2-k) . 11) k=0 Step 1: h(2—k) is same as h(-k+2). It indicates delay of h(—k) by 2 samples. This sequence is obtained by shifting h(-k) towards right by 2 samples as shown in Fig, 2.480). Step Il: The product x (k)-h(2—k) is obtained as shown in Fig. 2.48(b) [on next page]. Step III :y (2) is obtained by adding all product terms. 1 3 4 1.2.3.4 L$ y(2) = (Oxtye xy Bx 4G DG #1) #3454349 yQ)=5 Calculation of y(3) : Putting n = 3 in Equation (6) we get, 6 y(3) =D x(kyh(3-k) (2) k=0DSP. (E & TC) 2-106 Signals and Systems 23456 Fig. 2.48(b) : Product of x(k) +h(2-k) Step 1: Here h(3—k) is same as h(-k+3). It indicates detay of hk) by ‘3° samples. It is obtained by shifting h (—k) towards right by 3-samples as shown in Fig. 2.49(a). was) =H, 7, HF forett Fig. 2.49(a) : Sequence h(—k+3) Step I: The product x(k) (3k) is obtained as shown in Fig. 2.49(b).DSP. (E & TC) Fig, 2.49(b) : Product of x(k)-h(3-k) Step II zy (3) is obtained by aidding all product terms. =f(i 2 y(3) = (ba}+(F Calculation of y (4) : Putting n = 4 in Equation (6) we get, 6 y(4) ‘= YS x(kyh(4-k) | 03) k=0 Fig. 2.50(a) : Sequence h(-k +4) Step 1: h(4-k) is same as h(—Kk+4), It indicates delay of h(-k) by ‘4’ samples. It is =" obtained by shifting h(—) towards right by ‘4’ samples as shown in Fig. 2.50(a). Step I: The product of x(k) and h(~k+4) is obtained as shown in Fig. 2.50(b)."DSP. (E & TC) 2-108 Signals and Systems 123456 ®e Multiplication 5.00). yt ttt} ‘F hi-kH4) = (0, 0, Sedo ol ie23456 Fig. 2.50(b) : Product of x(k) and h(-k+4) Step Il :y (4) is obtained by adding all product terms. * =(1 2 3 4 3 ny(4) (+0)*(Se}e(S0 *ba}+(F4 2 y4)e2 Calculation of y (5): Putting n= 5 in Equation (6) we get, 6 y(5) = Y x(k)h(5-k) (14) k=0 Step 1: h(5-K) is same as h(-k+5), It indicates delay of h(-k) by 5 samples. It is obtained by shifting h (—k) towards right by 5 samples as shown in Fig. 2.51(a), Ste; StepD.S.P. (E & TC) Reza taboo i Ted) Fig. 2.51(@) : Sequence h(-k +5) Signals and Systems Step Il: The product of x(k) and h(-k+5) is obtained as shown in Fig. 2.51(b). a eee eee eee see eae 0 Fig, 251(b) : Product of x(k) and h(—k +5) Qe Mitipicaton- h(-k#5) = (0, 0,0,1, 1,1, 1,4 soi 123.4 5-87 Step TH :y (5) is obtained by adding all product terms. y(3) = ( a} fa }Go}($) 1 y(5)=6DSP. (E & TC) 2-410 Signals and Systems Calculation of y(6): Putting n = 6 in Equation (6) we get, 6 y(6) = & x(k)h(6-k) (1s) k=0 Step I: h(6—k) is same as h(-k+6). It indicates delay of h(-k) by 6 samples. It is obtained by shifting h (-k) towards right by 6 samples as shown n Fig. 2.52). o 12345678 Fig, 2.52(a) : Sequence h (~k+6) Step IT: The pro Fol 12945867 6 ‘® e Multiplication “e Fig. 2.52() : Product of x(k) and h(-k+6)DSP. (E & TC) 2411 Signals and Systems Step III zy (6) is obtained by adding all product terms. - (4 3 67 )244548-8 y(6) = (d)*(S9}($1) F45t37 3 y(6) Putting n= 7 in Equation (6) we get, 6 y(7) =D x(k) h(7-K) k=0 Step 1: h(7—K) is same as h(-K+7), It indicates delay of h(—K) by 7 samples. It is obtained by shifing h (~k)) towards right by 7 samples as shown in Fig. 2 53(a). Calculation of y (7): 6) 0,0,1,1,1, 4,1) 234567 89 ie: Fig, 2.53(a) : Sequence h (~k +7) Step Il: The product of x(k) and h(—k+7) is obtained as shown in Fig. 2.5300) fon next page]. Step It :y (7) is obtained by adding all product terms. Calculation of y(8) + Putting n= 8 in Equation (6) we get, 6 y(8) =D x(k)h(8-k) 0) k=0 .DSP. (E & Tc) 2412 Signals and Systems i2a45 8789 ©°@ & Mutiotcation rT #h(-k+7) = (0, 0, 0, 0, 0, 44,4, 4, 4) Fig. 2.53(b) : Product of x(k) and h(-k+7) Step T= h(8-K) is same as h(—k-+8). It indicates delay of A(-k) by 8 samples. It is obtained by shifting h (—) towards right by 8 samples as shown in Fig. 2.54(), MAH) 5 0,0,0.0,0, 114.1, 1) bara 12.3.4.5.6.7 8 9407 Fig, 2.54(a) : Sequence h(-k +8) Step I : The product of x(k) and hn(—-+8) is obtained as shown in Fig 2.540). 9 Step y(n)+r DSP. (E& TC) 2-413 Signals and Systems 0 @ e Mutiplcaion ' i ' ' ' ' 1.23.45 67.8 9.10 | i fod ' PCH 1G 0.0,0.00, 1.14.1) ; ' ' ' ' ‘ x{k)hi-k+8) at 0,0,0,0,0,,0,0,0, of “ 3 oly234567 8910 | Fig, 2.54(b) : Product of x(k) and h(-k+8) Step HI zy (8) is obtained by adding all product terms. | s y(B) = S1-$ \ “ y(8)=2 | Now result of convolution y (1) is writen as, yin) = fy (-2),y(-1),y(0),9 1), 9 (2),9 BY 4),9 3,9 (6) Y (Toy (8)} | sy(n) {o5 i 2 :DS.P. (E & TC) 2414 Signals and Systems Note : Arrow always indicates the sample at n = 0. Thus arrow should be marked at sample y¥(0). * The sequence y(m) is graphically represented as shown in Fig, 2.55. 2345678 Fig, 2.55 : Result of linear convolution Prob. 3: Prove and explain graphically the difference between relations (i) X(m)8(n=ng) = XCM) (i) x(n) #8 (m-ng) = x(n=Ng) {Dec. 2001, DSP (E&TO)} Soln. : @ x (m)5(n—n9) = x(n) Here x(n) is input sequence, Let the input sequence be given by, x(n) = {1,1,2,1, 1) t This sequence is plotted as shown in Fig, 2.56(a). Fig, 2.56(a) : Sequence x(n) We know that 8(n) represent unit impulse, It is defined as, 3(ny= 1 forn=0 = 0 elsewhere Fore Hen @DSP. (E & TC) 2445 Signals and Systems sequence 5 (n) is shown in Fig. 2.56(b). Fig. 2.56(b) : Unit impulse, 5 (n) 7 Now 8(n=ny) indicates delay of 5(n) by no units. So sequence 5(n=ny) can be represented as, 3(n-m) = 1. for n=n, 0 for nen For example, say n, = 1. Thus we get, 8(n-1) =1— forn=1 = 0 fornen, ‘The sequence 8 (n— 1) is plotied as shown in Fig. 2.56(0) “#8(0-1) = (0, 1) t Fig, 2.56(c) : Sequence 6(n-1) ‘Now multiplication of x(n) and 8(n-n,) ie. 8 (n—1) is shown in Fig. 2.56(d). Here we ae getting the resultant sample only at n = ny. We have assumed ny = 1. Thus result of multiplication is the sample at n = 1. This value is 1; which is the value of x (1). x(n)8(n-1) = x(1) ‘Thus in general we can say, x(m)8 (m=) = x( 0) Hence proved. i) x(n) *5(n—ng) =x(m—m9) Here ‘ indicates convolution. So we have to calculate convolution of x(n) and 5 (n—m) ). According to the definition of convolution, © y(n) = x(n)*h(n)= SY) x(k)h(n-k) wal)DSP.(E&TC) . 2416 Signals and Systems Fig, 2.56(d) : Product of x(n) (n—ny) Let h(n) = S(a-n) on(2) Now assume two sequence as follows, x(n) = (1,1, and h(n) = 8(n—n,) = {0, 1} t v also assume that value of ny Note : We cannot assume any arbitrary sequence for h(n). Since h(n) = 8 (nn); it should be in terms of delayed unit impulse. Step 1: Sequences x(k) and h(k) can be written by directly replacing ‘n’ by “K’. x(k) = 1,1) and h(k)=8(k-m) = (0, 1) t t ‘Thus different samples of x(k) and h (k) can be written a, x(-1)= 1 similarly, h(0) = 0 x(0)= 1 and h(1) = 1 and x(1) = 1 ° Plot of x(k) and h (i) is shown in Fig. 2.57. bella ae j Fig 257 (a) Step 1 Step Calc indicDSP. (E & TC) 2417 Signals and Systems Fig. 2.57(b) : Sequence h(x) Step I: Now range of ‘n’ is calculated as follows : lower value of =n => y,= x) +h, =-1+0 and higher valuc of 9=n=>y,= x, +h,=1+1=2 ‘Thys range of ‘n’ is from ~1 to 2. That means we have to calculate sequence y(n) from y(-1) t0 (2). Range of ‘k’ is same as x(k). Thus ‘k’ varies from -1 to +1. Step III : Putting the value of ‘k’ Equation (1) can be written as, 1 y(n) = DY x(k)yh(n-k) .