my cn a Leathe . S Poornachandra SCE CYL eI}Za iz uw Tata McGraw Hill . Published by the Tata McGraw Hill Education Private Limited, 7 West Patel Nagar, New Delhi 110 008. Digital Signat Processing, 3e Copyright © 2010, by Vijay Nicole Imprints Private Limited. No part of this publication may be reproduced or distributed in any form or by any mems, electronic, xechanical photocopying. recording. or otherwise or stored in a database or retrieval system without the prtor written Petsission of the publishers and eqpyright holders. The program listings (if any) may be entered. stored and executed it a Cpye, system, but they may not be reproduced for publication. This edition can be exported from India only by the publishers, ‘Tata McGraw Hill Education Private Limited, ISBN(13}: 9780-07.067279-6 ISBN(10): 0-07-067279-2 Information contained in this work has been obtained by publishers, from sources believed to be reliable, However, neither publishers nor copyright holders guarantee the accuracy or completeness of any information published herein, and neither publishers nor copyright holders shall be responsible for any errors. omissions. or damages arising out of use of this information. This work is published with the understanding that publishers and copyright holders are supplying information but are not attempting to render engineering or other professional services. If such services are required, the assistance of an appropriate professional should be sought. Laser Typeset at: Vijay Nicole Imprints Private Limited, Chennai - 600 042 Printed at: Pooja Graphics, Chennai - 6040 014 ‘ ROXCRROFDRBRR : se eho sContents Preface xv HM Introduction to Digital Signal Processing 1.1 What is DSP? 1DP-1 EE Introduction to Signals and Systems 2.1 Intreduetion to Modeling 1SS-1 21.1. Signals Iss-1 2.1.2 ne-dimansional Signal 155-2 2.1.3 Two-dimensional Signal 188-2 2.1.4 Multisdimensional Signal 155-3 24.5 Sampling 155-3 2.6 Quantization sn 24.7 Coding ISS-4 2.2 Classification of Signals IssS 2.2.1 Continuous-time Signal and Discrete time Signal ISS-5 2.2.2 Periodic and Aperiogic Continuous-time Signal 185-9 22.3 Evenand Odd Continuous-time Signal 158-16 2.2.4 Energy Signal and Power Signal 158-25 2.2.5 Deterministic Signal and Random Signal 155-31 2.3 Basic Operations on Signals 1SS-32 2.3.1 Amplitude Scaling of Signals 155-32 23.2 Addition of Signals 1SS-34 2.3.3 Multiplication of Signals 155-35 23.4 _ Differentiation on Signals 1SS-36 2.3.5 Integrationon Signals 1S$-38 23.6 Time Sealing of Signals 1SS-39 2.3.7 Reflection of Signals 155-41 23.8 Time Shifting of Signals 155-41 2.3.9 Time Shifting and Time Scaling ISSA 2.4_ Types of Signals Iss-48 2.4.1 Exponential Signal 1SS-48 2.4.2 _ Sinusoidal Signal (Gontinuous-time) 1SS-50 243° Step Function 15-53 24.4 Impulse Function 188-53 2.4.5 Ramp Function Iss-54 25 System ISS-55 2.6 _ Properties of Systems 1SS-56 2.6.1 Continuous-Time and Discrete-Time System 185-56 26.2 Stable and Unstable System 1SS-56 2.6.3 Memory andMemoryless System 1SS-63 2.6.4 _ Invertible and Noninvertible System 188-64 2.65 Time-invariant and Timo-variant System 155-646 © Digital Signal Processing 266 Linear and Nonlinear System. 2.6.7 Causal and Noncausal System 2.7__Interconnection of Systems Chay ‘Summary EBL Li Systems 3.1_Discrete-time Linear Time-invariant System Uh 3.1.1__ Representation of Discrete-time Signals in Terms of Impulses ura 3.1.2 Convolution Sum 3 3.2 Properties of TI System us 324 utive Property LTS 3.22 Associative Property U6 3.2.3 Commutative Property rd 3.3_Properties of Discrete-time LTI system Ur? 3.3.1 LTISystem with and without Memory um? 3.3.2 Invertibility of LTI System ny 3.3.3. Stability for LTI System ume 3.3.4 Causal System U9 3.4_Linear Convolution IL-9 3.5__Linear Convolution using Cross-Table Method LTLa7 3.6 Linear Convolution using Matrix Method L418 3.7 Step Response LT1-33, 3.8_Deconvolution TB 39 Basic Systems LT1-36 3.10 Linear Constant Coefficient Difference Equation LT1-37 3.10.1 Solution to Linear Constant Coefficient Difference Equation LT1-37 3.11 Introduction te Correlation TS 3.11.3 Properties of Cross-correlation and Autocorrelation LT1-54 Chapter Summary L755 Revlon Quoc Ts Fourier Series. 4.1. Continuous-time Fourier Series FSS 4.4.4 Fourier Series Representation of Continuous-time Periodic Signal Fss.2 4.1.2 Continuous-time Fourier Series Representation of Real Signal FSS-3 4.1.3 Fourier Series Representation of a Continuous-time Signal FSS-4 4.1.4 Trigonometric Representation of Continuous-time Fourier Series FSS-13 4.2_Properties of Signals FSS-21 42.2 Linearity Property FSS-21 42.3 Time Shilting Property FSS-22 4.2.4 Time Reversal Property FSS-23 4.25 Time Scaling Property FSS-23 4.2.6 Multiplication Property FSS-24Contents @ T 4.2.7 _ Conjugation and Conjugate Symmetry Property ESS.24 4.2.8 Parseval’s Relation for Continuous-time Periodic Signal ESS.25 4.2.9 _ Differentiation Property ESS.25 4.2.10 Integration Property ESS.26 4.3 Convergence of Fourier Series ESS.26 4S Gibb’s Phenomenon FSS-28 4.6 Fourier Series Representation of Discrate-time Periodic Signal ESS.29 4.61 To Calculate Discrete-time Fourier Series Coefficient EsS-20 4.7 Properties of Diserete-time Fourier Series ESS.33 4.7.1 Linearity Property ESS:33 4.7.2 Time Shifting Property ESS 4.7.3 "Time Reversal Property ESS-34 47.4 Time Scaling Property . FSS-35 47.5 Mubtiplication Property ESS.35- 4.7.6 Conjugation and Conjugate Symmetry Property ESS:36 4.7.7 Parseval’s Relation for Diserete-time Periodic Signal ESS.37 4.7.8 _ Difference Property ESS.37 Running Sum (Integration) Propert ESS-38 Chapter Summary FSS.39 5.2_Representation of Aperiodiic Signals FIM-1 5.3 Dirichlet Condition FTM-5 5.4 Spectrum Analysis FMS 5.5 Properties of Continuous-time Fourier Transform FIM-12 5.5.1 Linearity Property FIM-12 5.5.2 Time Shifting Property FTM-13 . Frequency Shifting Property FTM-44 ‘Time Reversal Property FIM-15, Time Sealing Property FTM.15, Multiplication Property FIM-16 Conjugation and Conjugate Symmetry Property TFIMAy Porsoval's Relation for Continuous-time Aperiodic Signal FIM-18 Differentiation in Time Property FIM.18 Differentiation in Frequency Property FIM-19 Integration Property FIM-19 5.5.12 Convolution Property FYM-20 5.6 Solution to Differential Equation using Fourier Transform FIM-24 5.7 Fourier Transform—Introduction M8 5.8 Representation of Discrete-time Aperiodic Signals FIM-28 5.9 Properties of Discrete-time Fourier Transform FTM.37 5.9.1 Linearity Property FIM -36 5.9.2 Time Shifting Property FTM-388 @ Digital Signal Processing EeeE E BE 5.18 5.20 5.9.3 Frequency Shifting Property ETM-39 5.9.4 Time Reversal Property ETM.39 5.9.5 Time Scaling Property . ETM-40 5.9.6 Multiplication Property ETM-40 5.9.7 Conjugation and Conjugate Symmetry Property ETM.41 5.9.8 Porseval's Relation for Discrate-time Periodic Signal cM-a2, 5.9.9 _ Difference in Time Property ETM.a2 5.9.10 Difference in Frequency Property EIM-43 5.9.11 Accumulation Property ETM-43 5.9.12 _ Convolution Property EIM-44 5.9.13 Convergence af the Fourier Transform EIM-45 System Characterized by Linear Constant Coefficient Difference Equation FTM-48 Band Limited Signals EIM-S5 5.11.1_ Mathematical Analysis EIM-55 5.11.2 The Ideal Bandpass Signal ETM-56 Parameter Estimation of a Band Limited Signal EIM:S7 5.12.1 Signal Eneray ETM.58 ‘Orthogonal Band Limited Signal EIM.s8 Discrete Fourier Transform EIM-59 Frequency Analysis of Discrete-time Signal EIM40 Properties of DFT EIM-73 5.16.1 Periodicity EIM:73 5.16.2 Linearity . EIM-73 5.16.3 Circular Shift of a Sequence (Time-domain) M23 5.16.4 Time Reversal of the Sequence IM-25 5.16.5 Circular Shift (Frequency-domain) ETM-74 5.16.6 Complex-conjugate Properties EIM-77 $16.7 Circular Correlation M27 5.16.8 Parsoval's Theorem EIM-27 5.16.9 Symmetry Properties of the DFT EIM-27 5.16.10 Multiplication of Two DFT ETM-78 = 5.16.11 Multiplication of Two Sequences EIM-22 5.16.12 Convolution of Two Sequences ETM.79 Circular Convolution ETM-80 5.12.1 Circle Method FTM.80 §.12.4_DFTIDFT Method ETM.84 5.18.1__Overlap-save Method ETM.£9 5.18.2 _Qverlap-add Method £1M.92, Fast Fourier Transform (FFT) EIM:98 5.19.1 _Radix.2 FFT Algorithm EIM.99 5.19.2 Decimation-in-time FFT Algorithm ETM.99 5.19.3 _Decimation-in-frequency FFT Algorithm ETM-107 Inverse Fast Fourier Transform (IFFT) FTM-115Contents @ 9 5.21 Radix-4 FFT Algorithm FTM-120 5.21.1 Divide and Conquer Technique for DFT Computation FTM-120 EPL7 The Rndied Docmationtntraa FT FrMa2 5.21.3 _Radix-4 Decimation-in-frequency FFT FTM-122 5.22 FFT Algorithm in Linear Filtering and Correlation FTM125 Chapter Summary FTM-127 Review Questions 130 Kl 2-Transform : 6.1 Introduction ZIM-1 62_Z-Transform ZIM 63 Region of Convergence (ROC) ZIMA 63.1 Relationship between Z-Transform and DTFT ZIM4 $.4__2-Transform of Finite Sequence ZIM-11 64.1 _ Right-hand Finite Sequence ZIM-A1 &42 Left-hand Finite Sequence ZIM-2 64.3 Two-sided Finite Sequence ZIMA4 65 Characteristic Features of Signals with their Corresponding ROC Z™M-14 5.1 __ Proporties of Region of Convergence (ROC) ZIMA6 66 Properties of Z-Transform ZIM-16 66.1 __ Linoarity Proparty ZIMAT 6.6.2 Time Shifting Property ZIM-T 6.3 _ Time Reversal Property ZTMA8 66.4 Time Scaling Property 7 ZIM-16 6.6.5 Multiplication Property ZIMA9 6.6.6 Conjugation and Conjugate Symmetry Property ZTM-20 6.8 Difference in Time Property Z1M.21 6.6.9 Differentiation in Frequency Property ZIM-22 6.6.10 Convolution Property ZIM-22 62 _ Initial Value and Final Value Theorem ZIM-23 6.8 Inverse Z-Transform 68.1 Power Series Method (Long-division) 68.2 Partial-fraction Method TD $8.3 ResiualMethod 00 ZT 6.8.4 Convolution of Two Signals Z1M.37 6.9 {TI System Characterized by Linear Constant Coefficient Difference Equation ZT™-40 6.10 Relationship between Z-Transform and Fourier Transform ZIM-44 6.11 Relationship between z-plane and s-plane ZTM-45 Chapter Summary ZTM-46 Review Questions \40.|\|\.0.- Mae40 @ Digital Signal Processing Finite Impulse Response Filter 21_Introduction RT 7.2_Phase Delay and Group Delay FIRA 7.3 Linear Phase Transfer Function FIRG 7.3.1 Symmetric Impulse Response with Odd Length FIRB 7.3.2. Symmetric impulse Response with Even Length FIR-0 7.3.3 Antisymmetric Impulse Response with Odd Length FIR-12 7.3.4 _Antisymmetric impulse Response with Even Length FIR-15, 7.4 Design of FIR Filter—Fourier Method FIR-18 7.5 Design of FiR Filter—Windowing Techniques FiR-32 7.6 The Triangular Window (Bartlett Window) FIR-37 7.2_Raised Cosine Window ra 7.8__Hanning Window FIR-50 7.9 Hamming Window FIRS? 7.10 Blackman Window 2.0 RB 7.11 Kaiser Window FIR-75 7.11.1 Kaiser Window Technique for High Pass Filter FIRT7 7.11.2 Kaiser Window Technique for Band Pass Filter FIR9 7.11.3 Kaiser Window Technique for Band Stop Filter FIR-80 7.12. FIR Filter Design Using Frequency Sempling Technique FIR-93 7.43 Equiripple Linear-Phase FIR Filter FIR-100 7.14 FIR Differentiators FIR-106 7.15 Hilbert Transform FIR-110 7.15.1 Advantages of FIR Filter over IIR Filter FIR-13, 7.15.2 Disadvantages of FIR Filter over IR Filtor FIR-113 7.16 Structure Realization of FIR Systern(Non-recursive System) FIR-14, 7.16.1 Direct Form Realization of FIR System . FIRA14 7.16.2 Cascade Structure Realization of FIRSystem FIR-1S 7.16.3 Realization Structure of Linear Phase FIR System FIR-118 7.16.4 Lattice Structure for FIR Fitter FIRA21 ‘Chapter Summary FIR-123 Infinite Impulse Response Filter 8.1_Introduction UR4 8.2 Analog Filters UR2 83 Analog Domain to Digital Domain Transformation UR4 8.3.1 Impulse Invariant Transformation Technique. IRA 8.3.2 Bilinear Transformation Technique IRS 8.4 Analog Frequency Transformation UR-12 44.1 Normalized Low Pass Filter to Desired Low Pass Filter Transfarmation ___IIR-12. 8.4.2 Normalized Low Pass Filter to Desired High Pass Filter Transformation MR-13 8.4.3 Normalized Low Pass Filter to Desired Band Pass Filter Transformation R13ROE BE 24 92 94 Contents. ital Frequency Transformation 8.5.1. LowPassto Desired Low Pass Filter Transformation 8.5.2 LowPassto Desired High Pass Filter Transformation 8.5.3 Low Pass to Desired Band Pass Filter Transformation: 8.5.4 Low Pass to Desired Band Elimination Filter Transformation Butterworth Filter Bal D ion of Order of the il Chebyshev Filter B7.1_ Magnitude Response of Type-| Chebyshev Filter 87.2_Mathematical Analysis 5 Reskzall . Implementation of Discrete-time System 8.9.1 Structure Realization of IIR System Recursive Structure 8.9.2 System Realization Using Direct Form-1 6.9.5 __ System Realization Using Parallel Form 8.10 Lattice Structure of IIR System 8.11 Lattice-Ladder Structure of IIR System Chapter Summary Review Questions | Analysis of Finite Word Length Effect Introduction Eixed Paint N G21 ‘Sign-magnitude Format 9.2.2 One's-complement Format 9.2.3 Two's-complement Format 9.2.4 — Fixed-point Addition 925 Fixed-point Multiplication Floating-point Numbers 931 JEEE Single Precision and Double Precision Format 9.3.2 Floating-point Addition 9.3.3 Floating-point Multiplication 9.3.4 Comparative Study of Fixed-point and Floating-point Representations 9.3.5 Dynamic Range, Resolution and Precision Product Quantization 9.4.1 Truncation Error in Fixed-point Representation 9.4.2 Rounding off Error in Fixed-point Representation 9.4.3 Quantization Error in Floating-point Representation 9.4.4 __ Truncation Error in Floating-point Representation 9.4.5 Rounding off Error in Floating-point Representation 9.4.6 Probability Density Function of Quantization Error Variance Estimation of Quantization Error 95.1 SNR Caleulation 9.5.2 Effect of Scaling on SNR en R-1S42 @ Digital Signal Processing 9.6 Finite Word Length Effect on lIR Filter FWL-16 9.6.1 Finite Word Length Effect on Filter Structure FWLA7 9.6.2 _ Effect of Finite Word Length on Direct Form-IStructure FWLA7 9.6.3 Effect of Finite Word Length on Direct Form-I! Structure FWL-16 9.7 Product Quantization Error in IIR Filter FWL9 9.8 Mathematical Analysis of Steady-state Output Noise FWL-22 9.9 Dynamic Scaling to Prevent Overflow FWL25 991 Alternate Method 97 9.10 _Limit-cycle Oscillations in Recursive Systems FWL30 9.11 Rounding off Error in DFT Computation FWL-34 9.12 Rounding off Error in FFT Computation FWL-36 Chapter Summary FWL.40 Review Questions 401 HE Random Signa! Processing 10.4 Introduction RSP 10.1.1 Probability Fun RSP 10.1.2 Joint Probability RSP-2 10.4.3 Conditional Probability RSP-2 102 Estimation Parameters RSP-3 Ae nn Funai Reba 102.3 Ensemble Averages RSP-4 102.4 Mean-squared Value RSP.5 10.2.5 Variance RSP-5 102.6 Standard Deviation RSP-S 10.2.7 Moments RSP-S 102.8 Correlation RSP-7 10.2.9 Covariance RSP-7 10.2.10 Orthogonality RSP-8 10.2.11 Autocovariance and Autocorrelation RSP-8 10.2.12 Cross-covariance and Cross-correlation RSP-9 10.3 Stationarity RSP-9 10.3.1. Strict-Sense Stationarity RsP.9 10.3.2. Wide-Sense Stationarity RSP-9 10.4 Ergodic RSP-10 10.5 White Noise RSP-10 10.6 Energy Density Spectrum RSP-10 10.7 Power Density Spectrum Estimation—Periodogram RSP-10 10.7.1. Power Density Spectrum Estimation RSP.11 10.8 The Periodogram RSP-11 10.8.1 Direct Method RSP.11 10.8.2 Indirect Method RSP-12 - 10.9 The Bartlett Method—Averaging Periodegrams RSP-15 10.10 TheWelchMethod RSP 10.10.1 Application—Analysis of Heart Beat Variability RSP.18 Chapter summary RSP-18 Review Questions 9° RSPContents @ 13 11.1 Introduction MDS-1 11.2 Decimator (Down-sampler) MDS.2 11.2.1 Frequency-domain Analysis of Decimator MDS-6 11.3 Interpolation (Up-sampling) MDS.8 11.4 Sampling Rate Converter MDS-11 11.4.1 Mathematical Analysis of Sampling Rate Conversion MDS-12 11.4.2 The Noble Identities for the Up-sampler and Down-sampler MDS-13 14.5 Polyphase decomposition : Mps.14 11.5.1_Efficient Transversal Structure for Decimator MDS-16 11.5.2 Efficient Transversal Structure for Interpolator MDS.18 11.6 Multistage Design of Decimater and Interpolator Mmps-19 11.2 Comb Filter MDS.23 11.8 Subband Coding MDS.24 11.9 Quadrature Mirror Filter MDS:25 11.9.1 Two-channel OMF Bank MDs-25 14.9.2 Mathematical Analysis of Two-channel OMF MDS.26 Chapter Summary MDS.28 Review Questions MDs.22 Introduction to Speech Processing W21_Introduction SP 12.2 Production of Speech Waveform ISP.1 12.3 Speech Recognition ISP-3 12.3.1 Training Stage SP. 12.3.2 Recognition Stage ISPS 12.4 Linear Predictive Coding Model for Speech Synthesis, 1SP-4 12.5 The Cepstrum: A Method for Speech Analysis IsP-6 125.1 Subband Coding of Speech Signals Vocoder sP.8 12.6 Transmuttiplexers 1SP-10 126.1_FDM to TOM Transmultiplexer ISP-10 12.6.2 TDM to FDM Transmultiplexer ISP-11 Chapter Summary ISP.11 Digital Signal Processor 18.4 Introduction SPA 13.2 Programmable Digital Signal Processing DsP-2 13.3 Multiplier Accumulator DSP-2 13.3.1 Overflow and Underline in MAC Unit DSP.3 13.4 Computer Architecture DSP. 13.4.1_Von-Ni in_ Architecture DSP-5 134.2 Harvard Architecture SP 1243 Modified-Harvard Architecture. DSP-614 © Digital Signal Processing 13.5 On-chip (cache) Memory DsP-7 13.6 Pipelining DsP-7 13.7 Pipeline Structure Processor DSP-11 13.7.1 Single instruction Multiple Data Architecture DsP-12 13.7.2 Very Long Instruction Word Architecture osP-12 13.7.3 Superscalar Processing Architecture DsP-14 13.8 Computer Configuration DSP-15, 13.8.1. Restricted Instruction Set Computer DsP-15 13.8.2 Complex instruction Set Computer DsP.15, * 13.9 Addressing Modes: DSP-16 13.9.1 Immediate Addressing DSP-16 13.9.2 Absolute Addressing DSP.16 13.9.3. Accumulator Addressing DSP-17 13.9.4 Direct Addressing DsP-17 13.9.5 Indirect Addressing DsP.17 13.9.6 Circular Addressing DsP-18 13.9.7 BitReversal Addressing. DSP-18 43.9.8 Memory-Mapped Register Addressing DSP.18 139.9 Stack Addressing DSP-18 13.9.10 Replication DSP-19 13.11_First Generation TMS320C1X Processor DSP-20 Dsp.22 13.11.2 Tho Arithmetic and Logie Unit 18.12_Second Generation TMS320C2X Pracessor CO SP24 TMS320C3x Digital Signal Processor 413.13 13.13.1_Architecture P28 13.14 TMS320C4X Digital Signal Processor Dsp-31 13.141 Architecture DSP.31 1314.2, Memory Architecture DSP-34 13.14.3 Internal Bus Operation DSP-36 13.14.4_ CommunicationPort SPT 13.145 DMACoprocessor DsP-37 13.15 TMS320C5X Digital Signal Processor DSP-37, 13.15.1_Central Processing Unit (CPU) DSP-39 3.15.2 _On-Chip Memory . DSP-40 13.15.3 On-Chip Peripherals (DsP-41 13.15.4 Addressing Modes DsP=42 13.155 TMS320C54XX Instructions Dsp.42 13.16 TMS320C6X Digital Signal Processor DsP-46 13.181 TMS320C64X Processor DSP-47 13.162 TMS320C67X Processor 05P-49 13.17 Code Composer Studio DsP.50 ‘Chapter Summary DsP-51 Review Questions DsP-57Preface ‘This book Digital Signal Processing, 2e is designed for a one-semester course on the subject for students of engineering at the undergraduate level. The book has its emphasis on ease of comprehension, lucid explanation of concepts, numerous examples, a solved problems approach and simple presentation. ‘The first four chapters of the book provides an introduction and reeap of the topics on signals and systems. Students who have studied Signals and Systems asa core paper may decide to skip these chapters A significant highlight of this book is the treatment