You are on page 1of 6

Digital signal processing

See also: Digital signal processor

1 Signal sampling

Digital signal processing (DSP) is any signal processing

conducted on analog signals by digital means (as opposed
to Analog signal processing where the signal processing
is is carried out by an analog process). It consist in the
mathematical manipulation of an information signal to
modify or improve it in some way. It is characterized by
the representation of discrete time, discrete frequency, or
other discrete domain signals by a sequence of numbers
or symbols and the processing of these signals.

Main article: Sampling (signal processing)

Digital signal processing can involve linear or nonlinear

operations. Nonlinear signal processing is closely related
to nonlinear system identication[3] and can be implemented in the time, frequency, and spatio-temporal domains.

choose the domain in which to process a signal by making an informed assumption (or by trying dierent possibilities) as to which domain best represents the essential characteristics of the signal. A sequence of samples
from a measuring device produces a temporal or spatial
domain representation, whereas a discrete Fourier transform produces the frequency domain information, that is,

The increasing use of computers has resulted in the increased use of, and need for, digital signal processing. To
digitally analyze and manipulate an analog signal, it must
be digitized with an analog-to-digital converter. Sampling is usually carried out in two stages, discretization
and quantization. In the discretization stage, the space of
signals is partitioned into equivalence classes and quanThe goal of DSP is usually to measure, lter and/or com- tization is carried out by replacing the signal with reppress continuous real-world analog signals. Usually, the resentative signal of the corresponding equivalence class.
rst step is conversion of the signal from an analog to a In the quantization stage, the representative signal values
digital form, by sampling and then digitizing it using an are approximated by values from a nite set.
analog-to-digital converter (ADC), which turns the ana- The NyquistShannon sampling theorem states that a siglog signal into a stream of discrete digital values. Often, nal can be exactly reconstructed from its samples if the
however, the required output signal is also analog, which sampling frequency is greater than twice the highest frerequires a digital-to-analog converter (DAC). Even if this quency of the signal, but this requires an innite number
process is more complex than analog processing and has of samples. In practice, the sampling frequency is often
a discrete value range, the application of computational signicantly higher than twice that required by the sigpower to signal processing allows for many advantages nals limited bandwidth.
over analog processing in many applications, such as error
detection and correction in transmission as well as data Some (continuous-time) periodic signals become nonperiodic after sampling, and some non-periodic signals
become periodic after sampling. In general, for a periDigital signal processing and analog signal processing are odic signal with period T to be periodic (with period N)
subelds of signal processing. DSP applications include after sampling with sampling interval Ts, the following
audio and speech signal processing, sonar and radar sig- must be satised:
nal processing, sensor array processing, spectral estimation, statistical signal processing, digital image processing, signal processing for communications, control of systems, biomedical signal processing, seismic data process- Ts N = T k
ing, among others. DSP algorithms have long been run
where k is an integer.[4]
on standard computers, as well as on specialized processors called digital signal processors, and on purposebuilt hardware such as application-specic integrated circuit (ASICs). Currently, there are additional technologies 2 DSP domains
used for digital signal processing including more powerful general purpose microprocessors, eld-programmable In DSP, engineers usually study digital signals in
gate arrays (FPGAs), digital signal controllers (mostly for one of the following domains: time domain (oneindustrial applications such as motor control), and stream dimensional signals), spatial domain (multidimensional
processors, among others.[2]
signals), frequency domain, and wavelet domains. They


the frequency spectrum. Autocorrelation is dened as the 2.2 Frequency domain

cross-correlation of the signal with itself over varying intervals of time or space.
Main article: Frequency domain


Time and space domains

Signals are converted from time or space domain to the

frequency domain usually through the Fourier transform.
The Fourier transform converts the signal information to
Main article: Time domain
a magnitude and phase component of each frequency.
Often the Fourier transform is converted to the power
The most common processing approach in the time or
spectrum, which is the magnitude of each frequency comspace domain is enhancement of the input signal through
ponent squared.
a method called ltering. Digital ltering generally consists of some linear transformation of a number of sur- The most common purpose for analysis of signals in the
rounding samples around the current sample of the input frequency domain is analysis of signal properties. The
or output signal. There are various ways to characterize engineer can study the spectrum to determine which frequencies are present in the input signal and which are
lters; for example:
A linear lter is a linear transformation of input
samples; other lters are non-linear. Linear lters
satisfy the superposition condition, i.e. if an input
is a weighted linear combination of dierent signals,
the output is an equally weighted linear combination
of the corresponding output signals.

In addition to frequency information, phase information

is often needed. This can be obtained from the Fourier
transform. With some applications, how the phase varies
with frequency can be a signicant consideration.