@) k=-1 To calculate y(n); put values of ‘n’ from —1 to +2. Calculation of y(=1) : Putting n= — 1 in Equation (3) we get, 1 y(-1) = D x(k)h(-1-k) ) k=-1 Here h(-1-k) is same as h(-k-1)-h(—k) is folded version of h(k) and h(~k-1) indicates advance of h (—k) by 1 sample. This sequence is shown in Fig. 2.57(c). ‘The multiplication of x(k) and h (—k-1) is shown in Fig, 2.57(4). y (~ 1) is obtained by adding all product terms.DS.P. (E & TC) 2418 Signals and Systems xikjhiciet) Zyl) = ye) =O E Let Jo4 Fig, 2.57(d) : Product of x(k) and h(-k-1) Calculation of y(0) : Putting n = 0 in Equation (3) we get, 1 y(0) = Yo x(kyh(-k) (5) k=-1 Sequence h(~K) is shown in Fig. 257(c). The product of x(k) and h(—k) shown in Fig. 2.58, y (0) is obtained by adding all product terms. y(0) Putting n= 1 in Equation (3) we get, Calculation of y(1) : 1 y¥Q1) = DL x(k)h(1-k) (6) kSignals and Systems Fig, 2.58 : Product of x(k) h(-k) Here h(1—k) is same as h(-k+1). It indicates delay of h(—K) by 1 sample. It is obtained by shifting h(-K) towards right by 1 sample. The sequence h(—K-+1) is shown in Fig, 2.59(a). The product of x(k) and h(—K-+ 1) is shown in Fig, 2.5900). Fig, 259(a) : Sequence h(-k+1) yQ) Putting n = 2 in Equation (3) we get, Calculation of y(2) = 1 y(2) = L x(kyhQ-k) 1) k=-1DSP. (E & TC) 2-120 Signals and Systems Fig, 2.59(b) : Product of x(k) and h(—k+1) Here h(2-k) is same as h(—-k+2). It indicates delay of h(-k) by two samples, It is obtained by shifting h(—k) towards right by two samples. The sequence h(—k-+2) is shown in Fig. 2.60(a). The product of x(k) and h(~k+2) is shown in Fig, 2,60(b). Y (2) is obtained by adding all product terms. yQDSP. (E & TC) 2124 Signals and Systems Fig, 2.60(b) : Product of x(k) and h(-k +2) ‘The output sequence y(n) is, y(n) = {y(=1),y(0),¥(1),y(2)} y(n) = O10 i This sequence is plotted in Fig. 2.61. Now the output sequence y(n) can also be written as, y(a) = {1b t Observe graph of x(k) shown in Fig. 2.57(a) and graph of y(n) shown in Fig, 2.61. It shows that output y (m) is obtained by delaying x(k) by ‘1” sample, y(n) = x(k-1) But x(k) is same as x(n)DSP. (E& TC) 2122 Signals and Systems y(n) = x(n-1) Initially we have assumed n y(n) = x(n-n,) y(n) = x(n)*8(n-n)=x(n-n) Hence proved. Note : This indicates that the convolution of input sequence x(n) with delayed unit impulse is equivalent to delay of input sequence by corresponding samples (ng). Prob. Obtain the expression for convolution of unit step sequence with finite duration sequence, Soln. We will assume some sequence x(n) as, x(n) = 22,1) (ly t Note that this is finite duration sequence. We have to obtain convolution of x (1) with unit step u(n). So let, h(n) = u(a) »-Q) The equations of h (k) and x(k) can be directly written as, x(k) = (2,2,1) and h(k)=u(k)= (1, 1,1... t t According to the definition of convolution we have, © y(n) = YD x(k) h(n-k) +B) k To obtain range of y(n) : Lowest index of y(n) => y, = x,+h,=0+0=0 Highest index of y (n) => y, =x, +h, = 2+00 = 0 Ie va Cc Fi,DSP, (E & TC) 2123 Signals and Systems ‘Tins range of y(n) is from ‘0° to ‘x’. That means we have to obtain y(n) for different values of from 0 to ©. Now the range of k is same as x(k). So range of K is from 0 to 2, Calculation of y(0) : y (0) is calculated by putting n = 0 in Equation (3). 7 y(0) = Y x(k)h(-k) a k=0 ‘The sequence x(k) is shown in Fig. 2.62(@). Sequence h(k) = u(n) is shown in Fig, 2.62(b). Now h(-K) is folded version of h(k), It is obtained by folding unit step u(m). Sequence h (~k) is shown in Fig. 2.62(0). Fig. 2.62(¢) : Sequence h (-k) =u(-n) ‘According to Equation (4) y(O) is obtained by multiplying x(k) by h(—K) and then adding all product terms. ‘This is shown in Fig. 2.62(d).DSP. (E & TC) D.S.P, 2.124 Signals and Systems —_— Calculation of y(1): — Putting n = 1 in Equation (3) we get, 2 y(n) = YD x(kynci-k) (5) k=0 No - k 5 ot ; Calculati Fig, 2.62(d) : Product of x(k) and h(-k) ai Here h(1~k) is same as A(-Kk+1), It indicates delay of h(~k) by ‘I’ sample. This uence is shown in Fig. 2.63(a), he samples as Fig, 2.63(a) : Sequence h (-k+1)D.S.P. (E & TC) 2-125 Signals and Systems ‘The multiplication of x (k) and h (—k-+1) is shown in Fig. 2.63(6). =y()=242=4) k lo 4 Fig. 2.63(b) : Product of x(k) and h(-k+1) Now y (1) is obtained by adding all product terms. y(1) = (2x1)+(2x1)=24+2=4 Calculation of y(2) : Putting n= 2 in Equation (3) we get, 2 y(2) = Y x(kyh(2-k) (6) k=0 h(2—k) is same as h(-k+2). It is obtained by shifting h(—k) towards right by ‘2 samples as shown in Fig. 2.64(a). 2-1 or 2 Fig, 2.64(a) : Sequence h (~k+2)DSP. (E & TC) . 2426 Signals and Systems ‘The product of x(k) and h(~k-+2) is'shown in Fig. 2.64(b). Fig, 2.64(b) : Product of x(k) and h(-~k+2) y(2) = (2x1)+(2x1)H(1x1) #2424155 6 y(2)=5 Calculation of y(3): Putting n = 3 in Equation (3) we get, 2 y(3) = XY x(k)h(3=k) (7) k=0 Sequence h(3-k) is same as h(-k+3). It is obtained by shifting h(—k) towards right by 3 samples. This sequence is shown in Fig. 2.65(a). Fig. 2.65(a) : Sequence h(-k+3) DS# ‘ magnity So for: Generzr DSP. (E & TC) 2 27 Signals and Systems The product of x(K) and h(~k+3 ) is stiown in Fig. 2.65(b). y(Q)=24+244= k Fig, 2.65(b) : Product of x(k) and h(-k+3)_ - y(3) = (2x1) 4(2x1) +(x 1)= 2424155 yG) Since unit step is infinite sequence; it is clear that all the next samples of y(n) will have magnitude equal to 5. This is because every time we will shift h(-k) towards right by 1 sample. So for every value of n the product of x(k) and delayed h (—k) will remain same. y(4) = 5,y(5)=5,y(6) = 5... Now the resultant output y (n) can be written as, y(n) = {y¥(0),y(1),¥(2),¥ Bs y(n) = eee (8) General Expression : The given sequencé x(n) is , x(n) = (22,1) andh(n)=u(n) = {1,1,1, 1, ..3 tee iDSP. (E & TC) 2128 Signals and Systems x(0) = 2 h(O) = u(0)=1 x(1) = 2 h(1) = u(1)=1 and x(2) = 1 h(2) = u(2)=1 (3) = u(3)=1 Now observe Equations (6) and (7) carefully. We can write the generalized expression to obtain Equation (8) as follows, y(n) = x(0)u(n)+x(1)u(n=1)+x(2)u(n=2)4... 9) Let us verify this equation. To calculate y (0 ) put n = 0 in Equation (9) ¥(0) = x(0)u(0)+x(1)u(-1)+x(2)u(-2) y(0) = (2x1) +(2x0)+(1x0) =2 This is because u(0) = 1, u(-1) = 0 and u(-2) = shown in Fig. 2.66, . This can be observed from unit step Similarly putting n= 1 in Equation (9) we get, ¥(1) = x(0)u(1)+x(1)u(0)+%(2)u(-1) y(1) = (2x1)4+(2«1)+(1x0)=2+240 yO) = 4 Similarly other values can be verified. Prob. 5: Obtain linear convolution of two discrete time signals given as, x(n) = u(n) h(n) = a®u(n), act Som. : Given x(n) = u(n) ie. unit step which is shown in Fig. 2.66. h(n) = a'u(n) is an exponential signal. Here a is exponential signal which is multiplied by u(n). We know that u(n) is present only for positive values of ‘n’. That means from n= 0 to n= co, So this multiplication indicates that a° is present for positive values of ‘n’. It will not affect the magnitude of a", because magnitude of u(n) is 1. This sequence a” is shown in Fig. 2.67. and ke will us ? infinite SeDSP. (E & TC) 2429 Signals and Systems Fig. 2.67 : Exponential sequence a Recall the-equation of convolution, o y(n) = x(k) h(n-k) eal) © In this equation h(n—k) is delayed version of h (Kk). Instead of that we can delay x(k) and keep h (k) as itis. © y(n) = h(k)x(n-k) -Q) k a ‘This is commutative property of convolution which will be discussed later in this chapter. We ‘will use Equation (2) in this example because it is easy to delay u ( n) than a"u(n). ‘Now we have derived generalized expression of convolution for finite sequence.with an infinite sequence. This equation is, y(n) = x(0)u(n)+x(1)u(n=1)+x(2)a(n=2)+.. -G) ‘When both sequences are infinite then Equation (3) can be easily modified as, y(n) = ox (-2)u(n+2) +x (-1)u(n41)+x(0)U(n) +x(1)u(n=1)+x(2)u(n-2)+. nd) Here x (1) is some infinite sequence and h(n) = u(n). Equation (4) is based on Equation (1). inthis case impot x (Kk) is Kept as it is and hk) is delayed, But if we want to use Equation (2) to obtain linear convolution then, y(n) = .ch(-2)u(m42)+h(-1)u(n4 1) +hCO)U(n) +h(1)u(n-1)+h(2)0(n=2)+-. ©) In Equation (5) we are taking x(n) = u(n). The given equation of h(n) is, h(n) = a'u(n) (6) ~ Ja forn20 ee {3 forn<0 a) Sequence h(n) can be written by putting values of n in Equation (7)DSP. (E & TC) 2-430 Signals and Systems That means, IT tn=0 =h(n)=h(0)= Forn=1 =>h(n)=h(1)= Forn=2 =>h(n)=h(2)= Fotn=3 = h(n)=h(3)= h(n) = (aaa? en(8) le Putting values of h(n) is Equation (5) we get, y(n) = 1u(n)+au(n-1)+a'u(n-2)+a'u(n-3)+... (9) Equation (9) gives the convolution of given sequences. To calculate sequence y(n), put different values of n in Equation (9), y(0) = u(0)+au(-1)+a’u(-2)+au(-3) Here u(0) = 1,u(-1) =n(-2)=u(-3)=0 “ y(O)=1 Putting n= 1 in Equation (9) we get, y(1) = 1u(1)+au(0)+au(-1) +a? u(-2) Here u(1) = u(0)=1andu(-1)=u(-2)=0 y(1)=14a Now putting n = 2 in Equation (9) we get, y(2) = u(2)+au(1)#au(0)4au(-1) Here (2) = u(1)=u(0)=1andu(-1)=0 » y(2)=ltata® Similarly for the n'* sample, y(n) can be written as, y(n)=T+atattae + (19) In terms of summation Equation (10) can be written as, a y(n) = Yak (any k=0 Now use standard equation of summation,DSP. (E& TC) 2434 Signals and Systems N_Nt! 2 ‘me For a#1 y #&= a