and coverage of topics on Finite Impulse Response Filters (FIR) and Infinite Impulse Response Filter (IIR). It reckons for students who find the subject tough and provides numerous examples with explanations. Similarly, the topic Finite World Length Effect has its emphasis on clear concepts and a simple and easy to understand presentation. The coverage of the topic has a prerequisite that the students are familiar with number and decimal systems. The topic Multirate Signal Processing has been discussed with necessary mathematical treatment. The basie concepts are explained in simple English to facilitate better comprehension. This textbook has a crisp and clear introduction to Estimation Theory starting from Estimation Parameters to Model Estimation. ‘The field of Digital Signal Processing has its impact on all areas of technology and science. It is of equal importance to industry and academia. In the engineering curriculum, this subject is now offered to students of electronics, electrical, communication, IT and-computer science , steams, We hope this book will serve as a basic resource for all students of engineering. We thank the management and staff members of our respective institutions for all their support and help, We thank Mr. P K Madhavan and othors of Vijay Nicole for their officientand tireless efforts in publishing this book. We welcome all objective criticisms and suggestions on the book. Dr S Poornachandra B SasikalaCopyrighted maierialIntroduction to Digital Signal Processing @ #11 WHAT IS DSP? - DSP or Digital Signal Processing, as the term suggests, is the processing of signals by digital means. A signal in this context means a source of information, In general terms, a signal isa stream of information representing anything from stock prices to data from a remote-sensing satellite, The signal here means an electrical signal carried by a wire or telephone line, or perhaps by a radio wave. In many cases, the signal is initially in the form of an analog electrical voltage or current, produced for example by a microphone or some other type of transducer. In some situations the data is already in digital form—such as the output from the readout system of aCD (compact disc) player. An analog signal must be converted into digital (i.e. numerical) form before DSP techniques can be applied. An analog electrical voltage signal, for example, can be digitized using an analog-to-digital converter (ADC). An analog signal on sampling results in a discrete signal followed byIDP-2 @ Digital Signal Processing ion and encoding in order to convert the discrete signal to digital signal. This generates a digital output in the form of a binary number whose value represents the electrical voltage input to the device. Signals need to be processed in a variety of ways. For example, the output signal froma transducer may be contaminated with noise. The electrodes attached to a patient’s chest when an clectrocardiogram (ECG) is taken, measure tiny electrical voltage changes due to the activity of the heart and other muscles. The signal is ofien strongly affected due to electrical interference from the mains supply, electromagnetic interference, muscle artifacts, ete. Processing the signal using a filter circuit can remove or at least reduce the unwanted part of the signal. Nowadays, the filtering of signals to improve signal quality or to extract important information is done by DSP techniques rather than by analog electronics. The development of digital signal processing dates from the 1960s with the use of mainframe digital computers for number-crunchjng applications such as the Fast Fourier Transform (FFT), which allows the frequency spectrum of a signal to be computed rapidly. These techniques were not widely used earlier because suitable computing equipment was available only in leading universities and other scientific research institutions. The introduction of the microprocessor in the late 1970s and early 1980s made it possible for DSP techniques to be used in.a much wider range of applications. However, general-purpose microprocessors such as the Intel x86 family are not ideally suited to the numerically-intensive requirements of DSP, and during the 1980s the increasing importance of DSP led several major electronic manufacturers (such as Texas Instruments, Analog Devices, and Motorola) to develop Digital Signal Processor chips—specialized microprocessors with architectures designed specifically for the types of operations required in digital signal processing. (Note that the acronym DSP can variously mean Digital Signal Processing, the term used for a wide range of techniques for processing signals digitally, or Digital Signal Processor, a specialized type of microprocessor chips). Like a general-purpose microprocessor, a DSP is a programmable device, with its own native instruction code. DSP chips are capable of carrying out millions of floating point operations per second, and like their better-known general-purpose cousins, faster and more powerful versions are continually being introduced, DSP technology is commonly employed nowadays in devices such as mobile phones, multimedia computers, video recorders, CD players, hard dise drive controllers and modepis, and will soon replace analog circuitry in TY sets and telephones. An important application of DSP is in signal compression and decompression. In CD systems, for example, the music recorded on the CD is in a compressed form (to increase storage capacity) and must be decompressed for the recorded signal to be reproduced, Signal compression is used in digital cellular phones to allow a greater number of calls to be handled simultaneously within each local “cell”. DSP signal compression technology allows people not only to talk to one another by telephone but also to see one another on the screens of their PCs, using small video cameras mounted on the computer monitors, with only a conventional telephone line linking them together. Although the mathematical theory underlying DSP techniques suchas Fast Fourier transform, Wavelet transform, Hilbert transform, Digital filter design and Signal compression can be fairly complex, the numerical operations required to implement these techniques are in fact very simple, consisting mainly of operations that could be done on a cheap four-function calculator. The architecture of a DSP chip is designed to carry out such operations incredibly fast, processing upto tens of millions of samples per second, to provide real-time performance, that is, the ability to process a signal “live” as it is sampled and then output the processed signal, for example, to a loud speaker or video display. All the practical applications DSP mentioned earlier, such as hard disc drives and mobile phones, demand real-time operation. In signal processing, the function of a filler is to remove unwanted parts of the signal, such as random noise, or fo extract useful parts of the signal, such as the components lying within a certain frequency range. ‘There are two main kinds of filters, analog and digital. They are quite different in their physical makeup and in their working. An analog filter uses analog electronic circuits made from components such as resistors, capacitors and op amps to produce the required filtering effect. Such filter circuits are widely used in applications such asIntroduction to Digital Signal Processing @ IDP-3 noise reduction, video signal enhancement, graphic equalizers in hi-fi systems, and many other areas. There are well-established standard techniques for designing an analog filter circuit for a given requirement. At all stages, the signal being filtered is an electrical voltage, or current, which is the direct analog of the physical quantity (example, a sound or video signal or transducer output) involved. A digital filter uses a digital processor to perform numerical calculations on sampled values of the signal. The processor may be a general-purpose computer such asa PC, ora specialized DSP (Digital Signal Processor) chip. The analog input signal must first be sampled and digitized using an ADC. The resulting binary numbers, representing successive sampled values of the input signal, are transferred to the processor, which carries out numerical calculations on them. These calculations typically involve multiplying the input values by constants and adding the products together. Ifnecessary, the results of these calculations, which now represent sampled values of the filtered signal, are output through a DAC (digital to analog converter) to convert the signal back to analog form, Note that in a digital filter, the signal is represented by a sequence of numbers, father than a voltage or current. ‘The main advantages of digital filters over analog filters are listed below. 1. Adigital filter is programmable, that is, its operation is determined by a program stored in the processor's memory. This means the digital filter can easily be changed without affecting the circuitry (hardware). An analog filter can only be changed by redesigning the filter circuit 2 Digital filters are easily designed, tested and implemented on a general-purpose computer or workstation, 3. The characteristics of analog filter circuits (particularly those containing active components) are subject to drift. and are dependent on temperature. Digital filters do not suffer from these problems, and $0 are extremely stable with respect to both time and temperature. 4. Unlike their analog counterparts, digital filters can handle low frequency signals accurately. As the speed of DSP technology continues to increase, digital filters are being applied to high frequency signals in the RF (radio. frequency) domain, which in the past was the exclusive preserve of analog, technology. 5. Digital filters are very much more versatile in their ability to process signals in a variety of ways; this includes the ability of some types of digital filter to adapt to changes in the characteristics of the signal, Fast DSP processors can handle complex combinations of filters in parallel or cascade (series), making the hardware requirements relatively simple and compact in comparison with the equivalent analog circuitry.Copyrighted materialCHAPTER Introduction to Signals and Systems @ 2,1 INTRODUCTION TO MODELING This book discusses signals and systems related to Engineering. It focuses on the modeling of physical signals and systems by mathematical functions, and the solution of such mathematical functions, when the system is excited by such signals. 2.1.1 Signals A signal is defined as a function of one or more variables which conveys information. A signal is a physical quantity that varies with time in general, or any other independent variable, It can be dependent on one or more independent variables. Dimension of a signal may be defined based on the number of independent variables.188-2 @ Digital Signal Processing Any variables which does not convey information is called Noise. Noise is arandom phenomenon in which physical parameters are time-variant, Unlike a signal, noise is usually does not carry useful information and is almost always considered undesirable, Some examples include channel noise in communication systems, transformer humming in electrical engineering and moving artifacts in biological systems. 2.1.2 One-dimensional Signal ‘When a function depends on a single independent variable to represent the signal, it is safd to be a ‘one-dimensional signal. The ECG signal and speech signal shown, 2.1(a) and 2.1(b) respectively are examples of one- dimensional signals where the independent variable is time. The magnitude of the signals is dependent variable. 80) 800 69) 600] . 40} $ 400] i i & £0 200) 2 -20 | 40 a -60 ~400| ‘0 80 100 160 200 250 300 350 400 450 ~%% 5 10 15 20 25 30 35 40 45 50 Samples Samples (a) () Fig. 2.1 One-dimensional Signal (a) ECG Signal (b) Speech Signal 2.1.3 Two-dimensional Signal When a function depends on two independent variables to represent the signal, itis said to be a two-dimensional signal. For example, photograph shown in Fig. 2.2 is an example of two-dimensional signal wherein the two independent variables are the two spatial coordinates which are usually denoted by x and y. Fig. 2.2. Two-dimensional PhotographIntroduction to Signals and Systems @ ISS-3 2.1.4 Multi limensional Signal When a function depends on more than one independent variables to represent the signal tabea ‘multi-dimensional signal, For example, space missile shown in Fig. 2.3 is an example of three-dimensional image. Fig, 2.3. 3D-Space Missile Definition Input signal A signal that enters a system from an external source is referred to as an input signal. For example, the voltage from a function generator, electrocardiogram from heart, temperature from the human body, etc. Output signal A signal produced by the system (may or may not be) in response to the input signal is called the output signal. For example, displacement due to force, output voltage from an amplifier, sinusoidal signal from an oscillator, ete. 2.1.5 Sampling Sampling is a process by which a continuous-time signal (continuous with respect to time) is converted intoa sequence of discrete samples, with each sample representing the amplitude of the signal at a particular instant ‘oftime, The sampling can be either uniform or non-uniform sampling. In uniform sampling, the space between any two samples is fixed throughout the signal under consideration uniform sampling s illustrated in Fig. 2.4, In nonuniform sampling, the space between any two samples varies throughout the signal under consideration based on their characteristics like frequency, etc. In general, uniform sampling is preferred over nonuniform sampling since it is simple to analyze and easy to implement. The hardware complexity is also Low in uniform sampling.1884 @ Digital Signal Processing Original Analog Signal Sampled Analog discrete) Sample Signal, Zin) 0123456789 ON 1213 415 16 Fig. 2.4 Uniform Sampting of signal 2.1.6 Quantization Quantization is a process by which each sample produced by the sampling circuit to the nearest level is selected froma finite number of discrete amplitude level as illustrated in Fig. 2.5 a peqeeeeeeqeceegeqeereqenees ‘Quantization, O(n) oN eres 0123 45 67 8 9 1011 1213 14 15 16 Fig. 2.5 Quantization of Signal 2.1.7 Coding Coding is needed in order to represent each quantized sample by a binary nuniber ‘0° or ‘I’. The ‘O” represents the “low” state or logical ‘0’, and ‘I’ represents the “high” state or logical *|’. The encoded version of quantized signal of Fig. 2.5 is showin in Table 2.1.Introduction to Signals and Systems @ ISS-5 Table 2.1 3-bit Quantization and its Binary Representation = © 1) 12 13 [4 [3 [6 17 [8 J9 |W [1 [mmm] i m |i {2 [5 |7 |7 |7 lo [5s |4 [4 [4 [5 [s [4 [4 [4 Js Bir inary | O01 | O10 | VOr [111 | 111 | 111 | nO] 101 | 100 | 100 [100 | tor [101 [100 | 100 [100 | 107] m . 2.2 CLASSIFICATION OF- SIGNALS. Signals are classified based on their fundamental properties. They are: ‘Continuous-time signal and Discrete-time signal Periodic signal and Aperiodic signal Even signal and Odd signal Deterministic signal and Random signal we YW ND Energy signal and Power signal 2.2.1 Continuous-time Signal and Discrete-time Signal Signal can be represented either by continuous or discrete values. Continuous-time signal A signal.x(¢) is said to be a continuous-time signal if itis defined for alltime z, The amplitude of the signal varies continuously with time. In general, all signals by nature are continuous-time signals. . The speech signal is a continuous-time signal, that is, conversation between persons is continuous with respect to time (Fig. 2.6a). 8 w 9 I: iu Liha mi I “a o 3 2 “ i ne a a ae) 0 an Fig. 2.6 (a) Continuous-time Signal Representation of Speech Signal (0) Discrete-time Signal Representation of Speech Signal1SS-6 @ Digital Signal Processing The electrocardiogram, which is the electrical representation of the cardiac muscle, is continuous with respectto time (Fig. 2.7(a)). ‘atro00 \ast000 800 ‘0 2 ne 00 150 0 50” Time 3) ) 700 150 Fig. 2.7 (a) Continuous-time Signal Representation of Electrocardiogram (b) Discrete-tinie Signal Representation of Electrocardiogram The sinusoidal signal as illustrated in Fig. 2.8{a) is continuous with respect to ‘time. wot wy 08. 08 oe os on on or 02 o ° 02 -02 04, -04 -08 -08 Fig. 2.8 (a) Continuous-time Signal Representation of Sirusoidal Signal (b) Discrete-time Signal Representation of Sinusoidal Signal Discrete-time signal Most of the signals that are obtained from their sources are continuous in time. Such signals have to be discretised since the processing done on the digital computer is digital in nature. A signal x(a) is said to be discrete-time signal ifit can be defined fora discrete instant of time (say n), Fora discrete-time signal, the amplitude of the signal varies at every discrete value n, which is generally uniformly spaced. A discrete-time signal x(n) is often obtained by sampling the continuous-time signal.x(() ata uniform or nonuniform rate, The discrete-time representation of speech signal, electrocardiogram and sinusoidal signal is shown in Fig. 2.6(b), 2.7(b) and 2.8(b) respectively.Introduction to Signals and Systems @ ISS-7 A continuous-time signal x(t) can be converted to discrete-time signal x(n) by substituting nT, that is, XC) = X(T) ag XE) - (2.1) n=0, 1, +2, +3... where = Constant integer, which can take positive or/and negative values isan imteger (normally Tis assume to be unity) T= Sampling perio SOLVED PROBLEMS» Problem 2.1. The continuous-time signal x(¢) = 5 sin( x1) for the interval 3 > ¢> 0. Plot the corresponding discrete-time signal with a sampling period T= 0.1 s. Solution x(1) = 5 sin( x) for3 > o[29[47[47 1291 0 [29/-47/-471-28] 0 [29147 [a7] 29 [0 The plot of continuous-time signal can be obtained by connecting each point by a line, as shown in Fig.2.9(). ‘The discrete-time signal can be obtained by a simple calculation given below: MO =XAT) buy (0) = x(0.17) | 20.1 x(1) = Ssin(n?) = Ssin(0.Ln7) walovTif2{;3 [4[ste[7][ sto (ny O TLS |29] 4 [48] 5 148] 4 | 2.9 | 15 15 La] The plot of the discrete-time signal is shown in Fig. 2.9 (b). 5 wo Fig. 2.9 (a) Continuous-time Signal (b) Discrete-time Signal188-8 © Digital Signal Processing Problem 2.2 The continuous-time signal x(/) = ¢ for the interval 2 > r > -2. Plot the corresponding discrete-time signal with a sampling period T= 0.1 s. Solution x(Q=e* for2>1>-2 (ry 2 [15] -1 [-os7 0 Tos 7 1 [is ][ 2 x9 [546[ 20 | 74 [27] 1 [037 [0.13 [0.05 [0.02 ‘The plot of the above continuous-time signal can be obtained by connecting each point by a line, as” shown in Fig. 2.10 (a). ‘The discrete-time signal can be obtained by (0 = 2007) beg 2(0) = 200.19) ho 4 x(n) = e720) ae n=O, +1, +2... —4|-3 |-2/-1/0]}] 1] 2]3 14 2.23} 1.82] 1.5 | 1.2 | 1 | 0.8 10.67] 0.5 [0.45] 0.37 ‘The plot of the discrete-time signal is shown in Fig. 2.10 (b). x 4x0) © 8 ” 2s w 2 7 1s 2 | : © as « —__—, | Ltii, = 1 4 2 t “5 4-3 -2 +41 jo 1 2 3 4 5S fh Y ® 3) Fig. 2.10 (a) Continuous-time Signal (b) Discrete-time SignalIntroduction to Signals and Systems @ ISS-9 2.2.2 Periodic and Aperiodic Continuous-time Signal A continuous-time signal x0) is said to be periodic if x(t) =x(¢ +7), T>0 (22) for all values of t, where T= period of a cycle, which is an integer value A= x(4 T= M(t+ 21) (t+ 37)= ...