Filtering, particularly in non-realtime work can also be

achieved by converting to the frequency domain, applying the lter and then converting back to the time domain.
A causal lter uses only previous samples of the This is a fast, O(n log n) operation, and can give esseninput or output signals; while a non-causal lter tially any lter shape including excellent approximations
uses future input samples. A non-causal lter can to brickwall lters.
usually be changed into a causal lter by adding a There are some commonly used frequency domain transdelay to it.
formations. For example, the cepstrum converts a signal
to the frequency domain through Fourier transform, takes
A time-invariant lter has constant properties over the logarithm, then applies another Fourier transform.
time; other lters such as adaptive lters change in This emphasizes the harmonic structure of the original
Frequency domain analysis is also called spectrum- or
A stable lter produces an output that converges spectral analysis.
to a constant value with time, or remains bounded
within a nite interval. An unstable lter can
produce an output that grows without bounds, with
bounded or even zero input.
A nite impulse response (FIR) lter uses only the 2.3 Z-plane analysis
input signals, while an innite impulse response
lter (IIR) uses both the input signal and previous
Main article: Z-transform
samples of the output signal. FIR lters are always
stable, while IIR lters may be unstable.
Whereas analog lters are usually analyzed in terms of
transfer functions in the s plane using Laplace transforms,
A lter can be represented by a block diagram, which can digital lters are analyzed in the z plane in terms of Zthen be used to derive a sample processing algorithm to transforms. A digital lter may be described in the z
implement the lter with hardware instructions. A lter plane by its characteristic collection of zeroes and poles.
may also be described as a dierence equation, a collec- The z plane provides a means for mapping digital fretion of zeroes and poles or, if it is an FIR lter, an impulse quency (samples/second) to real and imaginary z comresponse or step response.
ponents, where z = rej for continuous periodic signals
The output of a linear digital lter to any given input and = 2F ( F is the digital frequency). This is useful
may be calculated by convolving the input signal with the for providing a visualization of the frequency response of
impulse response.
a digital system or signal.



4 Implementation

Main article: Discrete wavelet transform

Depending on the requirements of the application, digital
In numerical analysis and functional analysis, a discrete signal processing tasks can be implemented on general
purpose computers.
Often when the processing requirement is not real-time,
processing is economically done with an existing generalpurpose computer and the signal data (either input or output) exists in data les. This is essentially no dierent
from any other data processing, except DSP mathematical techniques (such as the FFT) are used, and the sampled data is usually assumed to be uniformly sampled
in time or space. For example: processing digital photographs with software such as Photoshop.

An example of the 2D discrete wavelet transform that is used

in JPEG2000. The original image is high-pass ltered, yielding
the three large images, each describing local changes in brightness (details) in the original image. It is then low-pass ltered
and downscaled, yielding an approximation image; this image is
high-pass ltered to produce the three smaller detail images, and
low-pass ltered to produce the nal approximation image in the

However, when the application requirement is real-time,

DSP is often implemented using specialized microprocessors such as the DSP56000, the TMS320, or the
SHARC. These often process data using xed-point
arithmetic, though some more powerful versions use
oating point. For faster applications FPGAs[5] might be
used. Beginning in 2007, multicore implementations of
DSPs have started to emerge from companies including
Freescale and Stream Processors, Inc. For faster applications with vast usage, ASICs might be designed specifically. For slow applications, a traditional slower processor such as a microcontroller may be adequate. Also
a growing number of DSP applications are now being
implemented on embedded systems using powerful PCs
with multi-core processors.

5 Techniques
wavelet transform (DWT) is any wavelet transform for
which the wavelets are discretely sampled. As with other
wavelet transforms, a key advantage it has over Fourier
transforms is temporal resolution: it captures both frequency and location information (location in time).

Bilinear transform
Discrete Fourier transform
Discrete-time Fourier transform
Filter design


LTI system theory

Minimum phase

The main applications of DSP are audio signal processing, audio compression, digital image processing,
video compression, speech processing, speech recognition, digital communications, radar, sonar, Financial
signal processing, seismology and biomedicine. Specic examples are speech compression and transmission in digital mobile phones, room correction of sound
in hi- and sound reinforcement applications, weather
forecasting, economic forecasting, seismic data processing, analysis and control of industrial processes,
medical imaging such as CAT scans and MRI, MP3 compression, computer graphics, image manipulation, hi-
loudspeaker crossovers and equalization, and audio effects for use with electric guitar ampliers.