wn(12) n=N, N,-N,+1 For a=1 In the present example a< 1 thus a#1. ‘Now. from Equation (1), N, = lower limit of summation = 0 Ny = upper limit of summation = » Putting these values in the standard equation of summation (Equation (12) we get, feet} l-a AU) ¥(a) = Prob. 6 : Use discrete convolution to find the response to the input x(n) = a’ u(n) of the LT system with impulse. response h(n) = pTu(n). Soln. = . Given: x(n) = #Pu(n) and h(n) = b"u(n). Sequences x(k) and h (k) crm be directly written as, x(k) = Hu(k) and h(k) = bea (k) ‘According to definition of linear convolution we have, y(n) = Y)x(kyh(n-k) ) ke-2 We have h(k) = beu(k) (2) Sequence b(n) is obtained by replacing k by n~K in Equation @) a héaek) = Bo Ka(a-k) 3) ‘Thns Equation (1) becomes, yuy= > u(k)-b*u(n-k) y(n) = Dk bE uk) a (ak) () ke -# Sequence w(K) is unit sep. While u(m~K) = w(~k-+n) indicates delay of u(-k) by ‘n” samples, This sequence is obtained as shown’ in Fig. 2.68,DSP. (E & TC) 2432 Signals and Systems Fig. 2.68 : Sequence u(n—k) ‘Now the product of u(k) and u(n-k) is shown in Fig. 2.69. Fig. 2.69 ; Product of u(k) and u(n-k) This multiplication ie. u(k)-u(n-k) indicates that the sequences present only for the range 0 to n. So we will change the limits of summation in Equation (4). Yew limits of k will be from 0 ton. * y(n) = Yo ake™*u(k)-a(n-k) (5) k=0 But the term u(k)- u(n-k) = 1,Signals and Systems DSP. (E & TC) (6) y(n) = Daten’ Now Uk ae (7) oy yay = Dae k=0 outside the summation. Since the summation is in terms of ‘k’ we can take the term b” Bok ava anahvys y(n) Lik ko 1k (8) y(n) = BY (3) k=0 Use standard equation of summation, nN, _N,tH N, at-a? al yas} ta 0) n=N, N,-Ny+1 a Compare Equations (8) and (9). Here N, = 0, Ny=wanda=(§ a for 5#1 (10) Equation (10) gives convolution of a'u(n) and b"u(n). Prob. 7 : Impulse response of an LTI system is given as = h(n) =u(n)-u(n-N). Find the output y(n) of the system fo an input x(n) =a" u(n).DSP, Soin, (E& TC) 2-134 Signals and Systems 2 ‘The sequences x(k) and h(k) can be written as, x(k) = a®u(k) and h(k)=u(k)-u(k-N). According to definition of convolution, y(n) = Y x(k)h(n-ky (I) ks Equation (1) can also be written as, y(n) = Y) h(k)x(n-k) Q) k= -@ : x(n—k) is obtained by replacing ‘K by n-k u(k- x(n-k) = a®Ku(n-ky -B) Putting this value in Equation (2) we get, yiyy= (8a (n-k)]-[u(k)-0(k-N)] (4) k Consider the tern u(k)~u(k-N), Here ‘N’ is some constant, CK) is unit step and N) is, unit step delayed by NV samples. This sequence 1 (kN) i shown in Fig. 2.70(a). 1 2 3 NNH Neo Sequence u (kN) While the sequence uk) ~u(k~N) is shown in Fig, 2.700), Note that this sequence is present from ‘0’ to “N’. Now consider first term of Equation (4), Here u (nk) is same as.u(k-+n), Itis obtained as shown in Fig. 2.70(c),DSP. (E & TC) 2435 Signals and Systems NNH LQ) © Substraction |=. Oto N Fig, 2.70(b) : Sequence u(k)—u(k-N) ue) = (1, 11) 1) Fig, 2.70(¢) : Sequence u(-k+n) OtoN According to Equation (1) we have to obtain product of [u(n-k)] and [u(k)—u(k-N)], These two terms are basically unit steps shifted in discrete domain, This product is shown in Fig, 2.70(d) Note that we have assumed N Lowest index of x(n) + Lowest index of h(n) ye ythy -Q) and Highest index of ‘n’ for y(n) => Highest index of x(n) + Highest index of h(n) th --B) Decide range of ‘k’ : Range of k is same as range of x (). Step IIT : Once the range of ‘n’ and ‘k’ is decided; then put different values of ‘n’ in the equation of y(n). Examples (5), (6) and (7) are solved with the help of basic equation of convolution. Let us solve some more examples. Prob. 8: Compute linear convolution of the following : x(n)={1, 1, 14,1 }pand h(n) = (4, 4, 1,1} t t Soln. : We have, y(n) = > x(k)h(n-k) (1) k=-0DSP. (E & TC) 2.138 Signals and Systems Step 1: ‘The sequences x(k) and h(k) can be written as, xX(K)=L11 and h(k) = (1,1, it x(O)=1 and x(1)=1 x(2 x(3)=1 Step IL: Range of ‘n’ for y(n) is, Lowest index: y, =x) +h,=0+0=0 Highest index :y, =x, +h, =3+3=6 ‘Thus range of y(n) is from y (0) to y(6). Now range of ‘k’ is same as x(k). Thus range of ‘K’ is from k = 0 to k = 3, Step TI : Putting the value of ‘k’ in Equation (1) we get, 3 y(n) = YY x(k)h(n-k) ~Q) ko Calculation of y(0) : Putting n= 0 in Equation (2) we eet, 3. 3 y(0) = ¥) x(k)h(O-k) = x(k)h(-k) k=0 k=0 Expanding the summation by putting values of ‘k’ we get, ¥(0) = x(O)h(-0)-+x(1)h(-1)+x(2)h(~2) +x(3)h(-3) y(O) = (1x1)#(1x0)+(1x0)+(1x0) ash(-1) = h(-2)=h(-3)=0 y(0) Calculation of y(1) : Putting n = 1 in Equation (2) we get, 3 yQ) = ¥ x(k)h(1-k) k=0 Expanding the summation we get, y(1) = x(O)h(1)+x(1)h(0)+x(2)h(-1)4x(3)h(-2) y(1) = (1x1) #(1x1)+(1x0)+(1x0) 9 CakeD.SP.(E & TC) 2.139 Signals and Systems y(1) = 141... ash(-1)=h(-2)=0 : ya) Calculation of y(2) : Putting n= 2 in Equation (2) we get, 3 y(2) = XY x(k)h(2-k) k=o Expanding the summation we get, y(2) = xCO)h(2)+x(1)h(1)+x(2)h(0)+x(3)M(-1) y(2) = (1x1)+(1x1)#(1x1)+(1%0) w y(Q2) = 14141 ash(-1)=0 2 y(Q) =3 Calculation of y (3): Putting n = 3 in Equation (2) we get, 3 y(3) = & x(kyh(3-k) k=0 Expanding the-summation we get, y(3) = x(0)h(3)+x(1)h(2)+x(2)H(1) +x (3h (0) 2» y(3) = (xtt(xl)+(1x1)+(x1) 2 y(3) = 1414141 2. y@3) =4 Calculation of y(4): Putting n = 4 in Equation (2) we get, 3 y(4) = Yx(k)a(4-k) Keo Expanding the summation we get, y(4) = x(0)h(4)+x(1)h(3)+x(2)h(2)+x(3) (1) 2 y(4) = (1x0)+(1x1) (1x1) #11) c y(4) = 041441 ...ash(4)=0DSP. (E & TC) 2.140 Signals and Systems y(4) Calculation of y(5) : Putting n= 5 in Equation (2) we get, 3 y(5) = Y) x(k)h(5-k) keo Expanding the summation we get, y(5) = x(O)h(5)+x(1)h(4)4+x(2)h(3)+x(3)h(2) “ ¥(5) = (1x0)+(1x0)+(1x1)4(1x1) y(5) = 0404141... ash(5)=h(4)=0 Fe] Calculation of y(6) : Putting n = 6 in Equation (2) we get, 3 y(6) = Y) x(kyh(6-k) k=0 Expanding the summation we get, y(6) = x(0)h(6)+x(1)h(5)+x(2)h(4)+x(3)h(3) ¥(6) = (1x0)+(1x0)4(1x0)+(1x 1) y(6) = OF0+041 ..ash(6)=h(5)=h(4)=0 Step IV : The result of convolution y (1) is given by, y(n) = Cy (0), ¥C), ¥(2), ¥(3), yA), ¥(5), ¥(6) } [ y(n) ={1,2,3,4,3,2,1} t Note that arrow is marked at the sample y (0). Prob. 9: Compute the linear convolution of following : x(n) = 1 and h(n) ={2, 1,2, 1) Dee, 2000, DSP (E&TO)] Soln. : Step I: The sequences x(k) and h(k) can be written as,DSP. (E & TC) 2441 Signals and Systems x(k) = () and h(k) = @ 1,2, 9 * t x(0) = 1 Note that arrow is not present in the given sequences of x(n) and h(n). So by default it is a first (starting) position. Step II: Range of ‘n’ for y(n) is, Lower index : y, = x,+h,= 0 +0=0 Highest index : y,, = x, +h, = 0 +3=3 ‘Thus range of y (1) is from y (0) to y(3) Step IM : According to definition of convolution we have, a y(n) = Y x(k)h(n-k) ) k=-0 In this case h (n—k) indicates folding and delaying h(—k). Since x(k} is smaller sequence than h (k); it will be convenient to use following equation, © y(n) = Y h(k)x(n-k) Q) k=-0 Here sange of ‘k’ will be as per the sequence h(k). Thus range of ‘k’ is from 0 to 3. Calculation of y(0) : Putting n = 0 in Equation (2) we get, 3 y(0) = h(k)x(-k) k=0 Expanding the summation we get, y(0) = h(0)x(0)+h(1)x(-1)+h(2)x(~2)+h(3)x(-3) y(O) = (2x1)+(1x0)+(2«0)+(1x0) y(0) = 2404040 was x(-1)=x(-2)=x(-3)=0 2 y¥(0) = 2\ DSP. (E & Tc) 2442 Signals and Systems Calculation of y(1) + Putting n= 1 in-Equation (2) we get, 3 y(1) = h(k)x(1-k) k=0 Expanding the summation we get, YC) = h(O)x(1)+h(1)x(0)+h(2)x(-1)+h(3)x(-2) y(1) = (2x0)+(1x1)+2x0)+(1x0) y(1) = 0+1+0+0 wa X(+1)-=x(-1)=x(-2)=0 Calculation of y(2) : Putting n = 2 in Equation (2) we get, 3 yQ) = Y h(k)x(2-k) _ k=0 Expanding the summation we get, ¥(2) = h(0)x(2)+h(1)x(1)+h(2)x(0)+h(3)x(-1) ¥(2) = (2x0)+(1x0)+(2x1)+(1x0) as x(2)=x(1) = x(-1) =0 y(2) = 0404240 Q) Calculation of y(3) : Putting n = 3 in Equation (2) we get, = y(Q3) = Y h(k)x(3-k) k=0 Expanding the summation we get, ¥(3) = h(O)x(3)+h(1)x(2) +h(2)x(1)+h(3)x(0) ¥(3) = (2x0)+(1x0)+(2x0)+(1x0) ¥(3) = 0404041 8x (3)=xQ)=x(I)=0 [ye = 4] 3) Step Step Step 1DSP. (E & TC) 243 Signals and Systems Step IV : Now the result of convolution can be written as, y(n) = {y(0),¥(1),¥(2),¥3)} sy@)= 2120 t Note :, Here given sequence x(n) = {1}. We know that (1) is the unit impulse and is expressed as, 8(n) = {1}. That mearis we have calculated convolution of { 2, 1, 2, 1} with unit impulse. + Observe the result of convolution. It is the same as { 2, 1, 2, 1}. Thus we can conclude that the convolution of any sequence with unit impulse produces the same sequen 3) Tabulation method of linear convolution : This is the simplest method of performing linear convolution. Let x(n).