=x(t+ A) @3) where = any integer Hence, a periodic signal with period 7'> 0 is also periodic with period nT. Prove that the cosine signal is periodic with periodicity T Let us consider a cosine signal x(0) = A cos(wr +4) Let us assume period T which is same for all cycle in the given cosine signal xf!) = A cos (ait +6) 7 Fig. 2.11 Acosine signal a(E#7)= Acos {ar +7)+ 9} x(1+T)= Acos {art wl + ¢} U+T) = Acos {ar+2n+9} xU+T) = Acos{ar+o} = x00) The cosine signal is periodic as it satisfies periodicity equation (2.2).ISS-10 @ Digital Signal Processing SOLVED PROBLEM ___ ___-__._____—___———~—~——~~~~— Problem 2.3 Test whether the given signals are periodic or not. @ =e" (ii) x(Q= re Solution @ x=e"" From the definition of periodicity, x(1) = x(t+ 7) for T70 Substitute ¢ = (+ 7), xT) ser? Since T=2n, sin(e+ 1)= ‘Therefore, MEET) =e = es: Hence, the signal x(¢) = e™is peri Gi) x(N= re From the definition of periodicity, x(1) = x(¢+7) for 7>0 Substitute ¢ = (1+ 7), xU+T)= (ee me? SinceT=2n, — sin(¢ + T)=sin (¢+2n)=sin (4) Therefore, XCF TY (4 Temes (4 Ne 4 x09 Hence, the signal x(!) = eis aperiodic, Discrete-time pertodic signal A discrete-time signal is said to be periodic with period N, ifit is unchanged by atime shift of N, ie. x(n) = x(a+N), for all n (2.4) where ‘Vis a positive integer, Fundamental period ‘The fundamental period T, of the continuous-time signal x(0) is the smallest positive value of 7 for which equation (2.2) holds, Any signal x(/) for which there is no value of 7'to satisfy the condition of equation (2.2) s called an aperiodic signal. ‘The fundamental period WN, of the discrete-time signal a(n) is the smallest positive value of WV for which equation (2.4) holds. Any signal x(n) for which there is no value of N to satisfy the condition of equation (2.4) is called an aperiodic signal in discrete sense.Introduction ta Signals and Systems @ 1SS-11 SOLVED PROBLEMS—_—____»_»_>_s» Problem 24 ‘Test whether the given exponential is periodic or not, x= elt Solution By definition, x(4) will be periodic if, e/%(""") = ef a ITY ae gf gl In order to satisfy equation (2.2), e/%! =1 We know that, e/” = cds wT + jsinayT For @ =0, e/" =1(w,-0 defines only DC signal) For , #0, @,7 = 2xm (defines AC signal) then e"*7 =1 where m= 1, 2, 3,... integers) ‘Therefore, eee (7 = 270 ‘Therefore, periodicity is given by, T. = Q) Similarly, for discrete-time signal.x(n), the condition for periodicity is given by Q,N =2nm Nam where Vis an integer always, ° . Problem 2.5 Test whether the signal is periodic. If so, find the fundamental per = x xo-eo{e08) Solution The given signal resembles the general expression, (1) = cos( a" +9). ‘Therefore, on comparinggthe given problem with general expression the frequency @= 1. ‘The fundamental period for which the given signal exhibits periodicity is, ‘The given signal is periodic with periodicity 2n. Problem 2.6. Test whether the signal is petiodic. If so, find the fundamental periad. esin( sons}ISS-12 @ Digital Signal Processing Solution x)= so( tay The frequency, w,~ % Therefore, Te= -(5)- wy (3m ‘The fundamental period for which the given signal exhibits periodicity is 7, Problem 2.7 Find the fundamental period of x(t) ~oor{ 4 -osin( 2) xyzcos( Z)rasin( 2) = nO EXG) Solution The given signal x(t) isa composite signal which has two component signals x,(f) and x,(1). For x,(1). the frequency, a =F The fundamental period of x,(1)is 7, = a For x,(i), the frequency, a, = z ‘The fundamental period of ,()is 7; = 2% =—2™ 19 ‘The signal x(i) is said to be periodic fond only if the ratio of 7, to 7, isa rational, that is, R63 10” i (rational) ‘The fundamental period of given signal x(0) is, Ty = 57, =37, = Problem 2.8 Test whether the given signal is periodic or not. x(t) =cost+sinv21 x()=cos!+sinv2t a 30 20 Solution The given signal x(¢) is a composite signal which has two component signals x,(t) and x,(1). For (0), the frequency, 0, = 1Introduction to Signals and Systems @ 188-3 The fundamental period of (is 7) For.x,(0, the frequency, w, = v2 ‘The fundamental period ofx,@is 7; “The signal x() is said to be periodic if and only of the ratio of T, to T, isa rational, that is, 2x _ 2 1 yan 1 Kirationaly Since the ratio of T, to 7, exhibits an irrational ratio, therefore signal x(#) cannot be periodic. Note Any signal exhibits irrational ratio of their period is apertodic signal. Problem 2.9 Test for periodicity of x(t)= jel a Solution The frequency of the signal, w, = 10 ° ait mnt °"@, 10 5 ‘The signal is periodic with fundamental period m/ 5. If'so, find the fundamental period Problem 2.10 Test whether the given discrete-signal is period sree( 2 Solution ven si an For the given signal, the Frequency, Qy == where Qj is the frequency in radians in the discrete domain. 7 2nm _ 2nm ‘The fundamental period, Np === =3 (m= | peri 0 Oo, 2n/3 (m=) Note The fundamental period N of any diseréte signal must be an integer Problem 211 Test for the periodicity of x() = ow'(F } Solution ones E } +cos20 Hint cos? @=:188-14 @ Digital Signal Processing Since the frequency, =, N= 2 2 QA The given signal is periodic with fundamental period WV, = 8. 8 (m=1) Problem 2.12 Test for the periodicity of x(n) =!" Solution’ For the given signal, the frequency, , = 7 Note The general expression for fundamental period is Ny =—~-m In the previous problems, the value of mwas assumed to be 1. In this problem, in order to satisfy the integer condition for a fundamental period, m is assumed to be 7. That is, ‘Therefore, the fundamental period N, = 2 if m= 7 . led Problent 2.13 Test for the periodicity of x(n)=4e > Solution Frequency Note Phase term never contributes in the definition of periodicity, that is, irrespective of the phase term the given signal may or may not be periodic. 4 For the given signal, the frequency, Qo = =Introduction to Signals and Systems @ ISS-15 2mm _ 2mm _3m Therefore, OQ, aR 2 5 Period N, must be.an integer. Therefore m= 2. 5 x2=5 2 tim ‘The given signal is periodic with period 5 ifm =2 The fundamental period, a sen) Problem 2.14 Test for the periodicity of x(m)=4e* Solution For the given signal, the frequency, Q = ; Therefore, For m=7 (Substitute 7: Hence, the given signal is periodic. Problem 2.15. Determine the fundamental period of the signal Solution L™ Ul DC Component ‘The given signal x(1) is a composite signal, contains three signals: DC component, signais x,(0) and x,(0. The periodicity is calculated based only on signals x,(0) and x,(0). For the signal x,(n), the frequency ®, = * ‘The periodicity of x,(n)is N= 2m 4 aeISS-16 © Digital Signal Processing Form=2, Qe For the signal the *2(n),the frequency 2)= z ‘odicit is Ny oom ees (me The periodicity of (r)is Mz = Suma are =S (m=) The signal is said to be periodic if and only if the ratio of N, to N, is a rational, that is, Ny, 7 Ny 7 5 (rational) The fundamental period is N, = SN, =7N, = 35 Problem 2.16 A pair of sinusoidal signals with a common angular frequency is represented by x (0) =3 sin (3 an)andx,(m) = J3 sin@ zn) Specify the period for which the period N of both x(q) and x,(n) must satisfy them to be periodic. Solution ‘The common angular frequency of both the signal is given by aq = 3a rads Phase angle, 6=0 We know that the period NV for discrete-time signal is given by For x,(1) and.x,(1) to be periodic, their period NW must be an integer. This can only be satisfied for m= 3, 6,9, 12, 15... integer multiples of 3), which results in p iteger multiples of 2). Problem 2.17 Determine whether the signal x(n) = cos(0.L 7) is pe Solution x(n) = cos(@.1nn) ta, lic or not. For the signal a(n), the frequency 92, = 0.1m 2n_ Qn The periodicity of x(n) is NW = ‘The given signal is periodic with period 20. 2.2.3. Even and Odd Continuous-time Signal A continuous-time signal x(0) is said to be even, if it satisfies the condition 23() = (1), for all (5) Even signals are symmetric about the vertical axis, as illustrated in Fig. 2.12 (a). A.continuous-time signal (0) is said to be odd, if it satisfies the condition . 2x()= 1-9, forall r eoIntroduction to Signals and Systems @ (SS-17 (Odd signals are antisymmetric about the vertical axis, as illustrated in Fig, 2.12-(b). ll) mA @ o Fig. 2.12 (a) Even Signal (b} Odd Signal Let us consider a signal, x(0) which can be decomposed as odd and even signals, i.e. x(x) +20 7) where x,(0 represents the odd signal and »,(0) represents the even signal. From the definition of odd and even signal, XOX) (2a) x(0* 4-0) (2.80) Replace ¢=-# in equation (2.7), aD EC ta) (29) ‘Substituting equation (2:8) in (2.9), x)= +x (0 (2.10) ‘Solve equations (2.7) and (2.10). Adding (2.7) and (2.10), x(t) +x(—1) = 24, (0) Subtracting (2.7) from (2.10), x(Q—x(-1) = 2x, Solve for x,() and x,(), n= até) e a(t) @uy soettcsen 2.12) Equations (2.11) and (2.12) gives the relation between odd and even signals. Complex vatued continuous-time signal A complex valued signal x{0) is said to be conjugate symmetry if it satisfies the condition x=x() where x'()) is the complex conjugate of x(1)ISS18° @ Digital Signal Processing if a()=x,() +i5() (2.13) where x,(1) isreal part of x(1) and x(0) is imaginary part of x(0) ‘On conjugating equation (2.13), X0)=5, 020 It is shown that conjugation property affects only the imaginary part x(¢) of the signal x(), x,(0) ofthe signal. Itis also understood that a complex valued signal x(#) is conjugate symmett is an even signal and its imaginary part is an odd signal. Odd and even discrete-time signal A discrete-time signal x(12) is said to be even, if it satisfies the condition x(n) =x(-n), forall n Even signals are symmetric about the vertical axis.as illustrated in Fig, 2.13 (a), A discrete-time signal x(n) is said to be od x(n) = =x(-n), forall n it satisfies the condition (Odd signals are antisymmetric about the vertical axis as illustrated in Fig. 2.13 (b). x) x(n) tl I li ope I 6 -§ 4-3-2 -1 [01 45 6n TTT 1 aan) fa) (b) Fig. 2.15 (a) Even Signal (b) Odd Signal Let us consider a signal x(x), which can be decomposed as odd and even signals, i.e. x(n) = X,() + x_() (2.15) where x, (n)is odd signal and x,(n) is even signal From the definitions of odd and even signal, =x,(m) = x,(-n) 216) 2(n)=x,-n) Qin Replace n = ~n in equation (2.15), a(-n) = x, (-n) +x, (=m) : Q.18y (=n) =x, (1) +x,(n)Introduction to Signals and Systems @ 188-19 ‘Adding equations (2.15) and(2.18), x(n)+x(n) =2x,(n) Subtracting equation (2.15) from (2.18), x(n) x(-n) = 23, (7) x(n= Sapte 19) (2.20) Equations (2.19) and (2.20) gives the relation between odd and even conditioris. SOLVED PROBLEMS... == = Problem 2.18 Find the odd and even components of x(¢)=e!" Solution We know that, any signal comprises of even and odd parts, ie. xQ=x,()+x,(N=e™ The even signal is given by The odd signal is given by els Bae at) _ of - a ee x00 = J] |= sine 3()= 2,0) + ¥,(Q)= cos 2¢+ jsin 2¢ Problen-2.19 Show that the praduct of two even signals or two odd signals is an even signal and that the product of an even signal and an odd signal is an odd signal. Solution Let us consider two even signals x, (f) and x,(0). = 4,00 2,00 Replace =~, then ———_____ Hint For even signal x() = x(-#) ae =a) Hence, itis an even signal. Let us consider two odd signals x,(#) and x,(1). HO =3(9 2)1SS-20 @ Digital Signal Processing Replace ¢ =—4, then aC =a (-1) <0 x¢-9- EE at) = x) * 2) =) Hence, itis an even signal. Let us consider two signals such that x,(/) is even and x,(¢) is odd. Then xO =x (0 xx . Replace ¢ = —f, then a= 5 ea) x(-1) = x, (0) x[-x, (9) =-x(9 which is odd. Summary Even = Even= Even; Odd x Odd = Even; Even x Odd Problem 2.20 Find the even and odd components of x(2) = cos 1 + sin ¢. Solution : a(t) = cos r+ sins 3-4) = cos (-1) + sin(—1) = cos ¢—sin ¢ The even part is given by (0 cost Al) x(O+ x1) _ (Cost-+sins}+ (cos/—sins) 2 2 The odd part is given by x (92202 =A) eset in)— (608 SID) Sin x(0) = x,(0+,(0 =cose+ sine Problem 2.21 Find the even and odd components of x() = {3,2,1,4,5}. Note The arrow mark always shows the value for 0th position i.e. Position | -2 | -1 | xn) 3]2[1 Solution ‘The even partis given by a(n) = (n)+x(=n) 2Introduction to Signals and Systems @ ISS-24 For #=-2; x= a =3t5 7 For #=-1, (1) 7 73 For n=0, 1,(0)=2Otx0) 141, 2 2 = = M+ x(-2) _ 442 _ For m= 1, (I) 7 a For m2, 2-2)= 22220) _ 583g . x.(m) = 4,3,1,3,4} The odd partis given by For #=-2, For nel wea tax) 2-4 , (=O For n=0, 1-2) 9 For n=1, x) 2200 24 2 for 22, (dye 2H) 53 Xm ={-1 10,113 Adding x(n) and x,(n) position-wise, we obtain the original signal x(n). Problem 2.22 Draw the odd and even representations for the given signal. xt)IS$-22 @ Digital Signal Processing Solution For even signal, For odd signal, 2,() =O) (i) =O. x) gt) 2 _ Fig. 2.15 Problem 2.23 Draw the even and odd signals of the signal shown below. x(t)Introduction to Signals and Systems @ 1SS-23 Solution For even signal, For odd signal, (9220 3,2) 20) x0) xt) 4 4 x4 ¢ xt) J ) t t px exe) -x(-4) 4 . A 0-9) (t) + x1) ox X= HOHHED 2s Fig. 2.17 Problem 2.24 Draw the even and odd signals of the given signal. x(n)188-24 @ Digital Signal Processing Solution For even signal, For odd signal, a(n) = AOAC selr) OCD xo) . 4x) 3 2 2 witty, Holi w1es 45" Tee 45" xt—n) ‘x{-n) 3 3 2 2 1 I | u =g-4-9-2-1 | " T8n4= 88-1 " xn}ex(-n) x(n) 2 =5-4-3-2-1 en piled. TTT =s-4-9-e-1 Treo 45" ne Ls xo = Melon) xiny-xi-n) 15. 3 1 2 3 Sats toed =S=4-3-2-1 | 1 -s-s-o-2-1'| ] L.. T | Thess 2 3 {r= HOC) 15 1 cocaca-e-ol oT] Ly. T ] ] whe 22450 1 hs. Fig. 2.19Introduction to Signals and Systems @ 1SS-25 Complex valued discrete-time signal A complex valued signal x(1) is said to be conjugate symmetry if it satisfies the condition x(-n) =x" (nl) where, x" (n) is the complex conjugate of x(n) 7 (1) = xplnd+ Je) (n) 221) where, x,(n) is real part of x(n) and x; (n)is imaginary part of x(n) ‘On conjugating equation (2.21), ¥ = rel) ie) It is shown that the conjugation property affects only the' imaginary part x, (ri) of the signal x(n), not the real part ofthe signal x(n). It is also understood that a complex valued signal x(n), is conjugate symmetric if its real part is an even signal and imaginary part is an odd signal. 2.2.4 Energy Signal and Power Signal Let us consider a current i(2) flowing through a resistor R, developing a voltage (i) across it. The instantaneous power p(t) dissipated in the resistor is given by p= FOR 22) ro R In signal processing or analysis, itis customary to define power dissipation in a resistor of unit value, i.e. R= 10, regardless of whether the given signal x(4) represents a voltage v(1) or a current i(f). Hence, we can represent the instantaneous power pl) of the signal x(0) as A =VFN=PO=PO (223) ‘The total energy of the continuous-time signal x{¢) expended over time interval 7/2 < ¢ < +T/2is given by Li J pod ci ex uf pp a (0-1)=— 4 ‘The given signal is an energy signal as energy at «is finite, tf [x Pat Tin Poles thle *rucnfidr= Les tftIntroduction to Signals and Systems @ 1SS-27 ep Art Pou I( ara ue zrle), . Pau Hem -2) P,=0 The given signal is not a power signal as power atco is zero. Problem 2.26 Test whether the given signal is an energy signal or a power signal, ¥(}=(-0.5)" un) Solution £.=) |x(nl=¥ [0.56 E_=¥(0.5)" = 0.25)" m0 0 tA... i-0353 n energy signal as Eis finite. ri 2 P= Lt vo-2N +1 eet E,=: Po tt [osytaen|? 72 | Pel ae wii 029 =0 a plat The given signal is not a power signal as P, is zero. Problem 2.27 Test whether the given signal is an energy signal or a power signal. (= u(r) Solution ea ¥ Je(a)P? pole = Ut ¥ bacnye lu Ei mtd, win 24 wan 2s188-28 @ Digital Signal Processing 2 =a ¥ co] = — wen mad! lowe i$ ad ~ etna e Hence, the signal x(n) = (7) is a power signal as P, is finite. Problem 2.28 Test whether the given signal is an energy signal or a power signal. x(f) = u(t) Solution 12 12 B= te f [x@Par= ue f faa? a To" in To ie mm pyre B= te fed= le roe} rr 3 (5) Pelt rh RO 1 ve, mer, fe iF J Pade P= Lt = tf errae eid 3 on tt B Hence, x(0) is neither a power signal nor an energy signal as E. and P, are infinite. | vm | |Introduction to Signals and Systems @ (88-29 Problem 2.29 ‘Test whether the given signal x(t) =/'**’ is an energy signal or a power signal. Solution fp mexray|? ee Jee od Hint ra m1 ef ra ri rir e= ull =u (b+ r) Be) E,=0 1 7 pated f jee a reT fi 1 7 1 PeLts f lati tr YeT jy ToT Pet +(35) eT (22 Peis ™m ria Hence, it is a power signal as P, is finite. Problem 2.30 Test whether the given signal is an energy signal or a power signal. x(1) = cost Solution mp mm ‘ B= kt Jiota= we Jleosa?ae Hint cos?g = 100828 mn “mm eee a = u ft pare gt + fone ane ie 12 Pats f [xofdre Li f fcostal' at rool TR = Tian18$-30 @ Digital Signal Processing The given signal is a power signal as P_ is finite. Problem 2.31 Test whether the given signal is an energy signal or a power.signal. x(1)= u(r) —2(1-5) Solution Es lt F bea? se e=t Berns 4 E=Yl=5 1, then the signal amplifies).Introduction to Signals and Systems @ ISS-33 Let us consider a signal x(/) =a sin az, When a = 0.5, the signal reduces toy (1) = 0.5 sin af. Only magnitude of the signal reduced, whereas the period of the signal unchanged as shown in Fi When a = 1.5, the signal x(?) become y,(1)= 1.5 sin ar, Only magnitude of the signal increased, whereas the period of the signal remain unchanged as shown in Fig, 2.22(b). Fig. 2.22 Amplitude Scaling (a) a<1 (b) @>1 Consider a discrete-time signal x(m)as shown in Fig. 2.23. Let us introduce an amplitude scaling factor arto the discrete-time signal, ie. Yn) = a(n) (2.28) where, or = scaling factor (if a <1, then the signal attenuates; if a > 1, then the signal amplifies). Let us consider x(n) = u(n), 5220. When a = 0.5, y, = 0.5u(n). Only magnitude of the signal y,(¢) reduced, whereas the number of samples remain same as shown in Fig. 2.23(b). Similarly, when a = 1.5, ¥,(n) = 1.5u(m). Only magnitude of the signal y,(#) increased, whereas the number of samples remain same as shown in Fig. 2.23(c). x(n) yt) TOs yin) f : IL “{ttittit, {a) Original Signal (b) For a= 0.5 (0 For a= 1.5188-34 @ Digital Signal Processing BL» 4). 3 Xq(0) = 3:x(n) Mi Fig. 2.23 Amplitude Sealing (d) Original Signal (e) Scaled Signal (a=3) lullid re io 3 f a9 109 2.3.2 Addition of Signals Consider a pair of continuous-time signals x,(/) and x,(1) as shown in Fig. 2.24. Adding these two signals, x (0) and x (0), results in a signal y(2). It is important to note that the period of the output signal is unaltered. =a (+30) 2.29) x) xt) UO = x4(0 + x2(t + = . 