Transfer function
Goertzel algorithm

6 Related elds
Analog signal processing
Automatic control

Computer Engineering
Computer Science
Data compression
Dataow programming
Electrical engineering
Fourier Analysis
Information theory
Machine Learning
Real-time computing
Stream processing
Time series

Sen M. Kuo, Woon-Seng Gan: Digital Signal Processors: Architectures, Implementations, and Applications, Prentice Hall, ISBN 0-13-035214-4
Bernard Mulgrew, Peter Grant, John Thompson:
Digital Signal Processing - Concepts and Applications, Palgrave Macmillan, ISBN 0-333-96356-3
Steven W. Smith (2002). Digital Signal Processing: A Practical Guide for Engineers and Scientists.
Newnes. ISBN 0-7506-7444-X.
Paul A. Lynn, Wolfgang Fuerst: Introductory Digital
Signal Processing with Computer Applications, John
Wiley & Sons, ISBN 0-471-97984-8
James D. Broesch: Digital Signal Processing Demystied, Newnes, ISBN 1-878707-16-7
John G. Proakis, Dimitris Manolakis: Digital Signal Processing: Principles, Algorithms and Applications, 4th ed, Pearson, April 2006, ISBN 9780131873742
Hari Krishna Garg: Digital Signal Processing Algorithms, CRC Press, ISBN 0-8493-7178-3


[1] James D. Broesch, Dag Stranneby and William Walker.

Digital Signal Processing: Instant access. ButterworthHeinemann. p. 3.

P. Gaydecki: Foundations Of Digital Signal Processing: Theory, Algorithms And Hardware Design,
Institution of Electrical Engineers, ISBN 0-85296431-5

[2] Dag Stranneby and William Walker (2004). Digital Signal

Processing and Applications (2nd ed. ed.). Elsevier. ISBN

Paul M. Embree, Damon Danieli: C++ Algorithms

for Digital Signal Processing, Prentice Hall, ISBN 013-179144-3

[3] Billings S.A. Nonlinear System Identication: NARMAX Methods in the Time, Frequency, and SpatioTemporal Domains. Wiley, 2013

Vijay Madisetti, Douglas B. Williams: The Digital

Signal Processing Handbook, CRC Press, ISBN 08493-8572-5

[4] Oppenheim, Alan V.; Schafer, Ronald W.; Buck, John R.,
Discrete-Time Signal Processing (3rd ed.), p. 15
[5] JpFix. FPGA-Based Image Processing Accelerator.
Retrieved 2008-05-10.

Stergios Stergiopoulos: Advanced Signal Processing

Handbook: Theory and Implementation for Radar,
Sonar, and Medical Imaging Real-Time Systems,
CRC Press, ISBN 0-8493-3691-0
Joyce Van De Vegte: Fundamentals of Digital Signal
Processing, Prentice Hall, ISBN 0-13-016077-6

Further reading
Alan V. Oppenheim, Ronald W. Schafer, John R.
Buck : Discrete-Time Signal Processing, Prentice
Hall, ISBN 0-13-754920-2
Boaz Porat: A Course in Digital Signal Processing,
Wiley, ISBN 0-471-14961-6

Ashfaq Khan: Digital Signal Processing Fundamentals, Charles River Media, ISBN 1-58450-281-9
Jonathan M. Blackledge, Martin Turner: Digital Signal Processing: Mathematical and Computational
Methods, Software Development and Applications,
Horwood Publishing, ISBN 1-898563-48-9

Richard G. Lyons: Understanding Digital Signal

Processing, Prentice Hall, ISBN 0-13-108989-7

Doug Smith: Digital Signal Processing Technology:

Essentials of the Communications Revolution, American Radio Relay League, ISBN 0-87259-819-5

Jonathan Yaakov Stein, Digital Signal Processing, a

Computer Science Perspective, Wiley, ISBN 0-47129546-9

Charles A. Schuler: Digital Signal Processing: A

Hands-On Approach, McGraw-Hill, ISBN 0-07829744-3

James H. McClellan, Ronald W. Schafer, Mark A.
Yoder: Signal Processing First, Prentice Hall, ISBN
John G. Proakis: A Self-Study Guide for Digital Signal Processing, Prentice Hall, ISBN 0-13-143239-7
N. Ahmed and K.R. Rao (1975). Orthogonal
Transforms for Digital Signal Processing. SpringerVerlag (Berlin Heidelberg New York), ISBN 3540-06556-3.