= {x(0),x(1),x(2)} and h(n) = {co bC1),42)] t Step I: Form the matrix as shown in Fig. 2.71 : x(n) 1 x) X@) Fig. 2.71 : Matrix of x(n) and h(n) Note that we can interchange the positions of x(n) and h(n). Step Il: Multiply the corresponding elements of x(n) and h(n) as shown in Fig. 2.72. x(0) xQ) xQ) h(O) | h(O)x(0) h(0)x(1) h(0)x(2) hq) | h(1)x(0) -h(1) x(1) h(1)x(2) h(2) | h(2)x(0)-h(2) x1) h(2) x(2) Fig, 2.72 ? Multiplication of elements x(n) and h(n) Step III : Separate out the elements diagonally as shown in Fig, 2.73.DSP. (E & TC) 2144 Signals and Systems x) x) x@Q) 4] bOxQ = nx, - OK), : ed ¢ “7. - on hd) HQ) a og 7 Biaee - - - 12) “hex “nO, - “bxQ 2 2 Fig. 2.73 : Diagonally separation of elements Step IV: Simply add the elements in that Particular block. This addition gives corresponding values of y(n). Herg bh B ‘Wis: ¥ = xX, +h=0+0=0 andy, = a, +h=2+2=4 ‘Thus we get, ¥(0) = h(0)x(0) y(1) = h(1)x(1)+h(0)x(1) y(2) = h(2)x(0)+h(1)x(1)+h(0)x(2) ¥(3) =h(2)x(1) #h(1)x(2) and y(4).= h(2)x(2) After calculating all these values; write down the result of convolution as, y(n) = 1¥(),¥C1),9(2),93),9¢4)] Prob. 10 : Compute the convolution y(n) =x(n)*h(n) where x(n) = (1,1, 0, 1, 1} and h(n) = (1, ~2, -3, 4) ‘ T Wee. 2001, DSP (B&TC) Soln, : Here x(n) contains ‘5* samples and h(n) has “4” samples, To make the length of x(n) and h(n) same; rewrite the sequence h (m) as follows : h(n) = {1,-2,-3,4,0] t We can add zeros atthe end or at the beginning of sequence. This will not affect the given Sequence. This method of adding zeros, to adjust the length of sequence is called as zero-padding Range of ‘n’: ‘The range of ‘n’ for y(n) is calculated as follows : Lowest index of y(n) => y,=x,+h,=-2+(~3) =DSP. (E & TC) 2-145 Signals and Systems and Highest index of y(n) => y, =%,+h,=2+1=3 Thus y(n) varies from y(-5) to y(3). Now using tabulation method the convolution is obtained as shown in Fig, 2.74. x0) [xl 1x1 -2x1 -2x1 -3x1 -3x1 4x1 4x1 Oxl Oxl Fig, 2.74(b) : Convolution of x (m) and h(n) yGD yo) yS3) y2) Fig. 2.74(¢) : Diagonally separation of elements Different values of y (1) are calculated by adding corresponding elements as follows : y(-5)= 1 y(-4) = -241=-1 | y(-3) = -3-2+0=-5 \ y(-2) = 4-3+0+1=2 y(-1) = 04+440-2+1=3DSP. (E & TC) 2-146 Signals and Systems y(0) = 04+0-3-2=-5 y(1) = 044-31 y(2)= 0+4=4 y(3) = 0 Thus result of convolution y (n) is , y(n) = {y(~5).¥(-4),¥(-3),¥(-2),9(- 1), (0), ¥ (1), 9 (2), 9 (3)] + y(n) = {1,-1,-5,2,3,-5,1,4,0} i ‘Neglecting the last ‘0° term we have, y(n) ={1,-1,-5,2,3,-5,1,4} t Prob. 14: Compute y(n) = x(n)*h(n) ifx(n) =h(n) = (1, 2, -4, 3} t [May 2003, DSP E&TO)} Soln. : Range of n: Lowest range of y(n) >y,=x,+h,=0+0=0 Highest range of y(n) => y,=%,th, =3+3=6 Using tabular method; the convolution is obtained as shown in Fig. 2.74(d).. Fig, 2.74(d) : Linear convolution of x(n) and h(n) y(n) = (14,2, 2, 13, -6, 9} t Prob. 12: The impulse response of linear time invarient is h(n) = (1,2, 4,-1} t Pro SoinDSP. (E & TC) 2447 Signals and Systems Determine the response of system to the input. x(n) {1, 2, 3, 1} 1 ~t [May 2001, DSP (E&TO)] Soln. : The response of the systett is, : > y(a) = x(n)*h(n) Range of n: Lowest of y(n) = y,=x, i Highest range of y(n) => y,=x,+h, = Using tabulation method; the convolution is obtained as shown in Fig. 2.74(e) +3=6 a oO § e > s 5 © e oe Fig. 2.74) : near convolution of x(n) and h(n) y(n) = (1, 4,8, 8,3, -2,-1} t Use discrete convolution to find the response to the input x(n) of the LTI system with impulse response h(n) for the corresponding x(n) shown in Fig. 2.75(a) and (b). Prob. 13 : ) Fig, 2.75 : Given sequence x(n) and h(n) Soin. : (From Fig. 2.75(a) the sequences x(n) and h(n) can be written as,DSP. (E & TC) 2148 Signals and Systems x(n) = {1} and h(n) = {2, 1} t + Making lengths equal we have, x(n) = {1, 0} and h(n) = (2,1) + t x(0) = 1 co h(O) =2 x(1) = 1 h(1) =1 Range of ‘n’ for y(n) = Lowest index of n => y,=x,+h,=0+0=0 Highest index of n> y, =x, +h, =1+1=2 Thus y(n) varies from y(0) to y (2) By using tabulation method, convolution of x(n) and h(n) is obtained as shown in Fig. 2.76 x(n) ©. Fig. 2.76 : Convolution of x(n) = {1, 0} and h(n) = (0, 1} t t Different values of y(n) are obtained by adding comesponding terms. y(O)= 2° y(1)=1+0=1, y(2)=0 ‘Thus result of convolution is, y(n): = {y(0),y(1),¥(2)} t y(n) = {2, 1,03 t ‘Neglecting last term we have, y(n) = Q, 1} ths DSP. (E & TC) 2449 Signals and Systems (i) From Fig. 2.75(b) we can write sequences x(n) and h(n) as, x(n) = (1-1) and h(n) = E120) t t Making lengths equal we have, x(n) = {1,-1,0} and h(n) = 1,2, t t x(0) = 1 2 h(O) == x(1) = -1 h(l) =2 and x(2) = 0 and h(2) =1 Range of ‘n’ for y(n) : Lowest index => y, = x, +h, = 0 +0=0 Highest index => y, = x, +h, = 2+2=4 ‘Thus y(n) varies from y (0) to y(4). By using tabulation method; convolution of x(n) and h(m) is obtained as shown in Fig, 2.77. 1 -1 0 xq) (1x1) 1x1) (-1%0) (2-1) (2x0) (x-1) (1x0) ) © Fig. 2.77 : Convolution of x(m) = {1,—1, 0} and h(n) = {-1, 2, 1} t tDS.P. (E & TC) 2.450 Signals and Systems ( = La Different values of y(n) are obtained by adding corresponding terms y(0) = -1 y(i) = 2+153 y(2) = 1-25-1 y(3) = -140= y(4)= 0 ‘Thus result of convolution y(n) is, y(n) = {70.91.99} y(n) = {~1,3,-1,-1,0] ‘Neglecting last ‘0’ term we can write, 4) Multiplication method of tinear convolution : This is another easy method to. obtain convolution of two sequences. This mettiod is similar to multiplication of multidigit numbers. Write down the two sequences, x(n) and h(n) and obtain the multiplication by usual method, The result of multiplication will be equal to the convolution of two sequences, We will solve one example by this method, Prob. 14 : Obtain linear convolution of following sequences x(n) = (1,2, 4,2)andh(n) = @,2,-4, 1} t t Soln. : First we will decide the range of ‘n’ for y(n), Lowest index => y, = x,+h,=~-1+(=2)=~3 Highest index => y, = x,+h, = 2+1=3 Thus y(n) varies from y(~3) to y(3) We will solve this problem using multiplication method, x 1 2 1 2 Multiplication wa) -@Q2 2 1 121 2 € x(n) is muttiplied by (1) dition 3@ “1-2 -L -2 G3 x(a) is mutiplieady 1) % Zero digit similar to normal multiplication Og = x(n) is multiplied by (2) = two zero digits similar to normal ‘multiplication A, € In) is mules by (2) Three zero digits similar to normal multiplication These values are obtained by adding “all digits in corresponding columns. dditimno>@ 242 4 dditim »® 24 2 4 Sty ee)DSP. (E.& Tc) 2151 _ Signals and Systems ‘This result of multiplication is same as lifear convolution. We know that range of y (m) is from y(~3) to y(3). So the first term from LHS, represents the value of y (~3). Now mark the arrow at the position of y (0). Thus output y.(n) can be written as, Let us verify this‘result using tabulation method shown in Fig. 2.78. x(n) | 1 2 1 2 2 (ant) (2x2) (2x1) (2x2) 2 |ioxy (2x2) (2x1) (2x2) jor (-1x2) (-1e1) (=1%2) TP} (od) (1x2) (ied) (1x2) @ ) ¥G3) ¥O2) yOI) yO) © Fig. 2.78 : Convolution of x(n) = {1, 2, 1,2) and h(n) = 2, 2, ~ 1,1} t ‘Thus using tabulation method we get; y(-3) = 2 y(-2).