4| 4 4 | 4 —al 2 2 1 Wd ' 2a pte! 2apre!' j~a pte (@)Introduction to Signals and Systems @ 1SS-35 xt) xa) yt) = xr(0) + x20) 5 4 + 4 = 3 F) 1 1 1 ot ete! a! ara et (b) Fig. 2.24 Addition of Signals Consider a pair of discrete-time signals 2,(n) and x,(n) as shown in Fig. 2.25. Adding these two signals, x(n) and x(n), results in a signal y(n). The period of y(n) is unchanged. Ha =x (ni) +x 7) 230) ain) xn) iscecazasr ) Fig. 2.25 Addition of Signals 2.3.3 Multiplication of Signals Consider a pair of continuous-time signals. x,(¢) and x,(1)as shown in Fig. 2.26. Multiplication of these two signals, x,(¢) and x,(¢), results ina signal (2), ic. WD=z (x2, (231) Multiplication of message signal x,(#) over the carrier signal r,(t) results in a modulated signal, which is used to transmit over communication media,18-36 @ Digital Signal Processing xy ith VA) = yi) x aol) fa) x2() YA = y(t) exe) O12 3 (o) Fig. 2.26 Multiplication of Signals Consider a pair of discrete-time signal x,(7) and x,(n) as shown in Fig. 2.27. Multiplication of these two discrete-time signals, x,(n) and x,(n), results in a output signal y(n), i.e. . Hn) =x) *x,(1) . (232) 2.3.4 Differentiation on Signals The derivative of an input signal x(#) with respect to time is defined by WO= fo (2.33) Example ‘The. voltage v(2) developed across the it uctor is a derivative of the current i(¢) flowing through the inductor L, ic. dio) Hel - vy = LT (2.34) Differentiation of continuous time signal is equivalent to the difference of the discrete-time signal x(n). x(n)—x(n—1) (2.35) ‘The differentiation of square wave and sinusoidal wave are illustrated in Fig. 2.28 (a) and Fig, 2.28 (b) respectively.xl) Introduction to Signals and Systems @ ISS-37 iu 2 1 -8-S-4-8-2-10] 12945 6 M — — — 9 ola x =e Lo. = 1 ——a i ; : i t ° ’ -oae -6-5-4-3-2-1 p 123 4 6 6 Yume) = x4(n) x xg(0) Fig, 2.27, Multiplication of Signals:188-38 @ Digital Signal Processing x(t) x(t) =sinot 1 7 | AAR “4 J 4 o tt) w= at y= 2 scosut at 1 1 + i “1 “1 C) - Fig. 2.28 Differentiation (a) Square Wave (b) Sinusoidal Wave (c) Differentiated Square Wave (d) Differentiated Sinusotdal Wave 2.3.5 Integration on Signals The integration of an input signal x(¢) with respect to time (0) is defined by ' WO = fxr Example “The voltage v() developed across the capacitor is a derivative of the current i(9 flowing through the capacitor C. ie. it © 10k ne (2.36) The integration of a continuous-time signal x(/) is equivalent to summation of discrete-time signal x(n). Hed = FT xtn) (37) The integration of square wave and cosinusoidal wove are illustrated in Fig. 2.29 (a) and Fig. 2.29 (b)respectively.Introduction to Signals and Systems @ ISS-39 + xt) 4 x{t) 1, then the signal compresses). x2 et 1 t ot be = 305 [Pos 1 ett () Fig. 2.30 Time Scaling (a) B<1 (b) B>1 Consider a discrete-time signal x(n) as shown in Fig. 2.30. Let us introduce a time scaling factor fi to the discrete-time signal, i.e. yO) =x(Bn) (2.39) where f= sealing factor (if <1, then the signal expands; if >1, then the signal compresses.)188-40 @ Digital Signal Processing fa xine} HILL. Fig. 2.3% Time Sealing (a) < 1 (6) > 1Introduction to Signals and Systems @ [SS-44 2.3.7 Reflection of Signals Consider a continuous-time signal (0) as shown in Fig. 2.32(a). Let (J) denote a signal obtained by replacing 1 by -f (time-reflection) to the continuous-time signal, ic. w= (2.40) Ifx(-1) = x(/) forall values off, then the reflected signal is an even signal, or ifx{-1)=—x(9 for all values of 1, then the reflected signal is an odd signal. Consider a discrete-time signal x(n) as shown in Fig. 2.32(b). Let y(n) denote a sigrial obtained by replacing mby-n (time-reflection) to the discrete-time signal, ic, Wa) =x(-n) (41) If x(-n) = x(n) for all values of n, then the reflected signal is an even signal, or if x(n) =—x(n) for all values of n, then the reflected signal is an odd signal. x) x1) | Reflected signal | No * a a (a) x(n) xn) Fetected signal a . * 0 oy Fig. 2.32 Reflection of Signals (a) Continuous-time Signat (b) Diserete-time Signal 2.3.8 Time Shifting of Signals Consider a continuous-time signal 2(2). Let y{/) denote a signal obtained by shifting the signal x(¢) by (¢—1,), that is, HO=At-4) 4) If the signal x) is positive, and 4,>0 for all values of, then the signal is said to be right-shifted signal. In the example shown in Fig. 2.33(b), the signal is shifted to right side by 3 units.188-42 @ Digital Signal Processing If the signal x(1), is positive, and #,<0 for all values of f, then the signal is said to be left-shifted signal. In the example shown in Fig. 2.33(c), the signal is shifted to left side by 4 units. xt) 3 2 SSA sa preas4s 6 * (a) ya (Wax(tS}=x(-(+3)) 2 Fight shift by 3 units, 1 SSSA pres as 6 “ ‘Sunits | “Shifted right ©) Yaltlex(tr4pan(t{-4)) 2 Chto shitty 4 unite |) 4 1 SESS SOT eT aes *t Shifted left 4 units ) Fig. 2.33 (a) Original Positive Signal x(t) (b) Right-shifted Signal x(t-3) (c) Left-shifted Signal x(t+4) Ifthe signal x(~£) is negative, and ,>0 for all values of /,then the signal is said to be left-shifted signal. In the example shown in Fig. 2.34(b), the signal is shifted to left-side by 3 units. Ifthe signal x(—) is negative, and /,<0 for all values of f,, then the signal is said to be right-shifted signal, In the example shown in Fig. 2.34(c), the signal is shifted to right. fo 2 - Saar i (a) 23 at e7 *Introduction to Signals and Systems @ 188-43 ‘yi(dext-9) 2 4s + 4 . +t Sas pi 2s 4667 si Shitedien ~ "3 unit ) 4 Yaltn(-t+4) 2] 1 : “sas 43-breetee7 units Shifted right {) Fig: 2.34 (a) Original Negative Signal x(-1) (b) Left-shifted Signal x(-t-3) (c) Right-shifted Signal x(-t+4) The above analysis also holds good for discrete-time signals. Let us consider a discrete-time signal x(n). Let (n) denote a signal obtained by shifting the signal x(1) by (—m,), that is ym) =x(n- Ny) (2.44) ‘Ifthe signal x(a) is positive, and 1, > 0 for all values of n,, then the signal is said to be right-shifted signal. “ In the example, shown in Fig. 2.35(b), the signal is shifted to right side by 3 units. ‘Ifthe signal +() is positive, and 1,< 0 for all values of »,, then the signal is said to be left-shifted signal. In the example, shown in Fig, 2.35(c), the signal is shifted to left side by 4 units, x(n) ¥A(M=H(N-S)=X(M-{49)) Right shift by 3 units 1 2 3 4 8 6 7 8 9 1011 12 ‘Suniks Shifted ight o188-44 @ Digital Signal Processing yakn}ax(ne4)ymx(n(-4)) to Loft shift by 4 units 4-3 241 4234 56 7 8 8 10 11 12 ‘Shifted left ‘units () Fig. 2.35 (a) Original Positive Signal x(n) (b) Right-shifted Signal x(n-3) (c} Left-shifted Signal x(n+4) Ifthe signal x(-n) is negative, and 7, > 0 forall values of 7, then the signal is said to’be right-shifted signal. Inthe example shown in Fig, 2.36(b), the signal is shifted to left side by 3 units. Ifthe signal x(-n) is negative, and m, <0 for all values of m, then the signal is said to be left-shifted signal. In the example shown in Fig. 2.36(c), the signal is shifted to right side by 4 units. xin) | WU yaln}=xi-n-3) 4-3-2441 12345678 9 10 " Shitediet 4 “Tune ~ Yaln)=x(-n+4) SSB TTT 8 tT ae " unis ‘Shified right () Pig. 2.36 (a) Original Negative Signal x(-n) (b) Left-shifted Signal x(-n-3) (0) Right-shifted Signal x(-n+4) 2.3.9 Time Shifting and Time Scaling The time shifting operation is performed first on.x(0) resulting in an intermediate signal v(0), vi) =a(t—b) 46) ‘Next, the time scaling operation is performed on v(e). This replaces r by af resulting in the desired output, i= vat) =x(ar—5) *Q47)Introduction to Signals and Systems @ {SS-45 SOLVED PROBLEMS___ Problem 2.33 (i) Find x(2«+ 3) for a given signal x(1). Solution Step (i) x(¢+3)=x[1-C3)] xt) Step (ii) Time shiftx(r+ 3) x(t +3) ar) Step-(ii) Timescale x(21+3) Gi) Find y(n) = Step (x03) x(t+9) 1 +t rr)185-48 @ Digital Signal Processifig ‘Stepdi) a(S 3) Problem 2.34 Solution Step) Step (ii) Step (iii) x5 +3) 1 t 76843240 Lower Bound_| Upper Bound i+3=-1 ma-4 Ey Find x(2#—3) for the given signal x(). L_, at Bound Bound 11 Bound HII Lower | Upper | Lower | Upper | Lower | Upper Bound | Bound | Bound | Bound | Bound | Bound m-3=0 | 2-3-0] 2-3=8 [8-3-2 [2-3-2 Pr-3=3 483 tema} ea] tees | owes | 26 eae tnd e=n2] reas] reas] ees| Introduction to Signals and Systems @ ISS-47 Problens 2.35 Find {3) for the signal given in Problem 2.33 Solution Step(i) + l-2) 2 . 1 “oO 723456 «G4 2 Step (il) 1 . ann TShtse te ft Bound Bound It Bound IIT Lower [ Upper | Lower | Upper | Lower | Upper Bound | Bound | Bound | Bound | Bound | Bound 2=2=0fv2=2= 1] wz=2=1 a= 2=2] «2-2-3 w=2) w=3) W2=3 wm=4) 12=5 rea] te6 18) 1=10 Problem 2.36 Find x(31+2) and +(5+2] for the given signal x(t). Step@ Step qi) a 2 1 2-101 2 vt Step Gi) ‘ 4xGats 2) 3] 21S$-48 @ Digital Signal Processing Bound 11 Bound TIL Lower | Upper | Lower | Upper Bound | Bound | Bound | Bound Bede 1 [342=2/3+2—2] He I= 3 =H! 3=0 w=O ued Bound I Lower | Upper Bound | Bound B+2=0 [3421 Bie-2] reed 2-2 ted Step (i) es 48270723 ~ Bound! Bound 11 Bound | Lower] Upper | Lower ] Upper | Lower | Upper Bound Bound | Bound | Bound | Bound | Bound B+2= Of3+2= 1[/3+2=1] +2=2| w3+2=2/03+2=3 w=-2| 03 WB=-1) B=0} =O] w3=1 fom 1=0) ro] r= @ 2.4 TYPES OF SIGNALS . In this section, we shall study the various basic signals necessary to construct different signals, which are necessary for the system analysis. The basic signals are as follows: 1. Exponential signal () Real exponentialsignal Gi) Complex exponential signal 2 Sinusoidal signal 3. Step signal 4. Impulse signal 5. Ramp signal 2.4.1 Exponential Signal Real exponential signal (continuous-time signal) A real exponential signal x() in its most general form is represented by x(Q= Be" (2.48)Introduction to Signals and Systems @ ISS49 where B is scaling factor, (real parameter) and a. is real parameter, Depending on the value of a, a real exponential signal can be further divided into two more signals. For o:<0, the magnitude of a real exponential signal decays (decreases) exponentially and is illustrated in Fig 237 (a). For oc> 0, the magnitude of a real exponential signal rises (increases) exponentially and is illustrated in Fig. 237(b). x0) x(t) 0 t at @) (b) . Fig. 2.37 Continuous-time Exponential Signal (a) a < 0, (b) a >0 Example Charging ofa capacitorisan example of a growing exponential signal, and discharging of a capacitor is an example of a decaying exponential signal. Real exponential signal (discrete-time signal) A real exponential signal x(n) in its most general form is represented by x(n) = Ba" (2.49) where B is scaling factor, (real parameter) and a1 is real parameter. Depending on the value of a, a real exponential signa! can be further divided into four signals. For @ > 1, the magnitude of a real exponential signal rises (increases) exponentially and is illustrated in Fig. 2.38(a). For 1 > a > 0, the magnitude of a real exponential signal decays (decreases) exponentially and is illustrated in Fig. 2.38(b). For 0 > @ >I, the magnitude Of a real exponential signal decays (decreases) exponentially. For each integer value of n, the signal is represented alternatively and is illustrated in Fig. 2.38(c). For @ <-I, the magnitude ofa real exponential signal grows (increases) exponentially. For each integer value of 7, the signal is represented alternatively and is illustrated in Fig. 2.38(d). xin) x(n) [f>a>d] @) (b)185-50 @ Digital Signal Processing we O>a>-l] xin) Lt tee ss Lil a plore Arey 1O} @ Fig. 2.38. Discrete-time Exponential Signal (a}a>1, 1 1>a>0, @0>a>-1, Ma<-l Complex exponential signal (continuous-time) Let us define an exponential signal as xQ=e™ (2.50) An important property of this complex exponential signal is its periodicity, ie. cl agit? el gly ion" We know that, . ef" = cos(anyT)+ jsin(@yT) =I then x(1)= 1, which is periodic for any value of 7. Ifa = Uf @y #0, then the fundamental frequency a of the signal is the smallest positive frequency for which the equation wy = 21/7, holds good. Comtex exponential signal (discrete-time) Let us define an exponential signal as x(n) =Ca" (2.51) where C= scaling factor, complex parameter a=e", complex parameter 2.4.2 Sinusoidal Signal (Continuous-time) The continuous-time version of a sinusoidal signal x(¢) in its general form may be written as X(t) = Acos(ex +4) (2.52) Where A = Amplitude of the signal x(¢) to = Frequency of the signal x) (radian/s) 4 = Phase angle of the signal x(0) (radians) and is itlustrated in Fig. 2.39. .Introduction to Signals and Systems @ ISS-51 XtP-Acos (atse) Tek Fig. 2.39 Continuous-time Sinusoidal Signal Assinusoidal signal is an example of a periodic signal, whose period T™ 2 : In order to prove the periodicity of the sinusoidal signal, Iet us consider a continuous-time signal . x(t) = Acos(at +9) (2.53) which repeats every T second, i.e. x(U+7) = Acos(0x+7)+6) x(47) = Acos(tat+0T +0) For ree (one cycle) Then, the equation reduces to x47) = Acos(ar+9)= x(0) (2.54) Similarly, let us consider a discrete-time signal x(a) = cos(Qn+t 9) (2.55) which repeats every T second, ie. : x(a+ N)= Acos(Q(n+ N)+9) (1+ M)= Acos(Qn+ON +6) where (= Angular frequency for a discrete-time signal For N= 2m {one cycle) x(t N) = Avos(Qn+2nm+9) Then, the equation reduces to . (n+ N) = Acos(Qn+ 9) =x() (2.56)18852 @ Digital Signal Processing Complex exponentiatrepresentation of sinusoidal signal The complex exponential signal canbe expressed in terms of sinusoids as eT = cos(ax)+ jsin(ar) 257) 7!" =cos(ar)— jsin(ar) 258) Acos(cr+9)=—(c"™" +67") Acoster+9) = 5 (ee” +e" te") Acos(ar +9) = Re{4c™**)} (2.59) Asin(at +9) = Im (Ae) 2.60) Similarly, for discrete-time signals, the complex exponential signal can be expressed in terms of sinusoids as e/™ = cos(Qn)+ jsin(Qn) (2.61) 7! = cos(Qn)-. (Qn) (2.62) Similarly, the discrete-time sinusoidal signal can be written in terms of periodic complex exponentials, again with the same fundamental period, Q. ‘Acos(OQn+ =F {ero terns) Acos(Qn+9) =Hemer tomer) Acos(On+9)=Re(AeXYy (2.63) Asin(Qn+9)=Im{Ae™") (2.64) Exponentially damped sinusoidal signal When teal value decayed exponential signal is multiplied with a sinusoidal signal, it results in an exponentially damped sinusoidal signal. This can be represented by | x()= Ae" sinfor +4) a>0 2.65) and is illustrated in Fig. 2.40 (a). Similarly, for discrete-time signal, an exponentially damped sinusoidal signal can be represented by in(Qn+9) a>0 (2.66) x(n) = Ae and s illustrated in Fig. 2.40 (b).Introduction to Signals and Systems @ 1SS-53 x(n) Fig. 2.40 Exponentially Damped Sinusoidal Signal (@) Continuous-time Signal, (b) Discrete-time Signal 2.4.3 Step Function ‘The continuous-time step function is commonly denoted by 1(/) and is defined as wo=[p #20 267 0, r<0 and is illustrated in Fig. 2.41 (a). The discrete:time step function is commonly denoted by u(n) and is defined as 1, 20 0, n<0 268) wel and is illustrated in Fig. 2.41(b). u(t) @ Fig. 2.41 Step Signal (@) Continuous-time Signal (b) Discrete-time Signal ‘The step function shows discontinuity at = 0 (n=0 for discrete-time) in case of continuous-time step representation. 2.4.4 Impulse Function The impulse function is a derivative of the step function (/) with respect to time, Conversely, the step function u(t) is the integral of the impulse with respect to time. The continuous-time unit impulse Function is commonly denoted by &(¢)and is defined as188-54 @ Digital Signal Processing 5)=9, 120 26) and Jande=i am Equation (2.69) says that the impulse 3(/) is zero everywhere except atthe origin, and equation (2.70) says that the area under the unit impulse is unity. The impulse function is also known as the Dirac delta function and is illustrated in Fig, 2.42(a). ‘The discrete-time impulse function is commonly denoted by 5()and is defined as. 1 n=0 . 8(n)= ft no en) From equation (2.71), the impulse 5(n) is zero everywhere except at the origin, whose magnitude is unity. The impulse function is illustrated in Fig. 2.42(b). a) nr) 1 4 5 t a n=0 o () 3(n-Ng) G(n+ngy ite | 1 0 aS 9 a toy @ Fig. 2.42 Impulse Signal (a) Continuous-time Impulse (b) Discrete-tme Impulse (©) Discrete-sime Input Shift by +n, (@) Discrete-time Impulse Shift by - ng 2.4.5 Ramp Function The integral of the step function u(¢) is a ramp function of unit slope. The continuous-time ramp function is commonly denoted by r(1) and is defined as 1, 120 ro={ 120 (2.72) and is illustrated in Fig. 2.43(2).Inroduction to Signals and Systems @ 188-55 ‘The discrete-time ramp function is commonly denoted by r(m) and is defined as In, n20 0, n» ¥(0) Input signar >| S¥stem Qutput signal Fig. 2.44 Basic System In a communication transmitter, the system accepts message signal (audio, video or data), processes it (modulation and filtering), and gives an acceptable output for communication. A. typical prototype of ‘communication transmitter system is shown in Fig. 2.45. Wanton tio yee Data: Fig. 2.45 Communication Transmitter System An electrocardiograph is. system which collects electrical potential from the surface of the heart, processes and filters it, and gives an electrical output for diagnosis. A typical prototype is it shown in Fig. 2.46. £08 Greoaroren LOM Fig.2.46 Electrocardiograph Specifically, a system is an interacting group of physical objects or conditions called system components. The system accepts one or more input signals or parameters, and produces one or more output signals or188.58 © Digital Signal Processing @ 2.6 PROPERTIES OF SYSTEMS ‘The properties ofa system help us to understand the characteristic of the operator (say M) operating/representing the system. Following are a few basic properties of systems. Based on the properties, the system can be classified as Continuous-time and Discrete-time system Stable and Unstable system Memory and Memoryless system Invertible and Noninvertibie system ‘Time-variant and Time-invariant system Linear and Nonlinear system Causal and Noncausal system MA wR EN 2 Continuous-time system If the input and output of the system are continous-time signals, then the system is called ‘Continuous-time system’. ‘Let us consider an input signal x(/) to the system, If the system produces an output signal y(¢), then the system is called *Continuous-time system”, Input signal Ht 7 Output signal Fig. 2.47 Continuous-time System 1 Continuous-Time and Discrete-Time System Discrete-time system If the input and output of the system are discrete-time signals, then the system is called *Discrete-time system’. Let us consider an input signal x(1) to the system! If the system produces an output signal y(n), then the system is called ‘Discrete-time system’, x}———+}_ H+] 0) Input signal ‘Output signal Fig. 2.48 Discrete-time System 2.6.2 Stable and Unstable System Stable system A given systemvis said to be stable if and only if every bounded input produces a bounded output. The stable system is also known as ‘Bounded Input-Bounded Output” (BIBO). Lef us consider a system operated by an operator H[:]. The system is said to be stable if the bounded input x() produces bounded output 31), i.e. if |x| s Mf, <= forallr (2.74a) then [| sm, <© forall (2.74b)Introduction to Signals and Systems @ SS-57 A similar analysis holds good for a discrete-time system also, ie. if |x(@)|<.M, _-»_>__ Problent 2.37 Determine whether the given system h(n) =a" a(n) is stable or not. Solution For stability, the response of the system h(n) must be absolutely summable, i.e. Z Ae) || 2” [cw s DE Hence, the system is stable. P2425 4.42” S00Introduction to Signals and Systems @ ISS-59 Ayn) = e"u(n—4) The condition for stability $*| e¥2| : dt where C = Capacitance of the capacitor Note The analogous to differential equation in a continuous-time system is a difference equation in the discrete-time system. BO 2 4n-1)-y0) ‘Therefore, the differential equation is a snenory system. Memoryless system The given system is said to be memoryless if the output of the system depends solely ‘on the present value, Examples 0) = x) We) = 220) y(n) = xn) y(n) = mx(n) The memoryless system is also known as a ‘Static System’, A resistor is a memoryless system since the current flow i(2) flowing through it is proportional to the applied potential, i.e. i= o where R = Resistance of the resistor188-64 @ Digital Signal Processing 2.6.4 Invertible and Noninvertible System Invertible system A system is said to be an invertible system if the input signal given to the system can be recovered. The concept of invertibility is illustrated in Fig. 2.51. yt) tig gg a Fig. 2.51 Invertible System Let us consider an example of a communication system shown in Fig. 2.52,where the signal is modulated with a carrier signal and transmitted.through the transmitter system, The same signal is recovered at the receiver system by demodulating the signal from the carrier signal. Transmitter yi Receiver ty system system | __ ¢y = x49) Fig. 2.52 Communication System All the transforms used in signal processing are invertible system. \ Noninvertible system A.system is said to be noninvertible ifthe input signal given to the system cannot be recovered from the output signal of the system. The square-law system is generally a noninvertible system, except for the distinct inputs. 