Text and image sources, contributors, and licenses



Digital signal processing Source: Contributors: AxelBoldt, Tobias Hoevekamp, Brion VIBBER, The Anome, Tbackstr, Ap, HelgeStenstrom, Wayne Hardman, Rade Kutil, Ben-Zin, Boleslav
Bobcik, Olivier, Patrick, Yaroslavvb, Michael Hardy, Nixdorf, Kku, Easterbrook, MightCould, Loisel, Mac, Ronz, Glenn, Smack, Ww,
Dysprosia, Omeomi, Furrykef, Hyacinth, Greglocock, Omegatron, Phoebe, Ldo, Finlay McWalter, Jondel, Lzur, Giftlite, Wolfkeeper,
Ds13, Mcapdevila, SWAdair, Pgan002, Bact, Togo, Ukexpat, DJS, Abdull, Moxfyre, Discospinster, Style, Y(J)S, Martpol, Violetriga, Alberto Orlandini, .:Ajvol:., Rbj, Matt Britt, Hooperbloob, Vizcarra, Ryanmcdaniel, Atlant, Jnothman, Bart133, Cburnett, Tony Sidaway,
Mwilde, Tertiary7, Gimboid13, Essjay, Graham87, Jshadias, Coneslayer, Jehochman, Cyberparam, Wknight8111, Jeremyp1988, RexNL,
Turbotron, Tomer Ish Shalom, Alphachimp, Webshared, Adoniscik, YurikBot, Mysid, CLW, Vanished user 34958, JPushkarH, Nvd,
Johnpseudo, GrinBot, Zvika, Trolleymusic, SmackBot, PEHowland, The Photon, KocjoBot, Mdd4696, Ohnoitsjamie, Betacommand,
Andy M. Wang, Reza1615, Kurykh, Landen99, Can't sleep, clown will eat me, Gbuett, Spectrogram, Cybercobra, Savidan, Dreadstar,
Rekh, Soumyasch, RomanSpa, Dicklyon, Belizefan, Caiaa, Kvng, Hu12, CapitalR, Vermolaev, Van helsing, Rohan2kool, Jefchip, Indeterminate, DumbBOT, ErrantX, Briantw, Thijs!bot, Epbr123, Seth Nimbosa, Trlkly, Gkhan, MER-C, Raanoo, Bongwarrior, Recurring
dreams, Whoop whoop, Yewyew66, Allstarecho, Ged Davies, SamShearman, MartinBot, Glrx, JohnPritchard, 3dB dar, Dr Alan Hewitt,
Rickyrazz, Chinneeb, King Lopez, Jmrowland, Alinja, Ctmt, Mistman123, One half 3544, Inductiveload, Swagato Barman Roy, Turgan,
Pdfpdf, Iquinnm, Travelingseth, Yswismer, Joaopchagas2, Cgrislin, Dspanalyst, Pinkadelica, OSP Editor, ClueBot, Binksternet, Dawdler,
Yangjy0113, Niceguyedc, TypoBoy, Abdul muqeet, Drew335, Mutaza, Excirial, Mohammed hameed, Hezarfenn, Moberg, Versus22,
Pantech solutions, Johnuniq, Semitransgenic, Sandeeppasala, Skarebo, Stillastillsfan, Siddharth raw, Thebestofall007, Captain-tucker, Fgnievinski, TutterMouse, Fieldday-sunday, Stan Sykora, Cst17, Je8080, Shekure, WikiDreamer Bot, Frmatt, Legobot, OrgasGirl, TaBOTzerem, AnomieBOT, Eart4493, Materialscientist, Thisara.d.m, LilHelpa, Agasta, SkiAustria, Kesaloma, A.amitkumar, ShashClp, Prari,
SpaceFlight89, Allen4names, DARTH SIDIOUS 2, Helwr, ZroBot, Josve05a, GeorgeBarnick, Orange Suede Sofa, Tot12, Paileboat,
TYelliot, Rememberway, ClueBot NG, Vinras, Jack Greenmaven, AeroPsico, Frietjes, Mesoderm, Rezabot, Helpful Pixie Bot, Gfoltz9,
AvocatoBot, EmadIV, CitationCleanerBot, Jeancey, Humourmind, Physicsch, Rogueleaderr, ChrisGualtieri, YFdyh-bot, VladimirDsp,
Tcs az, Webclient101, Jogfalls1947, Striznauss, Stevebillings, Radiodef, Sisyphus110, Isarra (HG), David Lloyd-Jones, LCS check, Ugog
Nizdast, AndyThe, ScotXW, Arta1365 and Anonymous: 271



Jpeg2000_2-level_wavelet_transform-lichtenstein.png License: CC-BY-SA-3.0 Contributors: Own work Original artist: Alessio Damato
File:Question_book-new.svg Source: License: Cc-by-sa-3.0
Created from scratch in Adobe Illustrator. Based on Image:Question book.png created by User:Equazcion Original artist:
File:Wikibooks-logo-en-noslogan.svg Source:
svg License: CC-BY-SA-3.0 Contributors: Own work Original artist: User:Bastique, User:Ramac et al.


Content license

Creative Commons Attribution-Share Alike 3.0