= 244-6 y(-1) = -14442=5 y(O) = 1-24244=5 y(1) = 2-144=5 y(2) = 1-2=-1 yQG)= 2 '¥ (1) = {2, 6, 5,5, 5,-1, 2) tDSP. (E & TC) 2452 Signals and Systems 2.7.4 Properties of Linear Convolution : =>, ASKED IN EXAM - MAY 2003 I!! J We have discussed how to obtain the convolution of two sequences. In this sub-section we will study some important properties of linear convolution. These properties are : 1. Commutative property 2. Associative property 3. Distributive property 1. Commutative property : Statement : It states that Tinear convolution is commutative operation, That means x(n)*h(n) =h(n)*x(n). Proof: The linear convolution of x(n) and h(n) is given by, © y(n) = x(n)eh(n)= YD x(kyh(n-k) ~() k=-@ In this equation h(n—) is folded delayed sequence; while x (k) is unchanged. Put_m=n-k as new index of summation in Equation (1). k= a-m Limits will change as follows : When k=-0 >n-m=-0 +,-m=-o-n When k=+0 > n-m=0 “-M=o-n Putting these values in Equation (1) we get, y(m) = Y x(n-m)h(m) Q) m=+0 In case of summation operation limits from +0 to —co are same as — to +00, Thus changing limits of summation we get, © y(n) = x(n-myh(m) ~-@) m pation (3), m is dummy variable. It can be replaced: by any other variable. So replacing am’ by ‘k’ in Equation (3) we get, y(n) =) SY x(n-kyhn(k) wu(4) © In this equation x(n-k) is folded and delayed sequence; while sequence h(k) is aichanged. As per the definition of convolution Equation (4) can be written as,DSP. (E & TC) 2-153 Signals and Systems ra y(n) = Y h(k)-x(m-k)=h(n)*x(n) (5) k= ‘Ths from Equations (1) and’ (5) we get. y(n)=x(n)*h(n 6) Hence proved that linear convolution is commutative, Graphical Presentation : Graphically it is presented as shown in Fig. 2.79(a). Input Co oe Output = <=> Input Output x(n) y(n) = x{n) * hin) hin) y(n) = h(n) * x(n) Fig. 2.79(a): x(n) *h(n) = h(n) *x(n) 2. “Associative property : Statement : It states that [x(n)*h, (n)]*h,(m)=x(n)*[h,(n)*h,(n)} Explanation : We know that ‘*” indicates convolution operation. Every convolution operation indicates that certain input is applied to the LTI system and output is the convolution of input signal and the impulse response of system. Now consider L.H.S: term. That is the term, [x(n)*h, (n)]*h, (n). Let this output is y(n) : y(n) = [x(n)*h, (n)]*h, (0) Al) There are two convolution operations. The bracket term indicates one convolution of x(n) and hy (1). Let this oufput be denoted by y, (1). Thus Equation (1) becomes, y(n) = y,(n)*h,(n) -Q) ‘The convolution y, (n) = x(m)* hy (1) indicates that, we are applying input to LTI system whose impulse response is h, (n). The output of system is y, (n). Such system is represented in Fig. 2790). Input: Output x(n) ¥4(n) = x(n) * y(n) Fig. 2.790) : yy (m) = x(m)*hy (n) Now according to Equation (2); y(n) is obtained by computing convolution of y, (m) andDSP. (E & TC) 2.154 Signals and Systems 4, (1), Here y, (1) acts as input signal to the LTT system whose impulse response is h, (m). The output of this system is y(n). Diagrammatically it is represented as shown in Fig. 2.79(0), Input out yy(n) e y(n) = y4(n) * helm) Fig. 2.79(¢) :y(m)= yy(0)*h, (mn) Thus to implement Equation (1) we have to connect (cascade) two systems shown in Figs. 2.79(a) and 2.79(b). This cascade connection is shown in Fig. 2.79(4), Now consider the RH.S. term x(n)*[h,(n)*h,(n)]. We can design one LT! system whose impulse response h(n) is the convolution of h, (1) and h, (1). That means convolution of ‘two impulse responses. Thus h(n) = h, (m)*h, (n). x(n) *Eh,(n)*h,(n)] = x(n) *h(n) 0) ‘The diagrammatic representation of Equation (3) is shown in Fig. 2.79(e). Input (n)* hein) Output x(n) yén) = x(n) * [hy(n) * h(n] Fig. 2.79(¢) Meaning of associative property : This property indicates that we can replace cascade combination of LTI systems by a single system whose impulse response is convolution of individual impulse responses. It is represented in Fig 2.790). h, h(n) = y(n) * a(n) yin) Y¢0) = fxn) + hyn] fn ln) = a(n) * hale) * helm} Fig. 2.79(f) : Meaning of associative property o logsa com ThisDSP. (E & TC) 2155 Signals and Systems 3. Distributive property : Statement : It states that linear convolution is distributive. That means x(n)*[h,(n)+h,(m)]} = (x(n) *h, (n)]+[x(n) *h, (a)] (1) Proof : Consider the R.H.S. term of Equation (1). We will denote it by y (n). y(n) = [x(n)*h, (n)]+[x(n) *h, (n)] 2) Every bracket term in Equation (2), indicates the convolution operation. Thus we will denote output of each convolution as follows : y,(n) = x(n)#h,(n) @ and y,(n) = x(n)*h, (n) A) y(n) = y,(m)+y, (0) 6) Observe Equations (3) and (4). It indicates that there are two LTI systems whose. impulse responses are h,(n) and h,(n), Important thing is that, the same input signal x(n) is applied to both systems. While Equation (5) shows that we have to add the outputs of these two LTI system, This process is represented diagrammatically as shown in Fig. 2.80(a). ¥4(0) = x(n) # hy(n) x(n) pea yin) = van) + yalm) ya{n) = x(n) * ha(n) Fig, 2.80(a) : y(n) = y, (mn) +y,(m) = [x(m) *h, (n)] +1 x(n) * hy (0)] ‘As shown in Fig. 2.80(@), the same input is applied to both systems. So this is a parallel connection of two LTI systems. Now as shown in Fig. 2.80(a); two impulse responses are added. h(n) = hy (n)+h,(n) 6) Thus we can design a single LTI system having impulse response h(n) = hy (n)+h,{n). ‘This will perform the same operation performed by the system shown in Fig, 2.80(a). This system is represented in Fig. 2.80(b). I fron nee] a y(n) = x(n) * h(n) Fig, 2.80(b) : y(n) =x(n)*h(m) = x(n)*[h, (1m) +b, (n)]DSP. (E & TC) 2-186 Signals and Systems Meaning of distributive property : This property indicates that, we can replace a parallel combination of LTI systems by a single LTI system whose impulse response is the sum of individual impulse responses. It is represented as shown in Fig. 2.80(¢) : Input Output x(n) o> 4 i) x(n)— h(n) = hy(n) + ho(n) yin) yO) Fig. 2.80(c) : Meaning of distributive property Summary of properties : Table shows the’ summary of properties of linear convolution. Sr. | Name of property Mathematical equation No. Commutative x(n)*h(n)=h(n)*x(n) 2. | Associative [x(n)*h, (n)] *hy(n) = x(n)*[h,(n)*h,(n)] 3. | Distributive x(n)*[h, (n)+hy(a)] = (x(n) *hy (n)]+[x(n) #hy(n)] 2.8 Stability of LTI (or LSI) System : Earlier we have discussed the stability’ condition for discrete time system. It states that a discrete time system is stable if bounded input produces bounded output. Here the meaning of word “bounded” is; input and output signals should have some finite magnitude. ‘The magnitude of these signals should not be infinity. Now in this section we will derive the stability condition applicable to LTI systems; in terms of unit sample response, Proof of stability criteria for LTI (or LS!) systems in terms of unit impulse response : Statement : LTI system is stable if its impulse response is absolutely summable, Proof: Consider a linear time invarient system having impulse response h(n). Let x(n) be the input applied to this system. Now according to the definition of convolution; the output of such system is expressed as, val vah oper:DSP. (E & TC) 2457 -___ Signals and Systems y(n) = x(k) h(n-k) Al) k=-@ “Bawation (1) can also be waiten as, © : y(n) = Y h(k)x(n-k) -Q) k=-0 Now first we will define bounded sequence. A discrete time sequence is said to be bounded if the absolute value of every clement is less than some finite number. Thus input x(n) will be bounded if |x(m)| is less than some finite number. Let us denote this finite mmber by M,. Thus for input signal to be bounded. Ix(m)| s M, ~-Q) Here M, is a finite number, so definately its value should be Jess than infinity, Thus Equation (3) can be written as, |x(n)| < M,<® AA) Now taking absolute value of both sides of Equation (2) we get, es ly(@l= | DD h(k)x(n-k) 5) k=-0 We will read RLS. of Equation (5) as “absolute value of summation of terms”’. if we take es “DY” sign outside then the term becomes Sy |h(k)|]x(n-K)|, This is summation of absolute k=-0 values of terms. Always absolute value of sum of terms is less than or equal to the sum of absolute values of terms. © © Thus} YS h(k)x(m-k)| < Y [h(k)I]x(n-k)] 6) k=-0 k But from Equation (5); the LLH.S, term is fy (m)_ ly} sD Uh) Px(n-k)| @ ~o Here x(n-k) is delayed input signal, If input is bounded then its delayed version is also bounded. This is because delay or folding is related to time shifting operations. By performing these operations; the magnitude is not changed. Now for bounded input we have, Ix(a)}'s M,D.S.P. (E & TC) l 2-158 Signals and Systems Ix(n-k)| Present input x(n-1),x(n=2), «. => Delayed inputs that means past inputs. and x(n+1),x(n+2),.. => Advanced inputs that means future inputs We discussed that for the system to be causal we want only present