2.6.5 Time-invariant and Time-variant System Time-invariant system A system is said to be time-invariant if the input signal is delayed or advanced by any factor that leads to some delay or advancement in the time scale by the same factor, i.e. the system responds to an input which is given at any instant of time and results in an output. wegen . Roy fa) ho} xt) yey [Ee A : by xit=1 Time-invariant sma ep ro Fig. 2.53 Basic Time-invariant Systems (a) Delayed Time-invariant System (b) Time-invariant System (0) Advanced Time-invarians SystemIntroduction to Signals and Systems @ (SS-65 Let us consider three time-invariant situation depicted in Fig. 2.53. Fig. 2.53(a) illustrates a delayed pulse output (1 sec) produced by a system for a delayed pulse input of zsec. Fig. 2.53(b) illustrates a pulse output produced by a system for a pulse input where in delay is not incurred neither in input signal. Fig. 2.53 (c) illustrates an advanced discrete step output produced by system for an advanced discrete step input. Any other combination (input and output combination) of the system is time-variant system. SOLVED PROBLEM __ _. _ Problem 2.41 The input-output relation is given by (r) =sin| ew) . Determine whether the system is time- invariant or not. Solution (¢)=sin[x(0)] Let us assume the signal of the form nO=sin[y(o)] a) Let us introduce time delay ¢, in the input signal in equation (1), then (= H(-1,) The delay input therefore results in the output yp) =sin [x,(1)]=sinfx(e-19)] @) Let us introduce the same time delay ¢, in the output of the equation, i.e, Wilt (t-t0)] 8) On comparing equations (2) and (3), rlt)=lt=Fo) Hence, system is time-invariant. Time-variant system A system is said to be time-variant if the output signal is delayed or advanced with x(t) yet)188-66 © Digital Signat Processing SOLVED PROBLEMS == Problem 2.42 The input-output relation is given by y(t)=tx(¢). Determine whether the system is time- variant or not. Solution w= 1x(2) Let us assume the signal of the form AO =I) ay Let us introduce time delay to in the input signal in equation (1) n= xb) The delay in input results in an output IO=IO=FyC~) @ Let us introduce the same delay /, in the output of the system . HEH) (lt) 8 On comparing equation (2) and (3), IAD Wlt-b) Hence, the system is time-variant. Problem 2.43 Determine whether the following systems are time-variant or not. @ y@)=s(s)sinae Gi) (= x41) Gi) ya)=et™ dv) y= Px) Solution @ v)=x()sin ew ay WO) = T{x()]=sin ox x) Introduce time delay 1, in the input, ie. aWQ=xt-) iO =sin oF x,(1) =sin ax x(t—t9) @) Introduce time delay f, in the output of the equation, ° Wet=ty) =sin ax (¢—ty) x(t—ty) @) ‘On comparing equations (2) and (3), lt) + yt -to) The system is time-variant.Introduction to Signals and Systems @ ISS-67 Gi) (= a(49, 4) Introduce time delay ¢, in the input, ic. 14) =x 40-49] WO= x40) 65) Introduce time delay r, in the output of the equation, ic. Http) = x[4(t 4) 6) ‘On comparing equations (8) and (6), WO=VUt) The system is time-invariant Gi) pee q) y= Tleo)=e Introduce time delay f, in the input, ie. . =x) yee = eto) «) Introduce time delay f, in the output of the equation, ie. ylt=ty) =e) o On comparing equations (8) and (9), WO=¥E-h) The system is time-invariant. @) O=PxO (10) Introduce time delay ¢, in the input, i.e. x=) then WO=Px (0) = Patt) ay Introduce time delay f, in the output, ie. Ht) == x=) (12) (On comparing equations (11) and (12), IO # HE fq) The system is time-variant.18848 @ Digital Signal Processing Problem 2.44 Detetmine whether the following systems described by the given differential equations are time-invariant or not. 2 6 POs 28O. se =x09 iy. £20 42,94 = x(t) 3 iy PLO 29, 72, eos xr Note The differential equation indicates thatthe time-invariant system must have constant coefficiens. If the coefficients are time-dependent parameter, then the system is called a time-variant system. Solution 9 FO dO 425 y(t) = 0) Gr tS +) = x0) Since the coefficients are constant multiples, the system described by this differential equation is a time-invariant system. a 20 17199, yam since the coefficient of the second term is a time-dependent parameter, i.e. 21, the system described by the given differential equation is a time-variant system. 2 pfx 0 +e tx, +1824 900 x4 20. Since the coefficients of the first three terms of the differential equation is time-dependent, the system described by the given differential equation is a time-variant system. Problem 2.45 Determine whether the following discrete-time systems are time-invariant or not. (@ Ha)=cos {xt} Ga) = Ino) Gi) y= xl) aln—1)—Giv) 3h) = 0m) x04 1) «i se Solution @ ya)=cos [x(n] a) a)=TEx(7)] =cos [x] Introduce time delay 1 in the input, ic, Al) = x(n— 16) then . xr) = cos[x, (1)] = cos[.x(1-7g)] @) Introduce time delay 7 in the output of the equation (1), ie yng) = cos[x(e1~ 15 )] QB)Introduction to Signals and Systems On comparing equations (2) and (3), Jy) = (n= MH) The system is time-invariant. @)_ y@)= ln b(n) a) =F [x(a)] = In [x(r)] Introduce time delay ry in the input, i.e. oo (2) =3x(0—m) then ye) =In[ x4()]=In[x( 1) Introduce time delay 1, in the output of equation (4), ie. (a=) = In[xn—ry)] On comparing equations (5) and (6), Jy (2) = y(n= My) The system is time-invariant. Gil) y(a)= x(a) —xC0=1) Introduce time delay n,-in the input, i.e. xy(n)—3,0 PD x(n) x€n- 1-19) then yl) = xn) — 4 (n—1) = xl) — x11 =) Introduce time delay ry in the output of equation (7), i.e. (rng) = x(n my) — (1-119) On comparing equations (8) and (9), Jun) = y(n) The system is time-invariant. (iv) y(n)= x(n) x00 1) Introduce time delay in the input, ic. a(n) xi (+1) = x(n) x(n+1~my) then Ji) = 08) (14 1) = x(n 76) XCF L= m4) Introduce time delay 1% in the output of equation (10), i.e. Yn My) = x(1— 719) x41 — Mm) On comparing equations (11) and (12), : WO=He-m) The system is time-invariant, Iss-69 (4) (3) @ mM (8) Oo) (10) ay (12)188-70 @ Digital Signal Processing Problem 2.46 Determine whether the following systems are tim 0 oO =t Jamar [where C is capacitance of the capacitor] ariant or not. Gi 4O=F [where R(() is resistance of the thermistor] Solution f © yo=t fauna a wo=Theal-d fanart Introduce time delay fo in the input, i.e. x(t) =x(E=f9) if 1 AO=G J A=E J xen de Introduce time delay (o in the output of equation (1), ie. iy Oncomparing equations (2) and (3), x(t) WO= Ht 40) The system is time-invariant. 20. WO= FG) WO= TE] = Introduce time delay fg in the input, i.e. 3() = xf) then HwO= Ra) Introduce time delay /, in the output of equation (4), ie. Rl) HO _ Ml) RO J x@r-)ae a) Q) 8) @ o) ©Introduction to Signals and Systems @ ISS71 On comparing equations (5) and (6), wl # ptt) “The system is time-variant, 2.6.6 Linear and Nonlinear System Let us consider an input x,(¢) given to a continuous-time system which responds with y,(t). Similarly, let us consider another signal x, (t) given tojthe'same continuous-time system which results iif a response y, (1). “Then the system is said to be linear if 1. The response to x,(t)+x,(0) is y,(Q+y,(0 (additive property) 2 The response to ax,(#)+5x,(t) is ay,(t)+by, (1), where a and b are complex constants (scaling Property) Mathematically, T [ay (0) + bx, (Q)] = 07 [x,(9] +47 [2,00] T [ax (t) + or, (0) = 4 (+by(0) (2.76) If an input consists of weighted sum of several signals, then the output is the weighted sum of the responses of the system to each of those signals. , ‘Similarly, if two signals x(n) and x, (nm) given to a discrete-time system results in an output y,(n) and ‘yz(n) respectively, then the system is said to be linear if T [ax (n)-+ bx; (n)]=aT [,()]+ 67 [x27] T [ezr(n) + bx9(m)] = a(n) + y2(n) (2.77) ‘The superposition property holds for a linear system in both continuous-time and discrete-time, SOLVED PROBLEMS. = = Problem 2.47 Determine whether the given continuous-time system is linear or not. WD =tx) Solution Let us define the input signal x,(1) whose response y(t) is given by H(O=tat0) Similarly, let us define another signal x(t) whose response y(t) is . yAO=t (0 ‘The two defined signals are related by aC) = an (+ be,(0)188-72 @ Digital Signal Processing Then the output y,(1) is defined as Ys) = 13309) s(t) = thax,()+b49(0)] Pl) = at ay (4610 (0, yal) = ay4(0) +8y2(0) Therefore, the system is linear. Problem 2.48 Determine whether the given continuous-time system i: YOR +E where & is a constant. Solution Let us define the input signal. x,(t) which results in an output y4(¢). wi) =tx (+k Similarly, let us define another signal x, (¢) which results in an output y, (4). i) = y+ k The above signals x,(0) and x, (0) are related as 23(0) = ae (0) + 8x2 (1) where a, 6 = constants ‘Then, the output y3(¢) is defined as Yet) =x (Q +k s(t) =Tlax,(0) + be, ()]+k ld) = ats (0+ btx,()+k Is 4 ay O+by(0 Therefore, the system is nonlinear. Problem 2.49 Determine whether the given discrete-time system is linear or not. Yn) = nPx(n—1) Solution Let us define the input signal x,(n—1) which results in an output y,(n). Jar) = Py (2-1) Similarly, let us define another sigral x, (7-1) which results in an output y(n). Jala) =P xy(a-D)Introduction to Signals and Systems @ ISS-73 ‘The above signals x(n—1)and x,(n=1) are related as 239) = a (m) +bx, (7) where a, b = constants Then the output y,(») is defined by wlny= mx (a1) Jy) = 1? [ax (2-1) +bx, (9 1)] YC) = an? x, (8-1) +n? x, (1-1) ¥,() = ay,(n) + by,(n) ‘Therefore, the system is linear. Problem 2.50 Determine whether the given system is linear or not. y(n)= 3" (n) Solution Let us define the signal x, (m) whose response is given by vile) = 35 (0) Similarly, let us define another signal x, (n) whose response is given by vla=xi(n) ‘These two signals are related as ,(n) = ax,(n)+ bx, (1) where a, 6 = constants ‘Then the output y, (1) becomes ys =39 (0) C= [an (ry +be (oy Vs = Px} (rm) + b?xZ(m) + a(n) x9(n) y(n) # ayn) + bys (n) ‘Therefore, the system is nonlinear. Problem 2.51 Determine whether the following continuous-time systems are linear or not. @ 0 = xGing) @ 9 = sino) Gil) ¥() = Px(-1) (iv) PO=eO Solution @ xO=x(sing) Let us define x,(¢) and x,(0) as * 21 = x [sin(o) 2) > 2W = ¥, fsin(y]188-74 © Digital Signal Processing Let us define x, (0) such that then the output becomes, . Therefore, the system is linear. G0 =sin@(O] Let us define x (0) and x,(0) as Let us define x; (¢) such that then the output becomes, xy (1) = ax, (0) + bx, (1) ¥3(0 = x5[sin()] so =a [sinc] bm [sn] PC) = ay) + (1) (0 = ¥, 0 = sin[x(0] 4 () = y(t) = sin[x, (0) (N= aR) +67 ya(O.nsin [x,(0)] =sin [ax()+5x,() Hint sin (A+ A) =sin Acos B+ cos Asin B Isle) =sin [ax (t)}e0s[bx3(0))+ cos[ax; (¢)]sin [6x60)] dalt) # ay (1) + by2(t) Therefore, the system is nonlinear. @ yQ=Fx0-1) Let us define x,(Qand x2(¢) as Let us define 2x3(¢)such that then the output becomes, Therefore, the system is linear. x= = Px -D 20 = yD= Px 1) 33(0 = ax (0+ bx, (07 1 y= Px (-1) I3(0 =P [ax (-D 48x, ¢-D] Ys(0 = atx (1-1) +6222, (t- Ja = ay,() + by, (0)Introduction to Signals and Systems @ S575 (iy) yee Let us define x(t) and x2(¢) as 10> WOxe . 22) y2lf)=e Let us define x(t) such that 1 0) 4,00 = ay, (1) + br, (9) then the output becomes, nn wi=e IO Fay, (1) + by (0 Therefore, the system is nonlinear. . Problem 2.52 Determine whether the following discrete-time systems are linear or not. @ ya)=In Ex(a)) Gi) a) =x(4) -x(e-1) Gi) y(a)=2* (+71) iv) Hin) =2x(mp +4 Solution @ )=InB)] Let us define x(n) and x(n) as 2(n) => y(n) =In [xy(0)] x9) = ya(n) = In [xy (0)] Let us define x;(2) such that 2y(n) = a(n) + Bran) then the output becomes, Js@—)= Infx3()] Js(e)= Inf (n) + xy (0)] (a) # ay (m) + bya (a) Therefore, the system is nonlinear. @ Y=) —x(0-1) Let us define x,(m) and x, (7) as 4) > y(n) =f) —3 (0-1) 2, (0) = Jal) = (0) — 9-1)188-76 @ Digital Signal Processing Let us define x;(n) such that 2q(n) = ar(n) + be (n) then the output response becomes, ysl) = 25 (2) —4 (0-1) : dale) = [ax n) + bxz (n)] [ax (n= 1) + 8x (2—1)] Jnl) = aly (n)—x(n-1)] +4[xp(n)- a(n )] ale) = ay Cn) + bya (ny Therefore, the system is linear. i) ym) = 70) +27 (9-1) Let us define x,(n) and x,(n) as (n)— x} {n-1) 22) = 22m) = E(n) 3} (2-1) (8) => y(n) =3 Let us define x,(n) such that 35 (1) = ax (m1) + x (7a) then the output becomes, 230) = 5 (0) 4x5 0—D) n(n) = [ax (n) + bx (MP + [ax (2-1) + bx, (1)? aCe) = [ex (np 939 (9) + 2ab (nC) +a’ (1) + Pad (2-1) + 2ad x (2-Day) ys) =a [33 ny +37 2-9) ] +0 deaf =] + 2ab[x,(n).x, (0) +\(n-D,(7-1)] Js() # a(n) + by2(n) Therefore, the system is nonlinear, (iv) yn) = 2x) +4 Let us define x(n) and x9(vi) as Alm) = y(n) =2x, (0) +4 (0) => y (1) = 2x (0) +4 Let us defi th us define x,(n) such that ape ant+bs, (0) then the output response becomes, Ja = 2x5 (1) +4 . Y3() = 2 [ax (n) + bx (m)] +4 y(n) = 2ar,(n) + 2bxy(n)-+4 ¥s(m) * ay (1) + by, (2) ‘Therefore, the system is nonlinear.Introduetion to Signals and Systems @ ISS-77 Problem 2.53 Determine whether the following continuous-time systems are linear o not. 0 £0280 ,)- Moun @ FAO PO. y= 1202 Solution o£ SP 2D 9-Day Lats define en ies an SAO 2814 y, (= “a % (99 20 42th 2200.4 y, (y= 22, 20450 Let us define x3(¢) such that (05 ax (O+hO then the output response becomes, +40 a y,(0) 2 2250. ae t? tye £50,.910, rid =L fas (0) + bag(O]+ [a5 (0 +b, (0] : : x ee 1 ype aSill pO van neds en £20 a, A= fis. olufee (0) +0] Therefore, the system is linear. @ & 3 + 0 AO x0 Let us define (0) and x,(0) as 5092 4 BO» 9-1 HO 540 2) 20 4p BOs ye Or20188-78 @ Digital Signal Processing Let us define x,(1) such that xy) = ax (+52, (0) then the output response becomes, 2, LO MO. = dam ae = Shas + be O}+ fay (+6500) =o BO Joo Se HO | anierebig Fy), OO, noma AO sco}eel 20.0] de de Therefore, the system is linear, 2.6.7 Causal and Noncausal System Causal system Ina causal system, the output response of the system at any time depends only on the present input and/or on the past input, but not on the future inputs. In a causal system, the next input cannot be predicted. Hence, this may not be an essential condition for all systems. Examples for causal system are an) = x(n) =x) JO = (1) In the noncausal system, the output response of the system depends on the future input values also. Examples for noncausal system are y(ny= xin) JO = x(14 1) ~x() SOLVED PROBLEMS.» -§ /§ -> 5 5 Problem 2.54 Check whether the following systems are causal or not. © w= @) y=?) 4 (yy Me) =a¢sins) © vo fryer wi) y= 40 Solution ~ @ p=ato) Let 1= 0,then (0) = 0Introduction to Signals and Systems @ 1SS-79 t= 1, then y(1)=x(1), t= =I, then y(—-1)=-x(-1) For all values of ¢, the output depends on present and past values of the input. Hence, the system is causal. @ y=?) Let then y(0)=x(0) then y(+1)=x(1) i=—l, then y(-1)=x(1) 1=2, then y(2)=x(4) In the last two cases, the output value depends on the future value of the input. Hence, the system is ‘noncausal. Gi) = 70) Let 1=0, then 3€0)= (0) tel, then ¥-1)= C1) f= 1, then y= 8) ‘The output value does not depend on the future value of the input. Hence, the system is causal. (iv) QO) = xGsine) Let 1=0, then (0) =x {sin (0)] =I, then (1) = [sin (-)] 1 1, then 1) =xfsin(1)] The output value does not depend on the future value of the input. Hence, the system is causal. 4 x= J x@ar Let £=0, then WO)= J x(r)de= x(r)]!, = x(0)—x(-—) a= 1,then 7 VUD= f x(e)de= x(n), =3(4)- 2(-<) 4=-1, then W-D= fxd = x(n =3(-4)-x(-=)188-80 @ Digital Signal Processing In the second case, the output value depends on the future value of the input. Hence, the system is noncausal. (CO) 9-20 #0) Let 1= 0, then y(0)= = 1,then YC) 40 1-1, then yC1)= “cn The output value does not depend on the future values of the input. Hence, the system is causal. Problem 2.55 Check whether the following systems are causal or not. ws OH) =xe)—aer=1) GD mp= a") Gn) = Yate) & cm yom= S atm) () se)=x(0) (0-1) Solution “— 0) =3(0)—a(n—1) Let _m=0, then (0) = x(0)—x(-1) The output value depends on the present and past values of the input but not on the firture value. Hence, the system is causal. (@ Aln)= x(n!) Let_n=0, theny(0) = (0) n=l, theny(1)=2(1) _ ae aL, then y(—1) = 2(D"?] n=4 , then y(4)=x(2) ‘The output value depends on present or past values of the input, but not on the future value, Hence, the system is causal. y= Hae) is y(n) = sews x(0) + x(I)+2(2)+....4.2(0—) ‘The output depends on the present and past values. Hence, the system is causal,Introduction to Signals and Systems @ {SS-61 te wid= Fa) — yo= ‘The output depends on the future values of the input. Hence, the system is noncausal. @) Yea) = x(n) x(n 1) Let =0, then (0) = 2(0)2{-1) n= L.then (I= (1) 340) ==1, then y{-1) = 3-1) (-2) The output depends on the past and present values of the input, Hence, the system is causal. 