and past inputs. We do not want future inputs, Thus second bracket in Equation (4) should be zero to make the system causal Here x(n+1), x(n+2) «. are future inputs, We do not want to make input to the system zero, So the second bracket is zero if we adjust h(~1), h(~2), h(-3) .. to be equal to zero. ‘Then in this case we will have only first bracket terms. That means we will have only present and past input terms, And then the system will be causal. ‘Thus for causality, h(-1) = h(-2) = h(-3)= h(n)=0 forn<0 (6) Modification of convolution equation : =0 ) ‘That means, We have the equation of linear convolution,DSP. (E & TC) 2-460 Signals and Systems eo y(n). = SY) x(kyh(n-k) OR y(n) = Yo h(k)x(n-k) k=-0 We have derived that, the LTT system is causal then h(m) = h(k) = 0 for n (or k) <0. ‘Thus if LTT system is causal then the limits of summation will be from k = 0 to +00; because for negative values of ‘K’, h(k) is zero. Thus for causal LTI system the equation of linear convotution is, 0 0 y(a)= YS x(k)h(n-k)} OR fy(n)= Y h(k)x(n-k) k=0 k=0 Prob. 1: Determine range of values of parameter ‘a’ for which the linear time invariant system with impulse response. h(n) = a” n20andn even = 0 otherwise is stable [May 2002, DSP (E&TC)] Soin. + We know that LTI system is BIBO stable when, o DY |h(k)| 0 and n even and h(n) = 0 otherwise, That-means sequence h(n) is present from n = 0 to n = co and for even values of n only. So limits of summation will change from n = 0 to n = co, We will also replace h(k) by h(n). Thus Equation (1) becomes, . « LD tgyi= Z pay 0) k=-0 pai ee Since ‘n’ is even, to satisfy this condition we will put n= 2p. Here value of p is 0, 1, 2 .... 00, So ‘n’ is always even for any value of ‘p’. Thus Equation (2) becomes, © o YX tnage= Y i © p=0 k = Y lap p=0 IN & y@ (rea digiDSP. (E & TC) 261 Signals and Systems ‘Expanding the summation we get, o EY th) = al lal tla tars. © = 1+fal+faP+jaP.. ‘ «Q) This is geometric series. For geometric series we have, eo D # = ttatdsas... a if|aj<1 n=0 Thus Equation (3) becomes, DX tkdol= 7p Hlal |a| <1. Thus for the system to be stable |a] <1 is the condition. Otherwise the system will be unstable. if and only if 2.10 Difference Equations : Until now we have studied différent characteristics of linear time invariant (LTD) system, We haye also discussed that the output of discrete time system can be calculated using equation of linear convolution. o y(n) =} x(k)h(n-k) ww (1) k=-0 Here h(n) is the impulse response of system. By the use of h(n); we can calculate output y(n) for any input x (1) using Equation (1). Now in this section we will discuss the difference equation which are useful to implement (realize) the discrete time system. These difference equations are especially used for designing the digital filter. We discussed that impulse response h(n) is used to calculate output of discrete time system. Now if the system is causal then impulse response h(n) is zero for negative values of n. So for causal LTI system equation of convolution becomes, y(n) = Y x(kyh(n-k) ~-Q) k=0 Equation (2) can also be written as,D.S.P.(E & TC) 2-462 ‘Signals and Systems © y(n) = ¥ h(k)x(n-k) 3) k=0 Since output depends on impulse response h(n); the number of samples, present in h(n) plays an important role to determine output y(n). The number of samples present in h (n) is called as length of impulse response. Depending upon the length of impulse response; there are two types of-discrete time systems : 1. Finite impulse response (FIR) systems. 2. Infinite impulse response (IIR) systems. 4. Finite impulse response systems (FIR) : ‘As the name indicates; impulse response h(m) is having a finite length. That means h(n) contains some finite number of samples. Let the range of h(n) is from -M to +M. Thus equation of convolution for FIR system becomes, M y(n) = Y h(k)x(n-k) — (Non-causal) a) k=-M Note that in Equation (4), negative terms are present, so this equation is for the non-causal system, Now for the causal FIR system we will consider only positive samples of h(n). Thus range of h(n) becomes ‘0’ to M instead of -M to M. Equation (4) becomes, M y(a) = h(k)x(n-k) 8) k=0 Note that the summation is taken from 0 to M, So there will be total M+1 terms in h(n) or h(k). These terms are : (0), RCL), h(2) wh (M) .., Number of terms in h (1). ‘Now expanding Equation (5) we get, y(n) =h(0)x(n)+h(1)x(n-1)+h(2)x(n-2)+ .. th(M)x(n-M) (6) In Equation (6) x(n) is present input and x(n-1), x(m~2)... etc. are past inputs. Every input is multiplied by corresponding value of h(n). Thus Equation (6) indicates that output y(n) is weighted sum of inputs Note that in this case the system has to view input samples only from x(n) to x(n-M) and not all input samples. These input samples are most recent input samples. Thus this system has to store x(n-1), x(n=2) ... x(m-M), past samples in the memory. Since these are finite number of samples; FIR has limited or finite memory requirement, 2. Infinite impulse response (IIR) systems : ‘As the name indicates, impulse response’ h(n) contain ‘co’ samples. That means length of h(n) is infinite. Now the equation of convolution for non-causal IIR system becomes. lo 8S3eGH8 CcDSP. (E & TC) 2.163 Signals and Systems o y(n) = Y h(k)x(n-k) —_..(Non-cansal) 0) k=-0 For the causal system, limits of summation will be from 0 to «, a y(n) = Y h(k)x(n-k) ... (Causal) : (8) k=0 Here the range of h(n) = hk) is as follows : h(0), h(1), h(2) ..., h(co). So this is infinite impulse response system. ‘Now expanding Equation (8) we get, y(n) = h(O)x(n)+h(1)x(n=1) +h (2)x (0-2) + M()(D=) ~O) In this case TIR system has to store all past samples from x(n-1) to x(m—v). Thus TTR system needs infinite memory. This is practically not possible systems using convolution equation. But IIR systems cannot be designed using convolutos equation, This is because of the infinite memory requirement. So to implement such IIR Note: We discussed that FIR system needs limited memory. So it is possible to design FIR systems; difference equations are used. Now depending upon length of h(n) we have classified the systems as FIR and UR, Similarly depending upon output of system; the discrete time systems are classified as follows = 1, Non-recursive systems 2. Recursive systems. Why this type of classification is required 7 Equation’ of convolution indicates that for causal systems, the output depends on present and past input as well as it depends on the impulse response, h(n). But this is not the case always, ‘There are some systems whose present output depends on present and past inputs as well as past (previous) output, For such systems we should take info account, the previous output 10 calculate present output. This type of classification gives the idea about the ‘requirement of previous output 10 calculate present output. 4. Non-recursive systems : Non-recunsive systems do not require any past output sample to calculate the present output. Consider the equation of FIR causal system (Equation (5)). This equation is, M y(n) = h(k) x(n-k) k=0 To calculate present output we need present and past inputs only as shown by Equation (6). We will rewrite Equation (6) as follows : y(n) = h(0)x(n)+h(1)x(n-1)+...