3e(—2) + x(—1) + (0) + x1) + x2) +...2(- 1) @ 27 INTERCONNECTION OF SYSTEMS Interconnection of systems is considered to be one of the most important aspects in system design. In general, the actual system is always divided into many subsystems which are intercbnected such that the actual work will be executed smoothly. Computing a larger systems involves longer time duration and increased complexity to analyse the system. When the system is divided into smaller modules, the complexity involved in the smaller module is lesser than considering the entire system at time. Due to the advancement of parallel computation concept, the module based system analysis work faster than considering entire systern. Following are the general configurations, in which any subsystem can be connected. Fig, 255, Invreonnoction of Speers () Sertes (Cascade) Connection (b) Parallel Connection (c) Series-Parallel Connection188-82. @ Digital Signal Processing Fig. 2.55(a) illustrate the subsystems connected in series. Fig. 2.55(b) illustrate the subsystems connected in paraile!, A serial-paralle! combination of subsystems is shown in Fig. 2.55(c). Let us consider a general coramunication receiver system, as shown in Fig. 2.56. Fig. 2.56 Receiver of a Communication System ‘The receiver system consists of many subsystems like receiver (subsystem 1), local oscillator (subsystem 2) mixer (subsystem 3), filter (subsystem 4), and amplifier (subsystem 5). All these subsystems are connected either in parallel or in series, in order to execute the assigned process. Ifthe systems are not connected in proper configuration (series or parallel), then the output may not be the expected one, though each and every subsystem is individually working. CHAPTER SUMMARY © A signal is defined as a function of one or more variables, which conveys information. * Asystem is an entity that manipulates one or more input signals to perform a function, which results in a new output signal, © The signals can be classified as = Continuous-time (CT) signal and Discrete-time (DT) signal = Periodic and Aperiodic signals = Even and Odd signals = Deterministic and Random signals = Energy and Power signalsIntroduction to Signals and Systems @ © Differentiate discrete-time signal and digital signal Diserete-time Signal Digital Signal ‘A DT signal is obtained by sampling a CT signal ata uniform or non-uniform rate. Assignal x(n) is said to be DT signal ifit defines or represents an input at discrete instants of time. The DT signal is discrete in time only. For a DT signal, the amplitude varies at every discrete values of ‘n’. A digital signal is obtained by sampling, quantizing and encoding a CT signal. A signal is said to be a digital signal if it is represented in terms of binary bits ("0° orl’) The digital signal is discrete in time and quantized in amplitude. For a digital signal, the amplitude is represented as a high voltage if the bit is "1 and low voltage if the bitis «Differentiate energy and power signal. Energy Signal Power Signal The energy of CT signal x(/) over a period ~fersek is given by “ay E= Lt f |xafa Toei The energy of DT signal x(r») over a period -N SnS4N isgivenby a, E= lt A kel a A signal is referred to as energy signal ifand condition 0S £.. Se. Generally deterministic and aperiodic signals are considered to be energy signals. ‘only if the total energy of the signal satisfies the The power of CT signal x(r) over a period -fsts « is given by 1 ey Lies F Poe Ty b(n? ar The power of DT signal x(n)over a period ~N£nS#N is given by 1 2 swe > be eW ‘A signal is referred to as energy signal if and only if the total energy of the signal satisfies the condition 0S Fg $2. Generally random and periodic signals are considered to be power signals. 1SS-8318884 @ Digital Signal Processing Differentiate even and odd signals. ‘Odd Signal Even Signal ‘A signal is said to be odd if x(t) ==x(=1) for CT signal a(ny=x(-n) for DT signal “The odd coinponent of any signal is Jolt) = so for CT signal (p= AE) for DT signal Odd signals are anti-symmetric about the vertical axis Example: Sine wave Differentiate random and deterministic signal. Asignal is said to be even if x(t)=.x(-1) for CT signal a() = x(n) for DT signal ‘The even component of any signal is (= 2O2C2 for CT signal (np = OD for DT signal Even signals are symmetric about the vertical axis Example: Cosine wave Deterministic Signal Random Signal ‘A deterministic signal is one in which there is a certainty with respect to its values at any time. Future value of signals is predictable. Eg. Pulse train, sinusoidal wave etc, signals can be expressed mathematically. “A random signal is one in which there is a uncertainty with respect to its values at any time. Future value of signals is unpredictable. Eg, EEG signal, Noise, Speech etc. Random signals are_—_expressed mathematically in terms of impulses. The fundamental period of Noof a DT signal x(n) is the smallest posi value. je value of W for which x(n) = x(+¥) or the DT signal exhibits periodicity. It is defined as Ny = gente mis an integer ‘The fundamental period Ty ofa CT signal x(¢) is the smallest positive value of T for which x(t) = x(t+ T)the CT signal exhibits periodicity, It is defined as 7) = an Basic operations performed on dependent variables of signal = Amplitude scaling = Addition of signals = Multiplication of signals = Differentiation of signals = Integration of signalsoe Introduction to Signals and Systems @ 188-85 Basic operations performed on independent variables of signal © Time scaling = Reflection of signal © Timeshifting Amplitude scaling is considered as the multij ‘ion of scalar value a with continuous-time signal x(1) [x() for discrete-time signal], that is, c x(¢), [cc x() for discrete-time signal]. If the value of a > 1, then signal is said to be amplified. [f the value of ci < 1, then signal is said to be attenuated. During amplitude sealing the tine scale remains the same. When any two signals are added, their amplitude will be changed but the time scale remains unaltered. When any two signals are multiplied, their amplitude will be changed but the time scale remains unaltered. When any signal is differentiated, their amplitude will be changed but the time scale remains unaltered. When any signal is integrated, their amplitude will be changed but the time scale remains unaltered. Time scaling is considered as the multiplication of scalar value o with the time function of continuous-time signal x()[x(n) for discrete-time signal], thatis, x(a) x(a) for discrete-time signal] Ifthe value ofa.> 1, then signal is said to be compressed. If the value of a < 3, then signal is said to be expanded. During time scaling the amplitude remains the same. A continuous-time signal x(0) (x(n) for discrete-time signal) is said to be shifted right side, ifit satisfies the condition x(¢—1,)[2(—m,) for discrete-time signal] and shified left side, ifit satisfies the condition x(+1,) [x(ortn,) for discrete-time signal]. A continuous-time signal x(-1) [x(—m) for discrete-time signal] is said to be shifted right side, if it satisfies the condition x{t+/,) [x{+n,) for discrete-time signal] and shifted left side, if it satisfies the condition x(t—1,) [x(n—n,) for discrete-time signal]. ‘Areal continuous-time exponential signal in its more general form represented as x(t)= Be where B is the real scaling factor and o: is the real parameter. Fora. <0 the magnitude of the continuous-time exponential signal decays exponentially. For a >| the magnitude of the continuous-time exponential signal rises exponentially. ‘Areal discrete-time exponential signal in its more general form represented as x(n) = Ba" where Bisthe reat scaling factor and a is the real parameter. For a >! the magnitude of the discrete-time exponential signal rises exponentially. For 1 > a >:0 the magnitude of the discrete-time exponential signal decays exponentially. 120 0 r<0 The continuous-time step function is denoted by wom 1, 20 The continuous-time step function is denoted by u(m)= {i on ‘The impulse function is denoted by &(n) = ft n=0 0, n#0 120 0, 1<0 ‘The continuous-time ramp function is denoted by r()= in, n20 ‘The discrete-time ramp function is denoted by r(n) 0, n<0188-86 @. Digital Signal Processing * The system can be classified as © Continuous-time and Discrete-time systems = Stable and Unstable systems = Memory and Memoryless systems = Invertibte and Non-invertible systems = Time-invariant and Time-variant systems = Linear and Non-linear systems = Causal and Non-causal systems © Ifthe input and output of the system are continuous-time signals, then the system is called “continuous- time system”. © Ifthe input and output of the system are discrete-time signals, then the system is called “discrete-time system”. © Assystem is said to be stable if and only if every bounded input produces a bounded output. ‘A ssystem is said to possess memory if the output of the system depends on past and future values, # Asystem is said to be an invertible system if the input signal given to the system can be recovered from the output signal of the system. * Assystem is said to be time-invariant ifthe input signal is delayed or advanced by any factor that leads to same delay or advancement respectively in the output time scale by the same factor, that is, the system responds to an input at any instant of time and results in an output, ‘© Asystem is said to be linear if it satisfies superposition, scaling and additive property, that is, the response to x (0+¥,(is (0+ 42€0 (additive property) and the response to ax,(1) + by,(0) is ay,(4) + by,(9 (scaling property), where aand 6 are complex constants. The same definition holds good for discrete-time system also. © Assystem is said to be causal if the output response of the system at any time depends only on the present input and/or on the past input, but not on the future inputs. — REVIEW QUESTIONS — 1. Define signal. 2. Explain one-dimensional signal with suitable examples. 3. Explaintwo-dimensional signal with suitable examples. 4. Explain multi-dimensional signal 5. Distinguish between continuous-time signal.and discrete-time signal. 6. What is the basic difference between discrete-time signat and digital signal? 7. How do you classify signals? 8. Differentiate energy signal and power signal. 9. Differentiate even signal and odd signal. 10. Expl periodic signal.|. Differentiate determini: Introduction to Signals and Systems @ 88-87 ic signal and random signal. 1. Test whether the following signals are periodit or not. If'yes, what is its fundamental period? 2s ( @ xopasi' (Fae ) xm=zbe © x()aerur @) w=[si{Ere2] © x)= om( F}ese( 5} 3 3 }. Find the odd and even components of the following signals. (@) x(n) = {2,4,6,8, 10) (b) x(y=esar?) (©) x(0)=sin2¢+cos2r @) x(m=0" |. Draw the odd and even components of the given signal. nd ; \_ oT23a456 (a) x() =wose jor 23456 © . Test whether the following signals are energy signals or power signals. (a) x()=sin u(t) (db) x(t)=0? ede) £,.2 (c) x(t)=sin3¢ (@) xia)=e 4 Vuln) . -AEn4) ©) x(a) = ula) -u(n-8) O xijse 1° May (@) x(n) =cos4n un) xio=($] wen18888 @ Digital Signal Processing 16. Perform addition and multiplication on the following signals. x0 . xa) 2 2 1 1 =e = et ® at) 3 et tr 2 a7! 2 ae + 2 aS ) xn) An) 3 3 1 1 RO CORGGeee i. a-a-a-1 | tes 4 6 423-251 | 12 3 4 6 17, Find (0) for the given signal. “ @) WO=x51+6)Introduction to Signals and Systems @ 188-40 18, Find y(¢) for the given signal. @) yO=xH+4) © WO=x0-4) oe er) 19, Define the following signal mathematically and represent graphically. (@)_ Impulse signal (b) Ramp signal (©) Step signal (@) Sinusoidal signal (©) Exponential signal with various time periods . 20. Sketch the continuous-time signal x(¢)=10cos(2a1) for the interval 2 >> 0 and sketch the corresponding, discrete-time signal with a sampling period 7=0.1 s. 21, Sketch the continuous-time signal x() =e for the interval 2 > ¢ > ~2 and sketch the corresponding discrete-time signal with a sampling period T= 0.2. 22. Define a system with suitable examples. 23. Give a broad classification of systems. 24, Define stability of a system. 25. ‘Test whether the following systems are stable or not, (@) An) = b"u(-n) ©) A(n)=4""u(n~2) ©). Ai)=Pe" ule) @) We) =ésine ult) © AD=e™* MO 0 h(n) =e"? wn-6) 26. Define amemory system with suitable examples. 27. Test whether the following are memory systems or not. 2 @ HO=xO4+x0+) ) woe © WO=xa-Dexl-) — @) Date) © =F.) y(n) =e"u(n)188.90 © Digital Signal Processing 28. Define a time-invariant system with suitable examples. 29. Test whether the following systems are time-invariant or not. @ yO =coslx(e] () Cm) =mx(n) ©) y(n) =2(n)cosan () yny=er™ ©) yO=vt-D+2%yt-2) 29 ayant) © & Pam (hy yln)=tog|>0] w. Defines linear system with suitable examples. 31, Test whether the following systems are stable or not. @ pa)=ntx(ny (0) y(a)=nx(n-2)+k © yn=e™” +k @ Bo (0), 620 wo PO 5 2549 = 20. asx © 4 oo D pay Beh du 20 24 = 8049 i 32. Define causality with suitable exaniples, 33. ‘Test whether the following systems are causal or not. @ yep = mtn) (©) y(n) =2(6082n) © vm=r) @ xe Seat) © yQ=x)xt-2) Ow 2 34. Explain the various configurations in which systems are connected, justify their advantages and mention their applications. 35. Show whether the system y() < nx(7) is a linear time-invariant system.LTI Systems Linearity and time-invariant properties are considered to be important in analyzjng and real this chapter, the relation between input and output that satisfies the linearity and time-invariant properties of the system are discussed. In this chapter, term “convolution sum” is introduced, which gives the mathematical relationship for the input-output. The input-output relation is explicitly discussed both in discrete-time and continuous-time. @ 3.1 DISCRETE-TIME LINEAR TIME-INVARIANT SYSTEM The impulse response is the output of the Linear Time-invariant (LT1) discrete system when the impulse 81) is applied to it. 1, n=0 S(n)= 6.) 0, elsewhere The impulse response completely characterizes the behaviour of any LT! system. The discrete-time unit impulse can be used to construct any discrete-time signal,LTH2 @ Digital Signal Processing 3.1.1 Representation of Discrete-time Signals in Terms of Impulses Let us consider the product of a signal x(n) and the impulse sequence &(), written as x()(n) = x(0 (0) G2) From equation (3.1), itis clear that the impulse sequence exists only at n= 0. Therefore, 1x(n)5(n) can be rewritten as x(n) 6(0). Though the impulse sequence exists at n= O, the input signal x(n) exists in the remaining samples. Simitarty, x{-2)5(n+ 2) x(-1)6(0r+1) 2(0)5(n) x()5(a=1) 3(2)5(n-2) Therefore, the generalized relationship between the input signal x(n) and the shifted impulse sequence is given by x(n) (eK) = x(k) S(n—k) 63) In equation (3.3), x(n) represents the input signal but x(A) represents the magnitude of the input signal x(n) _at time A. In equation (3.3), the product of input signal x(n) and time-shifted impulse 5(n—&) results in the time-shifted impulse 5(n— &) whose magnitude is the value of the signal x(n) at time &, that is, x() represents ‘the magnitude of the signal at k and 5(—k) represents the position of impulse signal. Let us analyze the above statement graphically by considering a random signal as shown in Fig. 3.1. Itis clear from Fig. 3.1, that the signal x(1) can be decomposed into the product of time-shifted impulse and signal x(n}at k i.e. (0) = 4 245 (4 4) 4 0(-3)5 (443) + x(-2)5 (0+ 2) + x€-1).5(n +1) +2(0) 50) tx(l) (a1) + 3(2)6(2-2) 4.06 x(n)= = ES n— b Ga) ‘Any signal can be represented as a time-shifted impulse sequence 5(1—1). the input signal x(n) = n(n), ie. {i 20 u(r) = 0, elsewhere then any unit step signal can be represented as u(n)= F5(0- 8) osx(t) 8¢m-1) LTIsystems @ LT (6) ST TTT ‘7° te) Fig. 3.1 Signal Decomposition 3.1.2 Convolution Sum Let us consider a system, say H[>}, to which the input signal x(n) given in equation (3.4) is applied. The output signal becomes An) = Hat} vn)= uf 3 30-8] 36) x(o) ——>| Hie] j——> vin) Fig. 3.2. System RepresentationLTL4 @ Digital Signal Processing By linearity definition the term convolution describes, how the input signal interact with the system to produce the output signal. It is particularly useful to consider the output from the system owing to an impulse input. This is because any input signal may be represented as a sequence of impulses of different strengths. ‘The operator H[-] operates only on function but not on magnitude of the signal x(4). Therefore, yoy= 5 sy niser—¥] & ved= FY OAC en where A, (2) = H[5(0—4)] Ifthe system is an LTI (Linear Time-invariant) system, then the time-shifted input results in a time-shifted output, . h(n) =h,(n~k) G8) The output due to the time-shifted impulse is a time-shifted version of the output due to an impulse. Let f(a) = frm), then h(n) = hin-b G9) ‘Therefore, equation (3.7) can be written as, von= Fx Kn-#) G.i0) The equation (3.10) is called the ‘convolution sum’ and is denoted by the symbol +, i.e. vn) = Sx) h(n) = x¢n)+ hen) Gun Se . SOLVED PROBLEMS__-._»_=> = Problem 3.1 Express the given signal sequence as a time-shifted impulse. x(n) = {1,-2,8,4,5,3.7} tT Solution In general, the arrow * 4’ shows the value of the data for m= 0. * 0 1 2 3 a ar) [1 From equation (3.4), it is clear that any signal can be represented as linear combination of magnitude and position/type. a|[s ny z x(k) O(n) x(my= FY) x(k) B(n—k) = 2x7) = 2-3) 6(4 3) + x(-2)H(N4+-2) 4.x S(n+ 1) +0) 5(m) + (1) 5(r2— I) + x(2) 59-2) +x(3) (9-3) x(n) = 51+ 3) 26(n+2)+85(n+1) + 45(+1) + 58(n—1)-3 51-2) + 78(n-3)LTISystems @ LTLS Problem 3.2 Express the given signal sequence as time-shifted impulse. x(n) ={2,3,0,7,8,- 15,18, 20} t - Solution ™"|/2;1/o0;'1]/2/3][4][5 . BoT2 [3 Tolt7T[e [as] [20 From equation (3.4), itis clear that any signal can be represented as linear combination of magnitude and position/type. x(n) = Y) x(k)5(n- ky x(n)= = x(K) 5-4) Hn) = NCD) AU+2)+ MDA (N+N +410) 40) +HElM—N) +x(2)5 (1-2) + x(3)5(n—3) + x(4) 3(n— 4) + 2(5) 5-5) X(t) = 25(1+2) +3 (n+ 1) +7 5-1) +8 d(n —2) — 15 6-3) + 185 (n—4)+205(n—-5) @ 3.2 PROPERTIES OF LTI SYSTEM 3.2.1 Distributive Property Let us consider two LITT systems with impulse responses /,(n) and h(n) connected parallelly as shown in Fig. 3.3. = x—4 in}= y(n) + h(A) -——> yin) (b) Fig. 3.3 (a) Systems Connected in Parallel (b) Equivalent Representation By definition of the distributive property, £(19) [44 (1) + b(n) = x0) A () + 2) AC) G12) Proof The output of the first system is . Yin) = x(n)» h(n) larly, the output of the second system is by(a) = x() + A, (0) Si1 LTL8 © Digital Signal Processing The overall output of the system y(n) is given by Wei) = y(n) + y(n) Wn) = x(n) * hy (in) + x(n)o y(n) By the definition of convolution from equation (3.11) - ny= Dxbae- be 3 Amt Ay y= z x(b)TA (2- +h, (nk) = z x) A(n—k) where nk) = Weber hy = Win) =A (n+ hy) Therefore, Yer) = Sayin k= x(a) oh) i If two systems f,(n) and h,(7) are connected in parallel, then the impulse response of the system to the input signal x(n) is equal to sum of the two impulse responses. 3.2.2 Associative Property ‘Let us consider two LT1 systems with impulse responses f, (7) and h,(n) connected in series as shown in Fig. 3.4. ee a ETT) vie) ® tb) Fig, 3.4 (a) Systems Connected in Series (b) Equivalent Circuit By definition of the associative property, Ex) + 4) «Ay G) = XC) # (0) yD) 3.13) Proof The output of the first system. Y (1) = x(n) #8 (n) (3.14) Similarly, the output of the second system y(n) = HC) + Ay (2) (3.15) Substitute equation (3.14) in equation (3.15), hor) =[xCa)+ by (a)] + y(n) = x7) «fy (n+ fy (0) 3.2.3 Commutative Property Letus consider two LT! systems with impulse responses h,(n) and /,(n) connected in seties as shown in Fig. 3.5. xn) hy) thefoy|— yee) = 4m) — ele ny(o9 |} — yin Fig. 3.5. Systemis Connected in Series. . LTI Systems @ LTI-7 By definition of the commutative property, yin) # hy (72) = ha (nl) #4 (12) . G16) aye hy (ny =D) ACen (nk) ro Let n-kam—skan-m Alndohgiaye Y A(n—myhglm) A(r)# y(n) = z Ay(m)iy(n—m) Ay (a) hylan) = hy(n)a y(n) Equation (3.16) finds application in solving convolution problems. This property can also be extended to signals, i.e. U(n) = x(n) * h(n) = h(n)» x(n) GID @ 3.3 PROPERTIES OF DISCRETE-TIME LTI SYSTEM — 3.3.1 LTI System With and Without Memory A system is memoryless if the output at any time depends only on the present input (discussed in Chapter 2). This is true for the L771 system ifand only if Hn) =0, n#0 ‘Let us consider the impulse response of the form Hn) =k (ny where k= A(0), is a constant ‘The output of such a system is given by y= FY x(k) h(n-k) Hint &(n)=1, a=0 a 8(n woe ¥ x(k) k(n k) yr) = klk) G.18) Equation (3.18) isa memoryless LTI system. If h(n) # 0, #0, then the LTI system is called a memory system. 3.3.2 ‘Inverti of LTI System Asystem is invertible only ifan inverse system exists. Similarly, an LT§ system is invertible only ifan inverse LTI system exists.LTL8 © Digital Signal Processing Let us consider the following figure. x0) fin) | HOT ita) win) = x(n) System 1 system 2 Fig. 3.6 Invertibility of LTI System “The system response h(n) results in an output y(n) and the output of system | is given to system 2, whose response f(r) results in an output (7), which is equal to the original input x(n). This is possible if A(n)e h(n) = 6(n) G19) 3.3.3 Stability for LTI System A system is said to be stable if every bounded input produces a bounded output. The statement can be extended to LT systems also. Let us consider a bounded input x(n), ie. [x0n|< My = Problem 3.3 Find whether the system with impulse response h(n) =2e* is stable or not. Solution The concition for stability ig, EX [an] <~LT1systems @ LTH i bl= aed 7 Fn +E 0) E en) 2 [een= vine 584 4 Bl <1 . 1 Lettie Therefore, the system is stable. Problem 3.4 Find whether the system with impulse response h(n) = e* u(n) is stable or not, Solution The condition for stability is E hankco °F late Fe tek setacds 4 ‘Therefore, the system is unstable. 3.3.4 Causal System By definition, for a discrete-time causal LTH system, the imputse response h(n) must be zeto for n <0. The ‘causality can be extended to convolution sum as yeny= Y x(k) Ak) 25) =m Fora causal discretestime LTI system, h(n) =0 for n <0. Therefore, the output ofa causal system must be expressed as y(n) =F a h(a—K) 6.26) & ‘A causal system cannot generate an output before an input is given to the system. @ 3.4 LINEAR CONVOLUTION ‘The convolution equation defined in equation (3.10) can be given as an algorithm: 1, Plot both x(&) and (Ay 2 Reflectsi(t) about k= 0 to obtain A{-R) 3. Shift A(—-&) by n (towards left for —n and towards right for +n)LTO @ Digital Signal Processing Let the initial value of m be negative Multiply each element of x(&) with A(n-A) and add all the product terms to obtain y() Shift A(n-K) by incrementing the value of n by one, and do step 5 Do step 6 until the product of x(&) and A(n-A) reduces to zero nawes SOLVED PROBLEMS_._._-»_»_=>__ Problém 3.5 Perform the convolution of the two sequences x(nJ={1,2,3, 4}: h(M={1 11 atk) ni) , nek) 4 a | ! . UW LT 2 3 123 ° t Solution 4-3-2 Od ® to) © Whenn=-1 nin-ky Multiply the elements of Fig. (a) and Fig. (d) HI) =3€-A) AA) +x(-3) A-3) + x(-2) A(-2) I [ ‘ AIP HCT) +3(0) H(0) + CIV AID+ (2) MQ)+AGYHG) TaTST eT 7 ee a YEN = 041) +01) + 0(1) +001) + 10) +200) + 3(0) + 4(0) } dCD=0 , Whenn=0 hi-k) neo Multiply the elements of Fig, (a) and Fig. (e) HO)= (3) h-3) +342) A{-2)+ 1) AN) +340) (0) SAMI FAOVMD+OMG) RTS TE TT 30) =0(1)+0(1) +0(1)+0(1)+ 1(1) +2(0)+3(0)+4(0) te 0)=1 * Whenn=1 nik) [net Multiply the elements of Fig. (a) and Fig, (f) AL) = (2) A-2) + 2-1) A(—-2) + (0) (0) +x(I)A()+x2) 42) 4+28)4G) “karo TST TE AI) 0(1) + 0(1) + 1(1) + 2(1) +3(0) +40) cu AIR b+2=3 ‘Whenn=2 Muttipty the elements of Fig. (a) and Fig. (g) HQ)= XT) AL) + (0) A(0) + x1) ACL) + (2) H(2)+ 23) AGS) 2) =0(1) + 161) #2(1) +301) +4(0) MQ)V=14243=6 When n=3 Multiply the elements of Fig. (a) and Fig. (h) }43)= 200 (0) + (1) h(1)+x(2) A42) + x(3) 3) ¥3)= 10) +20) +301) + 401) yG)=14243+44= 10 Whenn=4 Multiply the elements of Fig. (a) and Fig. (i) 344) = £0) (0) +x(1) (1) + (2) AQ) + (3) AGB) +214) AA) 4) = 10) +2C1)+3€1) + 4(1) + 001) y)=24344=9 Whena=5 Multiply the elements of ‘Fig. (a)and Fig. j) 35) = (0) (0) + 2(1) ACL) + 2(2) 23+ (3) AB) + (4) H(A) + 25) H(5) 65) = 1€0) + 2(0} +361) +-4C1) + 0(1) + 0(1) HS)=3+4=7 ‘When 1-6 Multiply the elements of Fig. (a) and Fig. (&) 66) = (0) (0) + (1) (1) *:2(2) A(2) + x(3) ACB) + (4) Hd) + x(5) (5) + 2(6) 6) N= 1€0} + 2(0) + 300)+ 41) + 0(1) + C1) +001) Od Whenr=7 Multiply the elements of Fig. (a) and Fig. (1) YT= HOY HO) + (1) (1) +212) (2) +63) HB) +264) AA) + 3(5) MCS) + HCO) MO) +37) (7) ATI=1(O)+2(0)+3(0) + 4(0)+0(1)+ O(1) +061) +061) x=0 The result of the convolution is LTI Systems @ LTH nwo [n=2] asa age ek aie ” [aes] HAGE OF 234k nen [ees aan aa ae K--2-1 0 é 23 ware TEs aes a ni7-W) ne? (nm) = x(ni)* An) = {1,3,6, 10, 9.7.4}LTH12 @ Digital Signal Processing Note If the length of x(n) is m, and the length of h(n) is n,, then the length of the convolution sum is (ntnel). Problem 3.6 Perform the convolution of the given data sequences edi 2.2.1 Mnd=(1, 2, 2,2, 1) T Solution meth, “k-9-2-10 12 9k were a eek Od, When nt =-1, ‘h(n—ky! Aele-1 Multiply the elements of Fig. (a) and Fig, (d) 2 HARD) a(S) AS) + (4 AA) + 2(-3)A(-3) +a(-2) M{-2) 1 x1} AT) + x(0) (0) + .x(1) ACL) + x(2) (2) + 213) AB) WE-2)=0(1) + 0(2) + 0(2) +0(2) +O(1) + (0) +2(0) +200) +100) “K=5-4-3-2-T 0 T 2 TK e2)=0 When n+1=0 Multiply the elements of Fig. (a) and Fig. (¢) YET) x(-4) A-4) + (3) 3) + 2-2) 2) + (1) 1), + (0) A(0) + x( 1) ACL) + x(2) (2) + 23) (3) AD) O(1) + 0(2) + 0(2) +0(2) + 1(1) +2(0) +2(0) + 1(0) cursor es ak (=|) = (e) worl obs When +i=1 nea a Multiply the elements of Fig. (a) and Fig. (9) 2 Y(O)=x(-3) AE-3) + (2) A(-2) + -1) AI) +30) (0) + (1) ACI) + x(2) 2) +23) (3), HO) =0(1) +0(2)+0(2)+ 142) +2(1) +2(0)+ 100) Aimar fa ek O)-2+2=4 ot tnety When n+1=2 ner-#y/ mind] Multiply the elements of Fig. (a) and Fig, (g) 72 41) 3-2) AC-2) + a(-1) AC 1) + (0) 40) +1) (1) +(2) A(2)+x(3) A) ‘ | . W1)=0(1) + 0(2) + 12) +2(2) +2(2)+2(1)+ 100) Raat ae tak WID=2+442=8 * o (nt)LTI Systems @ LTH43 Whena+1=3 . hia-ky a Multiply the elements of Fig. (a) and Fig. (h) 2 21) HEL) + 3(0) (0) + (1) (1) + x(2) HQ)-+ 63913) 2)= OCI) + 12)+2(2)+2(2)+ 1) M2y=2+4+4+1=11 Nase 12s fk Whenn+1=4 wu Multiply the elements of Fig. (a) and Fig. (ip hya-kyt ¥3)= (0) (0) + x(1) ACL) +2(2) (2) 2 | 3 na tn4 +33) 43) +x(4) h(4) 1 EBV 11) + 2(2)+ 22) + 12) +01) Joye +4eae2e It Nae Te ty When rti=5 Multiply the elements of Fig. (a) and Fig. (j) “| mi=8 oF AY 30) A(0) +x€1) (1) +x(2) (2) +03) A(3), 2 +x(4) H(A) +065) (5) 1 HAY= 1(O}+ 2(1) + 2(2)+ 1(2)+ 0(2) + 001) WAHZ+4+2=8 koe When n +1=6 (nel) Multiply the elements of Fig. (a) and Fig, (k) bet 65) = (0) (0) + 2(1) (1) +3(2) 2) +3) (3) +0(4) 14) 2 +x(5) M5) +246) M6) HS) HCO) +2(0) +2(1)+ 12) + 0(2) +012) + 10) OVE 2+2=4 Katte eas 6 Fk ww inst = 6| When a + 1=7 Multiply the elements of Fig. (a) and Fig. (I) hi6-K) Y6}= x(0) (0) + x€1) A(1) + x(2) (2) +3) AB) +38) ACA) 2 +.x(5) (5) + (6) M6) +27) HM) 1 MG}= 1(0} + 2(0) + 2(0} + 1(1) + 0(2) + 0(2) + 0(2) +0(1) | | y46)=1 kotes«¢s67a8k Whenn+1=8 o Multiply the elements of Fig. (a) and Fig. (m) tos) (ne) HTD HCO} C0) + (1) H+ x(2) (2) + x03) (3) + x64) ACA) jee +2(5)M5)+x(6) #46) +3(7) (7) +18) HS) . WT) = 1(0)+ 2(0) + 2(0)+ 1(0) + OC 1) + 0(2) +.0(2) +0(2) + 0(1) 1 HD=0LT-44 © - Digital Signal Processing The result of the convolution is Yr) = x(n) A(n) = {1, 4, 8, 11, 11, 8, 4, 1 Problern 3.7 Perform the convolution of the given data sequences x(n) = {1,—2,3,-2}; h(n) = (2, -3,4} tT t Solution a) ne tk) 4 4 4 a 8 8 2] 2 2] 1 1 1 | 2 -k -1 jot k -k -1 fo1 2 k -k =-1 [0 t 2 k (net) 2 oa 3) o © © When n=-3 (or w+1=-2) mony mts-2 Multiply the elements of Fig. (a) and Fig. (d) Y-3)= aA) WA} + a3) {-3) + 2-2) H{2) 8 + x(=1) A(—1)+x(0) (0) +x(1) AC 1) +2) AQ) 2. -3)=0(4) +0(-3) +0(2) + 1(0)+ (-2)(0) +300) +-2)(0) ' Y-3)=0‘When n=—2 (or n+l =-1) Multiply the elements of Fig. (a) and Fig. (e) YD) = 3-3) M3) + (2) Af-2) + K-11) + x(O) (0) (1) A(E) +2) AQ) W-2)=0(4) +0 (-3)+ 1(2) + (-2) (0) +3(0)+(-2)(0) y2)=2 When n==1 orati=0 Multiply the elements of Fig. (a) and Fig. (f) C1) = a2) 2) +31) At D+ a0) H(0) + a1) (1) +212) A(2) M1) = 004) +1 (3) + (-2) (2) +3(0) + (-2) 0) Wel)=-3-4=-7 ‘When n=0(orm+l=1) . Multiply the elements of Fig. (a) and Fig. (g) 40) =x(-1) A=) + (0) (0) + 1) AL) +212) 2) 40)= (4) +(-2)(-3) +302) +€-2) 0 0)=4+6 +6=16 LTl Systems @ LTHMS " (ne) ® h-t-k) m=O) 4 3 2 1 xa r H+) -3 1 0 bi-k)t “k-a 3-2-1 | lo 1 k for) rLTH © Digital Signal Processing When n= 1 (or +1=2) Multiply the elements of Fig. (a) and Fig. (h) MU)=HE1D AC-1) +5(0) (0) + (1) C1) + 32) 42) AN)= 10) +(-2) 4) +33) MI)=-8-9-4=-21 When 1 =2(orn+1=3) Multiply the elements of Fig. (a) and Fig. (i) 2) = (1) A(-1) +x(0) (0) +31) (1) # (2) AQ) +1) AG) a 2 942)=1(0) + (-2) (0) + 3(4) +42) (-3) + 02) 2)=12+6=18 (S-kD} insted 4 When n=3 (or n +1=4) Multiply the elements of Fig. (a) and Fig. () a (=1) (1) +x(0) #40) + (1) A(1}+ (2) A(2) + x(3) AG) +004) ¥B)= 10) + (-2)(0) + 3(0) + 4-2) (4) + O(-3) + 0(2) 33)=-8LTI Systems @ LTH? Whenn=4 (orn+1=5) Multiply the elements of Fig. (a) and Fig, (k) WA) = a{-1) AD) + (0) (0) + (1) AL) +-2(2) WCQ) + (3) AG) Ht H4)= 1(0) + (230 + 3(0) + (-2)0-+ 044) + 06-3) + 0(2) w4)=0 The result of the convolution is nn) = x(n) #A(n) = {2, ~7, 16, -21, 18,-8} @ 35 LINEAR CONVOLUTION USING CROSS-TABLE METHOD Let us consider the convolution, (n) = x(n) +hin) where x(n) = {x,(0),x,(a),,(n),..} isthe input signal and h(i) * {F(n), (4), f(n),..} is the impulse response. The convolution of x(n) and A(n) can be performed as x0) x(n) x0) x(n) hyo) + * fo) hn) o-” Tey(rd ate) Hgfe ln) = g(a) han) Fig. 3.7 Convolution of x(n) and hin) Procedure |. Multiply each row element with column element 2. Draw a diagonal line as shown in the Fig. 3.7. 3. Add the diagonal terms Hea)= {x (0) A (r), {le (0) A) +40) hy} (0) 4) +0) Al) + 0) A}, 0dLTH18 @ Digital Signal Processing SOLVED PROBLEMS __-_=§>_-»- = EE Problem 3.8 Perform the convolution of x(n) and A(n), where x(n) ={1,2,3, 4} and A(m) = {1,1,1,1}. Solution xto) —<“sS hin) Fig. 3.8 dn) = x(n) * h(n) HQ) = (1 (142),(1+2+3), (+243 +4), (24344), G+ 4),4} (A) = {1 3,6, 10,9, 7,4} Problem 3.9 Perform the convolution of x(n) and An). x(n) = {1,—2, 3,4}, An) = {4,—3, 2-1} Solution xin) Ae * y(n) = x(n) *A(r) yn) = (4, 11, 20, 30, 20, -11, 4} @ 3.6 LINEAR CONVOLUTION USING MATRIX METHOD In this method, the data sequences are represented as a matrix. If the length of signal x(7) is N, and the length. of the impulse response A(n) is N,, then the matrix X can be obtained from x(n), whose order will be (N, +N -1)xM, and the matrix H can be obtained from h(n), whose order will be N,x1 such that ¥ = XH. Let us understand the convolution using matrix method with the following examples.LTI systems @ LTH19 SOLVED PROBLEMS__$-_-_-»_$»_»____ Problem 3.10 Find y(n) =x(n)* h(r) using the matrix method. x(n) ={I, 2, 3, 4}; A(m) =f, 1, 1, 1) Solution 1000 2100 i 3210 1 X=/4 3 2 1] #=| 0432 ' 0043 0004 1000 1 1 210 0f- J241 3 3.21 OF) [34241 6 Y=XH=|4 3 2 1] [=| 4+3+2+1/5]10 o 43 all!| |aese2 9 00 4 3] | ass 7 0004 4 4 Problem 3.11 Find the convolution of the following data sequences using the matrix method. x(M) = (1.2, 3.4}; Aa) = (4,-3,2,-1} Solution 10 0 0 4 2 1 0 of. |-u 3-2 1 off .} | 20 Y=NH=|-4 3 2 1 2 [=| -30 04 3 2 “4 20 00 4 3 =i oo 0 +4 4 Problem 3.12 The impulse response of an LTI system is won=(2) u(x). Determine the output of the system (1) at (i)n=—2 (ii) n=2 and (iii) m=+ 4, when input signal x(n) = u(n). Solution By convolution sum, ya) = Aan) *.x(2) = x(n) * A(n) (Commutative property) deny= YJ) x(nyhn—&)LT1-20 Forn=-2 Forn=2 Forn=4 Digital Signal Processing d= Z un) fy] so3) M-2) =0 22 81 0-()d"-a(35) Problem 3.13 ALTU system has the impulse response f() = /"u(), ore. Determine the output of the -system when the input x(n) = (7). Solution Va) = (1) * h(n) = h(n) x(n) i vn) higeahutk Whenn<0 xia fe 4), k<0 ‘ _J at, we={ 0, (Commutative propery) Fak) x(k) i) = uf) 1 pre e4sk x(n—k)f otherwise kz0 otherwise a a Since x(11-H)and A(d) has no overlap terms, therefore output is zero, that is, ¥(n)=0, neoLTI Systems @ “LTF21 Whena>0 “x(n-k) ym) = Ax (a= 8) im v= Sata ‘The lower value of x (vA) is still exist in the negative a-axis, where as the value of h (k) is zero in these region. Hence the lower value of summation takes the value 0. The upper value of summation is taken as variable 1 duc to the fact that 1 can take any value from to a9. Problem 3.14 An UTI system has the impulse response’ Ai(v) = au(2), O 20 [range of Mn — k)iscompared with x(&)] wey = Sxtky Mn) i g2nz0] hin-k) vn) = The lowest value of /41-k) still exists in the negative x-axis, where the value of x(k) is zero. Hence the lower value of summation takes the value zero. The upper vatue of i(t-&) can take any variable ‘n’ between 0 109, hence it is simply 1. Hence the upper value of summation takes the variable value n.LTI Systems @ LTL23 ‘When n>9 NO iacat fora se ee 7 BOND a 2 ven) =F x(hyiKn~ ky & The lowest value of h(n-k) exists in the negative x-axis, where the value of x(k) is zero. Hence the lower value of summation takes zero. The upper range of h(n-k) has exceeded the maximum range of x(k), (Le. 9}, hence A(ra-K) becomes zero for > 9. Therefore, the upper value of h(e-A) is limited to 9, vend Sa . e= aS (Von) Ss 0, ned or] Wal) gz nz0 ynd= 1-(Vay aro Problem 3.15 Repeat problem 3.13 by interchanging the position of x(v) and A»). Solution mn) xk) Lp ews | or 2 346 k ok 10-9 0 2 6 ce elLTL24 @ Digital Signal Processing When n <0 . x(dek) 9 1 oru(n), k20 A= “ { 0, otherwise | uk), kEn -)= vn) { 0. otherwise i < a ee es Mny= FA) x(k) =0 ~ t (0-9) A Both /(n) and x(n) have no overlap terms. Therefore output response is zero. When 9> 120 nent 1 Ak) x(n) . nl } 9>n20 - 1a ~ Recs ser for 2 8 ole ! The lowest value exists in the negative x-axis, where the value of x(k) is zero. Hence, the lower value of summation takes 0. The upper range of x(/1-A) is well below the maximum range of f(A), hence the upper value ‘of summation takes the variable value 1. When #29 x(n-k) of 23456789101 2 k | (n-9) n In this case, the lower range of x(k) has exceeded the lower range of i(k), hence the lower range of summation must be (rt~9). The upper range of x(n-A) is less than the maximum range of fi{) (j.e.c0), the upper range of summation must be n itself. . a wa)= DF hnxin—k)= DY at ars aes Letk—-a+9=0=m ’ . y= Sa. LTisystems @ LTH2s en) nal S a" 3(n) = ( } =a 0 jan (==) nz9 Ima Problem 3.16 An UTI system has the impulse response h(n) = a"{u(n)=n(n—4)] 0 n 23 , hence the upper value of the summation is a varible n, When 72024 4 4‘ 7 w= J xhn-’= ah evs) . kat 3) kar 1 dny= an am Let k—n+3=0 w(ny= wopaae (LY T2n24LT Systems @ LTH29 ‘The lower range of (nk) has exceeded the lower range of x(). Therefore, the lower Value ofthe summation is (n— 3). The upper range of h(n — k) varies in the range 7 2 nz 4. Since the upper range of A(n—k) exceeds the upper range of (4), the upper value of the summation is 4. = When n28 yn)=0 28 0 1 w (J yOn)=} a? e 42n23 -(2) 012345678692 (0-3) n 0 nze Both x(4) and h(#-K) are not overlapped. Therefore output response is zero. Problem 3.18 An LT1 system has an iripulse response A(n) = u(n + 1) ~ u(n— 10). Determine the output of the system when the input x(n) = —u(n) + 211-3) — a(n 6). Solution + inet) PELELLLLL [- -1for23 456789011 © u(n-19) 1 -- of P23 4s 67 8 8 101 1213 4hLTH30 @ Digital Signal Processing h(n) = u(n+1}-(a-10) 1 ayes ese 7 8 8 ain) 123456789 ° ” a -- 2u(n-3) 2 r-- 1 optr2s456785 on -uln—8) 12345678 90H ° m 4 x(n) = =u(n)+2u(n=3)-u(n—6) x(k) 4 (n-9) (-k) “3-2 J ot (net)LTI Systems @ LTE31 When (n+1) <0 -l, k=0,1,2 a(ky=j+l, k=3,4,5 1 0, otherwise -k-12-11-10-9 -8 -7 -6 -5 4 -3 -2 -1 lL, kel Ho-ne{ 0, elsewhere (n-9) (net) y(n) =F a(&) nA) it Both x(4) and /(n—k) are not overlapped for (+1) <0. Therefore, the output response becomes zero. peny= 0 When 22(n+1) 2012 02-1 eo tek) wa)= ¥ x A(n-k) Ga al vn) = Yee iso 9 87 6-5 4-3 2 215 -(n+2) ino (n-8) When 52(n+1)23 = 42022 . te y(n)= x(k) inh) (nek) i A ew y(n) = Yk A(n— WY, x(k) heh) is S vn): Levens Faw K-67 SSA 123465 k (n-9) (net) vey= Yt LOD wa)= (0-4)LT132 @ Digital Signal Processing When 92(n+1)26 => 82025 5 hin-k) | ved = F x(n) - a 1 2 5 lny= D x(kyhdn— hy Y x(n— kink) i is Asaea fies 4s eta gw k 2 3 vem= > 1+ L=-34+3=0 Zone we) on When 122(n+1)210= 12029 ‘ (nt) v= Date in-4) x(n) 1 2 ‘ pny= Yo x(k) bird) + Yi x(k) nk) kata) bey wk2-1 [0123465 67 6 9 01 12 2 5 wa= > C4 YO=-241 =(n-9]43 betes) yn) =(n-9) in-9) (nen) When 152 (a+1)2 13 = 1420212 hin=k) Besse 7 8 9 101 12 18 4 (o-3) (ns) 3 y= Da Mr-&) fore) yaya S. We-mtts 9)LT Systems @ LT/33 When (r+ 1)2 16=5 #215 28 4 6 6 7 8 9 10 1f 1213 14 18 16 (a) (1) wa=o There is no overlapping of x(&) with h(n-k), Therefore, output response is zero. 0 ne-l Ant+2) -1Snsl n+4 2sns4 wn =40 Ssns9 n-9 — 10SnS11 8-9 12Sns14 0 n>ia @ 3.7. STEP RESPONSE By using the convolution sum, we can easily represent the step response in terms of the impulse response. Let us consider ihe output response of the system (71) as, Y= a(n) *hKn)= Yay A) = J) H(R)2(n—k) . G27) kee ime where, x(n) i input signal and f(a) is impulse response of the system. °= =~ The step response of the systemy means. applying a unit step function as a signal io the system. that is, a(n) un) 1, nzk - 1, neo where un) -{ otherwise “("~*) -{r otherwise Then, step response S(n)= J h(k)u(n—k) im St) = FHA) (3.28) ist Equation (3.28) explains that the step response is the impulse response.LTL34 © Digital Signal Processing SOLVED PROBLEMS Problent 3.19 Find the step response of the system if the impulse response is h(n) = at"u(n),0_> =) Problem 3.21 What is the input signal x(n) that will generate the output sequence y(n) ={1,5,10,11,8,4,1} t for a system with impulse response h(n) = {1,2,1} (A.U.April 2003) Solution T Yn) ={1,5,10,11,8,4,1}; h(n) = {2} tT The total number of samples in the output response is N,+ N,-1=7. ‘The number of samples in the impulse response is V,=3. ‘The number of samples in the input signal is W,=7~ N,+1=5.LT36 @ Digital Signal Processing Let us consider the general deconvolution equation eI HY e(m\in— m) = x= FO) = 22) Ll For n=0, 0-49 “TT 1)—a(O)A(9) _ S12 For wt, = ADEA Sas For n=2, (2) = 2A=AOMA)— ADA. _ 10132 _ WO) For n=3, (3) = 2D MOE) aCVH(2P— DVM) _ = 103x134? _ 9 (0) D For n=d, aay= 2D HOMO ADE) “AEN 7 HED = 10 ~3x0-3x1 = 2x2» | 1 ‘The input sequence is.x(m) = (1,3, 3, 2, 1}. m@ 3.9 BASIC SYSTEMS The system can be broadly divided into two categories. They are 1. Finite impulse response system 2. Infinite impulse response system Finite impulse response (FIR) system is one which exhibit zero response outside a finite time interval. Example hiny 1, [ols 0, otherwise ow WN a + Finite duration ———-4| A(n) = Fig. 3.10LT! Systems @ LTI37 Infinite impulse response (IIR) system is one which exhibit an impulse response of infinite duration. Example ny {eonse (0, otherwise Trend te Infinite [Infinite duration Fig. 3.11 @ 3.10 LINEAR CONSTANT COEFFICIENT DIFFERENCE EQUATION ‘The difference equation generally represents the relationship between the input and output signals for a system. The Nth order difference equation is wv i Baver- = Paste (3.33) ‘The equation (3.33) can be rewritten as y a y(n) + Dag (nk) =D) by x(n—K) a & is 1 y= Ya x(n-O-+ Yavin) (3.34) % =o Fy ket where a, = |, isa constant. ‘The output of the system can be obtained from the input and past output signal. This is clear in equation (3.34). Hence, for most discrete-time system implementations, we use equation (3.34). 