+h(m)x(n-M)DS.P. (E & TC) 2-164 Signals and Systems This equation docs not contain any previous output term like y(n-1), y(n=2) ... ete, So this is the equation of non-recursive system. Thus we can conclude that causal FIR system is non-recursive system, In other words we can say that non-recursive implementation of FIR causal system is possible, ‘Now for non-causal FIR system; the limits of summation will be from -M to M. Thus this system contain present, past and future inputs. But it does not contain future term. So this is also non-recursive system. In case of non-recursive systems; past output terms are not required. Past (previous) output terms are obtained by taking feedback from output to the input. Thus non-recursive system do not require any feedback, Is it possible to implement non-recursive IIR system 7 We know that, Non-recursive <> No feedback or no past output term present and IR <2 Infinite impulse response. ‘The equation of convolution for causal IIR. system is, 5 y(n) =) h(k)x(n-k) k=0 Expanding the summation we get, y(m) = h(0)x(n)+h(1)x(n-1)+h(2)x(n-2).... +h(o)x(n—o) We have discussed earlier that such systems require to store all past inputs. That means it is Tequired to store infinite input terms in the memory. So infinite memory is required. This is not possible practically. Thus non-recursive implementation of causal IR system is not possible. The representation of non-recursive system is shown in Fig. 2.81(a). : i x00) — eee yn) ik) x(n =k) Fig, 2.81(a) : Non-recursive system 2, Recursive Systems : A discrete time system in which output y(n) depends on present input, past inputs as well as previous outputs; is called as recursive system. Let us consider one example, Consider a system represented by block schematic as shown in Fig. 2.8100)DSP. (E & TC) 2465 Signals and Systems mlz ay (1) vin) Fig, 2.81(b) : Example of recursive system First we will write the equation for system. (@) As shown in Fig. 2.81(b); a connection is drawn from output y(n) to the input side. This is a feedback Gi) ‘This feedback signal, y(n) is passed through an unit delay (z') block. So the output of delay block is y (n—1). This is past output compared to y(n). The signal y (n~ 1) is multiplied by constant multiplier ‘a’ to obtain ay (n—1). (iv) An adder is used to add input x (m) and signal ay(n-1). (v¥) Now the addition of x(n) and ay(n-1) produces the output, Thus the output of system is expressed as, - y(n) = x(n)+ay(n-1) (10) ~ At instant n = 0 we get, (il) y(O0) = x(0)+ay(-1) Here y(0) => present output x(0) => present input and y(~1) => past output Similarly at n = 1 we get, y(1) = x(1)+ay (0) ‘Thus y (0) is past output So at any instant present output depends on past output. Hence it is recursive system. The representation of non-recursive system is shown in Fig. 2:81(C). yin) Fig, 2.81(c) : Representation of recursive systemDSP. (E & TC) 2-166 Signals and Systems 2.10.1 LTI Systems: Characterized by Constant Coefficient Difference Equations : In the previous section, we discussed that, the behaviour of system depends on its impulse response, In this subsection we will discuss how LTI systems are characterized by input-output relations; called as difference equation with constant coefficients. Consider a recursive system shown in Fig. 2.81(b). The equation of output is, y(n) = x(n)+ay(n-1) (1) This is first order difference equation. Here ‘a’ is a constant coefficient. Now output y(n) is calculated by putting different values of n in Equation (11). Forn=0 => y(0)=x(0)+ay(~1) ~@ Forn=1 =>y(1)=x(1)+ay (0) wai) Putting value of y (0) from Equation (i) we get, y(1) = x(1)+a[x(0)+ay(-1)] y(1) = x(1)+ax(0)+a¥y(-1) Rearranging the terms, y(1) = ay (-1)+ax(0)+x(1) alii) Forn=2 =>y(2)=x(2)+ay(1) _ (iv) Putting Value of y( 1.) from Equation (ii) we get, y(2) = x(2)+aly (-1)+ax(0)+x(1)1 y(2) = x(2)+aby(-1)+a?x (0) +ax(1) Rearranging the terms, y(2) = ay (-1)+a?x(0)+ax(1)+x(2) auf) Observe Equations (iii) and (v) carefully. Now we can directly write the equation for value y(n) = tly (<1) +a"x (0) 4a? x (1) +...4ax (n= 1)+x(n) (12) In terms of summation Equation (12) can be written as, 1 y(n) = atty(-1)+ D akx(n-k) (13) k=0 Note that we have obtained output y(n) by putting only positive values of n. That means for n> 0. So these are the equations of causal system. Every equation contains the term y(~1). It is assumed that, we are starting the operation of system at n= 0. Then the term y(~1) is called as initial condition of a system. Important terms related to Equation (13) : We know that Equation (13) is representiig the behaviour of causal system. It consists of two parts as follows :DSP. (E & TC) 2-167 Signals and Systems @ The first term, ¥(~1) is called initial condition of the system, Gi) The second term, is the Tesponse of system to the input x (n). Now depending upon initial conditions, two types of responses are obtained, (1) Zero state response or forced response : If initial condition, y (1) of the system is zero then the Tesponse of the system to the input signal is called as zero state response. It is denoted by y,, (1). In this case, whatever output we are getting, that is because of input x(n) only, And since the system is forced to the input; it is also called as forced response, Since y(~1) = 0; the system is said to be relaxed system. ‘Thus putting intial condition, y (—1) = 0 in Equation (3) we get, n y(m)=y, (n)= DY &x(n-k) . for n> 0 (14) k=0 (2) Zero input Fesponse or natural response : AS the name indicates; input is not applied to the system and inital condition y(-n) #1, ‘Thus, this kind of outpot without applying input x(n) is called as ero input response or natural by me Y(-n) # 1, this is called as non-elaxed initial condition. This oupit is dented byy,;, Putting x(n) = 0 in Equation (13) we get, This equation indicates that output is not dependent on input x(n ) but it depends on initiat condition, y(—1) (3) Total response : The total response of system is obtained by adding zero state response and. ero input response. “+ Total response, y(n) = ¥,i(n) +y,, (n) (16) Equation (11) is basically a linear constant coefficient difference equation. The general form of this equation can be expressed as, N M 2X ay(n-ky+ DY oyx(n-k k=0 12)\ DSP. (E & TC) 2-168 Signals and ‘Systems Here a, and bare constant coefficients. And ‘N’ is the order of difference equation. That means ‘N’ represents the order of system. Stability, linearity and time invariance property of recursive systems + We will discuss these properties of recursive systems described by constant coefficient difference equations. (1) Stability : A recursive system is table when we get'a bounded output by applying bounded input. (2) Linearity : A reousive system desribed by Equation (17) is suid t0 be linear if it satisfies following conditions : : {The otal response of system is adtion of zero input response and zero state response, y(n) = y(n) 9,08) (i) Zero state response should be linear that means principle of superposition should be applicable. (ii) Zero input response should be linear that means principle of superposition should be applicable. 3) Time invariance: As shown by Equation (17), the output y (n) contains constant coefficients a, and b,. These coefficients are not affected by the change in time (shifting). So the recursive system is time invariant. 2.10.2 Solution of Linear Constant Coefficient Difference Equation : In this subsection we will discuss how to obtain exploit solution of a given linear constant coefficient difference equation. Basically there are two methods to obtain this solution. ‘These methods are as follows : (1) Direct method (2) Indirect method Indirect method is based on z transformation. So will study that method in the later chapter. Presently we will discuss direct method for obtaining the solution. In the direct method; it is assumed that the solution consists of two parts : (1) Homogeneous or complementary solution. It is denoted by y, (1). 2) Particular solution. It is denoted by y, (7) ‘Thus the total solution is expressed as, y(n) = y,(a)+y,() (18) (1) Homogeneous solution of difference equation : Recall general form of difference equation (Equation (17)), Itis, N M yin) = - Day a-k)+ LD bx(n-k) (19) k=l k=0DSP. (E & TC) 2-469 Signals and Systems In the first term the limits of summation are from k = 1 to N. We can change these limits from k = 0 to N as follows. Take the limits from k = 0 to N then, N N LD ay(n-k) = ay(ay+ YL gy (n-k) k=0 k=1 N N - Y ay(n-k) = |y(n)- LY ay n-k) (20) k=l k=0 Pulting this value in Equation (19) we get, . N M y(n) = ay(n)~ D gy (n-k)+ Y x(n-k) x k=0 k=0 Assume ay = 1 N M y(n) = y(n)- Y ay(n-k)+ D byx(a-k) k=0 k=0 N M D ay(n-k) = >» bx (a-k) : (21) k=0 k=0 We will obtain homogeneous solution by assuming input x(n) = 0. Since x(n) = 0 > x(n-k) = 0, Thus Equation (21) becomes, N D ay(n-k) = 0 (22) k=0 ‘We will assume that the solution is in the form of exponential. ‘Thus let, homogeneous solution, bor = ® (23) Replace n by n-k y,(n-k) = 207K Putting this value in Equation (22) we get, N Dra eae BO (24) \ k=0 Expanding the summation we get, ayeea wea 24 ra aeN = 0DSP. (E & TC) 2-170 ‘Signals and Systems We have assumed that ay = fe Maa Pea te ta eN = 0 Taking 2°~™ common we get, NLR aM oa, DELIA gg PP oN tata] = 0 oN aMaa Naa N84 tay | = 0 08) ‘The term inside the bracket is called as characteristic polynomial of the system. This polynomial has ‘N° roots denoted by Ay , Ay... Ayy. These roots can be real or complex valued. Case 1: Roots are distinct : ‘When the roots are distinct then we can write the solution as, (26) Here ¢, , 6)... Gy are constants, Case TI : Multiple roots : In case of multiple roots; the solution can be written as, - at naan Yy(M) = OM teal Hey AE. +6 hy Solved Problems : Prob. 4: Determine the homogeneous solution of the system described by y(n)~3y(n-1)~4y(n-2) = x(n). Soln. : Step I: Assume input x(n) = 0 y(n)~3y(n-1)-4y(n-2) = 0 ~ n(l) Step II: Assume the the homogeneous solution is in the form y,(n) = 2 0) Thus we get, y(n-1) = a8! and y(n-2) = 27? Step III :Putting these values in Equation (1) we get, a3 1-44n-? = 9 B) ‘Taking X"~? common we get, 3B a[a. arattggp-dgcnt2_g 4] wat a3n- 4] =0 B-34-4DSP. (E & TC) 2474 Signals and Systems Now obtain roots of Equation (4) using the formula, ~ba Vor 4ac roots = Here a=i,b=-3andc 2385 he 2x1 2 2 3-5 345 d= yo andy 7 a, = Land 4, =4 Step IV : The homogeneous solution is given by, y(m) = 9 neh oY, (0) = (1p ae -) a Now we will obtain values of c, and cy. Step V : From Equation (5) we get, Forn=0 >y(0)=%+¢, (6) ang forn=1 > y(1)=~% 440, (1) Step VI: Now take original equation, y(n)-3y(n-1)-4y(n=2) = 0 y(n) = 3y(n-1)+4y(n-2) (8) Forn=0 >y(0 sy (- 1) +4y(-2) 9) Forn=1 = y(1)=3y(0)+4y(-1) (10) Putting Equation (9) in Equation (10) we get, - yQ) = 3b ay (1) +4y (-2)] +4y(- 1) es 9y (= 1)+ By (-2) 44y (1) y(1) = 1By(-1)+1y(-2) 1) Step VEL: Equte equations in step V and VI. That means equating Equations (6) and (9} we get, oy +e, = 3y(-1)+4y(-2) 12) ‘And equating Equations (7) and (11) we get, mo, t4e, = 1By(-1)+12y(-1) 13) Now we will solve Equations (12) and (13) to obtain c, and C>. ‘Adding Equations (12) and (13) we get,DSP. (E & TC) 2.472 Signals and Systems 5c, = l6y(-1)+6y(-2) AA) Putting this value in Equation (12) we get, oy Fy (1) +8 y(-2) = ay(—1) 449-2) = 3y(-1)-- By 1) 445 (-2)- 8 y(-2) ; 4 gy) +4y(-2) 03) Step VIII : Putting these values in Equation (5) we get, y(m) = [-s2-pegrea car] Bsenelfyca ar 06 If we assume initial conditions zero that means, y(~1) = y(-2) =0 then Equation (16) becomes, Ya) = (-1 +4)" fl?) Note : To obtain this solution we have assumed input x(n) = 0. So this output is also called as Zero input response. Prob. 2: Find the impulse response of the system shown in Fig. 2.82 ir) Fig, 2.82 : Given discrete time system Soin. : Impulse response means, the response (output) of system when input x(n) = 8(n). Using homogeneous solution we can find the impulse response, assuming that the system is initially relaxed, First we will write the difference equation from given Fig. 2.82. At the output of first delay we get y(n-1) and it is multiplied by + While at the output of second delay we get y ( n-2) Ste\ D.S.P. (E & TC) 2473 Signals and Systems and it is multiplied by i ‘These two outputs are added with input signal x(n). y(n) = x(n)-Sy(n-1)4gy (2-2) © Now we will find out the homogencous sotution. Step I: Assume input x(n) = 0. Thus Equation (1) becomes, y(n 3y(n-1)-fy(a-2) = 0 ® Step II: Assume that the homogeneous solution is in the form y(n) = 8 -@) ‘Thus we get, y(n-1) = "7! and y(n-2)= 297? Step III :Putting these values in Equation (2) we get, wad grt hye? =0 Taking 2"~? common we get, a) ay Step IV : The homogeneous solution is given by, y, (im) = eh teh 1 aon = a(t} econ We will denote y,,(n) by h(n)DSP. (E & TC) 2-474 Signals and Systems nny=e,(§) +6,(-1) 5) Now we will obtain values of c, and c, Step V+ We will apply initial conditions. In this case x(n) = 8(n) and y(n) = h(n). Putting these values in Equation (1) we get, h(n) = 3¢n)-Zh(n-1) +E n(a-2) Forn=0 =>h(0)=8(0)-Zh(-1)+2h(-2 Here 6( 0) = 1 and since initial conditions are zero thus h(~1) = 0 and h(~2)=0 h(0) = 1-3x0+2x0 h(O)=1 Forn=1 =>h(1) =5(1)-2n(0)+2(-1) Here 8(1) = 0,h(-1) =0 and h(0) = 1 - = 0-Sx142 h(1) = 0-2x143x0 a hcy=-2 Step VI : Putting n = 0 and n= 1 in Equation (5) we get, ° Forn=0 =>¢,{1) +¢,(-1)°=n¢0) 116 | *% 1 | +c,(-1)'=h(1) (7) We will obtain values of ¢, and ¢, From Equation (6) (2) san Im, @ Gi)D.S.P.(E & TC) 2475 Signals and Systems Putting this value in Equation (7) we get, -1+¢, Putting c, and c, in Equation (5) we get, =a) =4(6) +8" This is impulse response of given system. (2) _ Particular solution of difference equation : ‘The particular soluion is denoted by y, (nn). It is calculated by assuming that the output has same form at that of input, For example if input x(n) = u(n) then the output is assumed to be Yp(n) = ku(n); where‘k’ isarbitrary constant. Important rules to assume the solution : @ If any term ina particular solution (yp) is also present in homogeneous solution then multiply ‘y,” by n Ef such term appears ‘r’ times in a homogeneous solution then multiply ‘yg’ by Some commonly used input signals and the particular solution that we have to assume is shown in following table Sr | Input x(n) Particular solution = No. 1._| Constant term (A) - 2 facn) ku(n) 3. [Ae ke 4] apne byt? tk, 5. [Acosoyn "| ky c0s0)n+k, sinognDSP. (E& TC) 2-476 Signals and Systems ‘A particular solution is basically used to calculate the zero state response (y,,) of the system, ‘The following steps are used to calculate the forced response of system : Step I; Find the nature of zero input response from the roots of characteristics polynomials. This is y,(n) ory,,. Step If: Assume the particular solution %, (n) Step III : Determine the coefficients in the zero input response. Then the forced response Yyy = YC) ty, CB). Solved Problems : Prob. 1: Find out the particular solution for the following differential equation y(n)+3y(n=1) = x(n) Assume x(n) = u(n) ‘Solin. : Here input x(n) = a(n) = unit step Now corresponding fo this input, we will assume the particular solution as, y,(n) = ku(n) () Herek is some constant. Now the given equation is, y(n)+#3y(n-1) = x(n) - Q) From Equation (1) we can write, y(n-1) = ku(n-1) @) Putting Equations (1) and (3) in Equation (2) we get, ku(m)+3kn(n—1) = a(n) -@) Now we will calculate the value of k, We know that input is unit step, u(m). Its value is 1 for n> 0, Putting n = 1 in Equation (4) we get, ku(1)+#3ku(0) = u(1) Remember that while putting value of ‘a’ any term should not vanish. But u(0) = u(1)=1 k+3k " 4k = l>k=4 Potting this value in Equation (1) we get, y,(a)=fu(n) Prob. 2: Determine the particular solution of the difference equation, y(n) = 8y(n-1)-4y(n-2)+x(n) When the forcing function x(n) = ", n> 0 and zero everywhere. seD.S.P. (E & TC) 2477 Signals and Systems Soln. = Refer the table; for input x(n) = 3" we will assume the particular solution, yp (a) = ke ~l) ‘Now the given input x(n) = 2" is present only for positive value of ‘n’ (n20). Thus we canwrite, x(n) = Pu(n) ~Q) Accordingly Equation (1) is modified as, Y¥_(m) = ka n(n) -G) From Equation (3) we can write, y(n-1) = k2°"' (n= 1) and y(n-2) = k2"7?u(n=2) Putting these values in the given equation of y(n) we get, k2u(n) = 3k2"7) u(n-1)-4k 2°"? u(n-2)+2"u(n) (4) Now we will calculate value of k. Given input is present for n> 0. But while putting,value of ‘a’ any term in equation should not vanish. For example if we will put n = 1 then the second term of RHS. becomes, ak PF y(n=2) 4 (2)! a(-1) But u(- 1) = 0; so this value of ‘n’ is not allowed. Thus we will put n= 2 in Equation (4) so that any term will not vanish. k?u(2) = 3k (2)! a (1) ~ 4k (2)?u(0) #27 0 (2) Here u(0) = u(1)=u(2)=1 4k = 6k-4k+4 2 ks4 > k=2 Putting this value in Equation (3) we get, spiny 2000 | Note : If total solution is asked to calculate, then first obtain y,(} and y,(n) separately. The total solution is y(n) = yy(n)+Y,(n). 2.41 Correlation : Basics of correlation : ‘The word co-relation means comparison, Basically it is comparison between two signals or sequences. A good application of correlation is the radar system.