3.10.1 Solution to Linear Constant Coefficient Difference Equation Itis convenient to express the output of the system described by a difference equation as a sum of two responses: i. Natural Response (9, 2. Forced Response ()!) 3.40.1.1 Natural response ‘The natural response of the system output can be obtained by considering the initial conditions. While calculating the natural response of the system output, the input is made zero. Therefore, the difference equation (3.33) is reduced to homogeneous equation, which is given by x Yay n-*)=0 (3.35) &LT138 @ Digital Signal Processing where superscript “(1)’ represent the natural response. The natural response /"a) is the solution to the homogeneous equation and is given by : (n= a" (3.36) where 4 is M roots of the discrete-time system. Substituting equation (3.36) in (3.35), we get (3.37) ta A" + aA" + a,a"> 0 NAN ba Ah ta AN 2 4. tay) =O Since 2°-" 20 BN a AN" 4a,aN2 4. ay =0 (3.38) Equation (3.38) is the characteristic equation of Mth root. Case (i) Ifroots 4,,4;,A,,... are distinct. The general solution to the homogeneous equation whose roots are distinct is IMC AT +CyhZ +O, +. CyAt, (3.39) Example \froots are distinct, say A, =4 and A, =6, then according to equation (3.39), the solution to the homogeneous equation is y,(n)=C,(4)" +C2(6)" Case (ii) Ifroots are repetitive. ‘The general solution to the homogeneous equation whose roots are repetitive m times is PAM = ANC ACNE Cyt HCP HC) (3.40) Example \f A, =8 repeated 2 times and A, = 4, then according to equation (3.40), the solution to homogeneous equation is Yl) = (C, +Cpm)(B)" +C,(4)" Case (iii) Ifroots are complex. ‘The general solution to homogencous equation whose roots are complex, ie. 2, = a+ jband Ay =a— jd alr) = 7" (A, C05 8+ 4, sind) (3.41) where r= va?+5? vm A, 4, = constant coefficientsLTE40 @ Digital Signal Processing : Let n= =p yy tye HO) = EINE) 512 HORE 63 5 1 WD = 390) ~Z-0) $(2) 1_7 so=4(2) 6 18 Substitute the value of y(0) and y(1) in the equations (4) and (5) respectively, we get menses i 2 1.7 Ct 2Q == my ‘On solving, we obtain Cn he,--13 Substituting the values of C, and C, in equation (3) vin 9 which is the natural response of the given difference equation. 31012 Forced response The forced response of the system output can be obtained by considering the input signal alone (initial conditions are assumed to be zero). The forced response consists of two parts, (ly Homogeneous solution Ly.(™)] Q) Particular solution L,(7)] ‘The particular solution is denoted by y,(n) and it represents the solutions to any difference equation when the input is given. The particular solution y (1) can be obtained by considering the input signal x(n), n 20, ie. w. w dp) =—Yayypln—k)+ DY byxtn—) (3.42) & a The following table shows particular solutions for some of the standard signals.LTI Systems @ LTH Table 3.1 Standard Particular Solution Tnput Signal Particular Solution x) Jel) Step signal, 4 K Ac" Ka" Ant gn hyn 4 hyn? tt hy A castor 9) cosa ky sin on Asin(eon+$) : The forced solution of the system is obtained by adding both the homogeneous and particular solutions. SOLVED PROBLEMS... Problem 3.23 Find the forced response of the system described by the difference equation who) =Eyln-D= Ln 2) + a(n) (0-1) when the input signal is x(n) =2"u(n). Solution y(n) n)—2 ned yn) ven) Z Her Dt E72) wln—l—2 vn) x(n)—a(n—1) =x(n)— (4-1) For input signal x(n) =2"u(n). the particular solution from the table 3.1 is of the form Vplat) = K2" Substituting equation (2) in equation (1), xo Sota t 6 6 Ky tage Obtain the value of *K* of equation (3) by substituting any valuc of ‘n’. Let 9, Spotgh poe K2'->K2'+2K2 6 6 K= ‘Therefore, the particular solution is given by Ip 4 5 a) (2) Q) (4)LTH42 @ Digital Signal Processing The homogeneous solution (obtained from equation 3.36) is wi-4 yn pst y(n—2) =0 [inputsignal, x(7)=0] The solution to the homogeneous equation is =a" Therefore, Since 2" #0 Since the roots are distinct, we use equation (3.39) ety niernci(3)} +<5(3) YPN) = y+ 9, (M) a o yoor-le(3) +a(3) jee (5) ‘The forced response is given by Letn=0, YPO=C, ro+t (6) n=l, a Let us consider equation (1). y(n) yun p+ yn-2) = x(m)-x(n-1) 6 6 1 Forn=0, w10)-2 9-042 6-2) = x(0)- x=) ro) Since initia! conditions are assumed to be zero for calculating the forced response, y{—1 xl) =0. Consider the input signal, x(n) = 2" Forn= 0. x(0)= 1 ‘Therefore, equation (8) can be written as: : O)=ILTI Systems @ LTM43 Similarly, for m= | in equation(1), 5 1 )-2 (0) +=, 1)-3(0) w on y+ et )—x(0) s)-2ey40=2-1 [Since x(1) = 2] u wae Substituting the values of y(0) and y(1) in equations (6) and (7) * 4 Cats 4,0 38. 2°35 6 GehG= Substituting the values of C, and C’, in equation (5), WMeny={ LY 42) 4 So yn) (3) (3) +22 n20 3.10.13 Complete response ‘The complete response of the system can be obtained by adding both natural response and forced response. Wen) = y(n) y(n) (3.43) The complete response can be obtained directly without separately finding the natural response and forced response as we did in Problems (3.18) and (3.19). SOLVED PROBLEMS. _—§_ 3» = Problem 3.24 Find the complete response of the system described by the difference equation when the input signal x(n) = 2" n(n). The initial conditions are y(—1) =1 and y(—2)=1. ‘Solution wn=1)—Ly(n=2)4 x)= x(n=1) 6 a) The complete response af the system wr) = (n+ YP (nyLTL44 © Digital Signal Processing To obtain natural response set input signal to zero, ic. Hed =Z oD =Z yl 2}=0 The solution to the homogeneous equation is of the form daly = a” Therefore, equation (2) becomes The solution to the homogeneous equation is ml=C, () ea(5) To obtain the forced response, consider the input signal x(n), ic Home yhn=2) = x(n) —¢e— 1) The solution to the particular equation is of the form [from Table (3.1)] Yp()= K2" Therefore, equation (4) becomes Kar Sorta h grt aot art 6 6 Obtain the value of” of equation (5) by substituting any value of ‘n’, Leta=0, xo Seats lata 6 6 kat Therefore, the solution for the pasticular equation becomes Soe Yplm)= 52 2) (3) (4) (5) (6)LTi Systems @ LTH45 ‘The complete response is sum of the homogeneous solution and particular solution, Hn) = yal 04 (0) a4, 2Y 6, fl! vn=gatec(3) +3] Forn=0, Hoa Sci 4cy 8 1 I Forn= 1, vo=£ec(t}+04) To obtain the value of y(0) and y(1), let us reconsider the difference equation (1). Hla) = pln) ylnm 2p a(n) a(0=1) anna, s(0)=1 x{1)=2; x(-1)=0 (not given) Forn=0, 1O=2 5(-D-Ly(-2)+0)—1-D ‘Now substituting initial conditionsy(-1)= 1 and y(-2)= 1 5 ¥O)= j -40)+1-0 3 (0) == 2O=5 5 Forn=1, w)=2 0) ~ 231-4 x(1)~H10) Now substituting i ial conditions {-1) = 1 and y(—-2)=1 5(5)_1 ay=2(3)-2 (42-1 ¥) (3) got 20 y= Substituting the values of )(0) and y(1) in equations (8) and (9) respectively 17 . 15 Therefore, the complete response of the system is given by pando) -BC On solving, G22 G=- 3 y 2 (8) oLT-48 @ Digital Signal Processing Problem 3.25. Find the system response described by a difference equation 1 ya) yn = x0) when the input signal x(n) =sin2n Initial condition »(-1) = 1. Solution The difference equation is 1 YA—ZVA—N= x(n) ‘The complete response of the system is vn) = yale) + Yp(r) - ‘To find the solution to the homogeneous equation, set the input signal x(n) = 0. wa-$ 0-0 =0 ‘The solution to the homogeneous equation is a(n) =A" The homogeneous equation becomes “Therefore, the solution of the homogeneous equation is (n= (ty Yn 3 ‘To obtain the forced solution let us consider the input signal, x(1n) = sin2n ya) ye D= xn) The solution to the particular equation is p(t) = Kj cos 2n+ Ky sin2n (K,cos2n+ Ky sin2n)~ AEA; cos 2(n=1)+K; sin2(n=1)] sin 2n Hing ©05(4= B)=c0s AcosB+sin Asin B sin(A—B)=sin Acos B—cos Asin B» LTiSystems @ LTT K,cos2n+ K, sin In— 31K; (cos2n cos2+sin2n sin2) + K,(sin2n cos2—cos 2n sin 2)]=sin2n K,cos2n+ Ky sin penn cos? “ in2n sin2— 2 sinan cos2 +Zeos2n sin2=sin2n On comparing left hand side and right hand Fide ers, Ry cos2n—“L cos? cos2n+A2cos2n sin2=0 cos2n( ~ eos sina) = K\~“Leos2-+-S2sin? = 0 (1-22) + (222) 0 1.208X, + 0.4547K, =0 q) Similarly, Ky sin2n~ Asin 2 sin2n~“2c0s2 sin 2n=sin2n Ky ~“bsin2- “eos? =1 « {22} ,(\-2)- 1 = OASAT K, +1.208 K, =1 @ On solving equations (1) and (2), &,=-02729; K,=0.725 ‘Therefore, the solution to the particular equation is . Ypltt) =-0.2729 cos2n+0.725 sin2n 6) “The complete response of the system is Wad=y, Cty, (a) xec(z] -0.2729c0s2n+0.725 sin 2n Forn=0, 0)=C-0.2729 (4)LTL48.@ Digital Signal Processing To find the value of (0), let us consider the difference equation vta)~$ v(0—D=x(0) where x(n) =sin2n » Forn=0, x(0)= (-1)+x(0) © Substitute the value of y(0) from equation (5) to (4) aC-02ns or C=0.7729 The complete response of the system is given by vem=o.rra9{£) ~0.2729¢0s2n + 0,725sin2n Problem 3.26 Find the system response described by a difference equation plo) xn Des. ig n—2)=2 forn20 The initial conditions are (-1)=2 andst 2)= Solution Given that JO 75. Zon bee pre 2)=2 a To obtain the natural response, set ted input sien to zer0, ‘The homogeneous equation becomes ma tr ves y(n—2)=2 (2) The solution to the homogeneous equation is Yale) =a" (3) The homogeneous equation becomesLTtSystems @ LTI-49 Since °°? #0: Since the roots are distinct. the solution to the homogeneaus equation is flo. iy yM=Ch =} Hh p= SOG “) To obtain the forced response, let us consider the input signal, which is a step signal who~c amplitude is 2, that is, x/n)=2um). The particular equation becomes y(n). sy Da gla 2Y=2Wny (3) “The solution tote particular solution [from Table (3.1)] becomes ¥p(m= Ku(a) (6) Substituting equation (6) in equation (5) Kn) - ZK utn—Dpeck K n—2)= 200) Therefore, yp) =4 un) ‘The complete solution of the system becomes wr) = yp) + yp) xny=cy (3) +e(t +4 wi) Forn=0, MO~Cr4Cs+4 a Forn=1, yes boat Cy +4 @) In order to find the valvesofx0) and 1), let us consider the difference equatio 7 1 vn) aype- )- oe Forn=0, IO=T. Zo Y)— pe 2)+2 yO) = 329167LTES0 © Digital Signal Processing Forn=1 y= y(0)- X-1) +2, one “Ta 2” =1(35)_1oayi22 v= Z(3} yg +2 = 3.5347 Substitute the values of y(0) and ){1) in equations (7) and (8), 29167=C +044 ‘We obtain, C= 2.3337; C,= 1.2504 ‘Therefore, the complete response of the system is given by xire[aain(3) s12soe(5) whan @ «3.11 ‘INTRODUCTION TO CORRELATION ‘The correlation is another mathematical operation to measure the degree of similarity of any two signals/ images. Correlation is used in RADAR, digital communication, remote sensing engineering, etc. Letus explain correlation with respect to the following example. The signal sequences.x(n) and y(n) are the transmitted and received signals respectively. TARGET Fig. 3.12 Target Detection Using RADARLTI Systems @ LTb51 If the target is present in the space during RADAR search, then the received signal y(n) is the delayed input signal x(n~D), i.e. Ma) = ax(n~ D) + w(n) where a= attenuation of signal x(n) w(n) = additive hoise pick up along with the signal at RADAR D= delay factor The delay is directly proportional to the distance between the RADAR and the target. In practice, the received signal x(n ~D) is heavily corrupted by the additive noise to the point where a visual inspection of (n) does not reveal the presence or absence of the desired signal. Correlation helps us to extract thi information from y(n). 3.11.1 Cross-correlation Letus consider two different signal sequences x(7) and y(7) which has finite energy. The cross-correlation of x(n) and y(1) is given by, Yay () = > x(n) y(n m), m= 0, £1, 42,43, ... (B44) or equivalently Yq(m)= 5, x04m) y(n), m=0,41,42,43, .. G45) where, m=lag parameter The subscript parameter ‘xy’ in 7, indicate the direction in which the sequence is shifted by m. In equation (3.44), the signal x(n) is unshifted while y(n) is shifted by *m" units to the right (m is positive). In equation (3.45) the signal y(n) is unshifted while x(n) is shifted by ‘ms’ units to the left (m is negative). Both equations (3.44) and (3.45) are identical, that is, both relations yields identical cross-correlation sequence. Reversing the roll of x(7) and (12) in equations (3.44) and (3.45) result in equations (3.46) and (3.47) respectively. Yoalmy= Y) yln)x(- Mm), m= 0,1,22,43, .. G46) ‘or equivalently Ap lond = By sr)ylor+ Mm), m= 0,41 22,33, Gan On comparing equation (3.44) with (3.45) or (3.46) with (3.47), we conelude that Yay = Ype(—m) (3.48) Itis clear from equation (3.48) that 7, isthe folded version of 7,,. If the length of the sequence x(7) is N, and the length of the sequence y(n) is N, then total length of correlation sequence is N,+, =.LTb52_@ Digital Signal Processing ‘The major computational difference between convolution and correlation is that in case of convolution, one of the sequence is folded, then shifted, then multiplied by the other sequence to form the product sequence for that shift and the product terms are added. In case of correlation, except folding all other process remain the same, that is, one of the sequence is shifted, then multiplied by the other sequence to form the product sequence for that shift and the product terms ate added, Mathematically, the correlation and convolution can be related as Y= x(mh4 (m1) . G49) SOLVED PROBLEM Problem 3.27 Determine the cross-correlation sequence of the sequences x(n) = {1,2,3,4,5}; y(n) = 15,6, 7,8,9} tT tT Solution Let us consider equation (3.44) Yylm = LY x(n) yur-m) For the given problem equation (1) reduces to Yu (d= Dx y= my When m= 0 Y(0)= x00) 0) = For m= 0, the cross-correlation is the product of x(v) and y(n) and sum of all the products, that is, Tol (-2) y(-2) +.4(—1) (1) + 20) (0) 4-x(1) (1) +-2(2) (2) Yq (OV=1XS +26 +3X7 +4x8 +59 S115 When m= | 10 Yop = (2) ol-3) + (1) 9-2) ¥(0) y(—1) + x(1) 90) + (2) 9) Yq ()=OF2X543X644x745x8 +0=96 When m=2 y= Dx) 9-2) Vy (2)= x(-2) (A+ XC) 3-3) + (0) 2) + xD) (1) + 12) (0) Vy (2)= 0+ 0435446457 4040574LTI Systems @ LTLS3 When m= 3 ¥_8)= Zxeove- -3) ¥_0) 2) y(-5)+(-1) 1-4) +-1(0) cayeanye 2)+x(2) (1) Yq B)=0+04044%545x6+040=50 Whenm=4 ,4)= x x(n) (n-4) Y= atc 2) f-6)4-D) 9 5)+3(0) (4) +x(0) (-3)4 (2) 1-2) ¥,,(4)=04+.040+0+5x5+0=25 When m=5 A ¥oy(5)= Y x(n) yln—5) a oy (S)= (2) H(—T) + X(N) 8) + (0) HS) + (1) (A) + 202) YC-3) Yul Yq (m25)= When m =-1 Yo(-D= Ye) ynel) 4 Yop(“1=X(—2) y(—D+ x(-1) 90) + (0) yl) +3(1) y(2)+ 22). 3) Yo(-W=l X6+24743x8-+4%9+ 0=80 When m = -2 (2) yn +2) Ty (2 (=2) y(0) + 3(-1) (1+ ¥(0) y(2) +000) (3) +212) 14) Yo(-2)=1 X7+2x8+3%9+ 04 0=S0 When m =~3 Yag6-3) = Dx) 9243) Yay (3) = ¥(-2) VHC) A2)+ (0) 3) + 2() E44 DO) Yy(3) = 1xB+2x940404 0=26 When m= Yn 4) = 2 x(n) y(n 4) To) = -2) yey x 1) y(3) + x(0) yC4}+ (1) 65) + 42) 76) Yy(A)=1X9+040404 0=9LT-54 @ Digital Signal Processing When m=-5 14-9) Yu (5) =O ¥,,(m$-5)=0 ‘Therefore, the cross-correlation sequence of x(n) and y{i) is Yay (om) ={9,26, $0, 80, | 15,96, 74,50,25} 2) y(B)+ x(=1) (4) + x(0) (5) + x(1) 9(6) + 3(2) (7) 3.11.2 Autocorrelation If») = x(r2) then equation (3.43) and (3.46) reduces to rath ¥ a(n)x(n-d (3.50) or equivalently Pally= ¥ xn xer (3.51) Equations (3.50) and (3.51) are the autocorrelation equations of the sequence. 3.113 Properties of Cross-correlation and Autocorrelation Let us consider two data sequences x(n) and y(n) whose linear combination is given by An) = Aa(n) + BYn—f) Where, A and Bare scalar 1s the shift The energy of the sequence 2(n) is given by £2 20 : E= 3 [Axle + By(n—DP Onsimplification, E=4 Sy eew E yn D424B > x(n) vn=1) -(0)+B*y,, (0)+2ABY,, (2) Ifthe energy of signals x(n) and y() are finite, then the energy of Z(n) must also be finite, that is, EA Ye (O)+ BY y (OVE ZABY sp N20LTI Systems @ LTS On simplification, by inequality l.W|sf7O+Y,,0) (Proofis leftto the reader) Ifx(n) = Xn), then . bls V7.) =E, ‘The autocorrelation of sequence attains maximum energy condition at zero lag (that is, /= 0). SOLVED PROBLEM’) => -§ —§_-§ > Problema 3.28 Find the cross-correlation of two finite length sequences x(n) = {1,2,3,4} and (2) = {5,6,7,8}. Solution x) =41,2,3.4}; 1-D= {8.7.6.5} By definition, 7,(=2()* (-/) Yop (D={8,16 +7, 2441446, 32421412 +5,28+18 +10, 24 +15, 20} Yup (C)={8,23,44, 70, 56,39, 20} (CHAPTER SUMMARY © Any signal can be represented as a time-shifted impulse sequence. * The convolution gives the relation between input signal, impulse response and output response, It also explains how output response is obtained from input signal when it is passed through a system impulse response. The convolution sum is denoted by »() = > x(k)h(n—k) = x(n) h(n). im © The LT! system satisfies basic three properties, thats, distributive, associative and commutative properties. © Distributive property: x(n) [04 (n)+ h(n] = x(n) y(n) + 2(0) # Ay)LTL5S © Digital Signal Processing Associative property: [x(r2)*/4()}* hp (m2) = x(n)* [Ay () * A, (2) Commutative property: [x(7) * A(n)] = [A(7) * x(n] © The procedure for linear convolution is as follows: Penne ay ae 2 Plot both x(k) and A(x). Reflect (k) about g = 9 to obtain h(-k). Shift h{-K) by n(toward left). Let the initial value of n be negative. Multiply cach element of x(k) with A(n—A) and add all the product terms to obtain y(n) - Shift h(n - &) by incrementing the value of m by one and so step 5. Do step 6 until the product of x(k)and A(n—k) reduces to zero. The solution to the difference equation is analyzed by considering natural response and forced response. The natural response can be obtained by considering the initial values alone. The input signal to the system is agsumed to be zero. The natural response can be obtained by solving homogeneous equation. The forced response can be obtained by considering the input signal alone. The initial values of the system are assumed to be zero. The forced response can be obtained by solving homogeneous equation and particular equation. The correlation is another mathematical operation to measure the degree to which the signals are simitar. The autocorrelation refers tothe correlation of same signal where ascross-correlation refers tothe correlation of two different signals, ‘The cross-correlation of two different signal sequences.x(n) and y(n) which has finite energy is given by Py = Deyn), m= 0,21,22,.. The autocorrelation of a signal sequence x(n)is given by Yux(!)= > xr)x(n—2), 1=0,41,42,... REVIEW QUESTIONS _— Define convolution sum, Derive an expression for convolution sum. Explain the properties of convolution sum. Express the following signals in terms of impulse function. (a) xfa)={0,2,4,8,10} (b) x0n)=14 7,9, 12} Define invertibility of LTI system. Define stability in LTI system.