Digital Signal Processing II `Advanced Topics’ Marc Moonen Dept. E.E.ESAT, K.U.Leuven marc.moonenesat.kuleuven.be www.esat.kuleuven.bescdDSPII Version 20092010 Lecture1 Introduction p. 2 Lecture1 : Introduction • AimsScope Why study DSP ? DSP in applications : GSM, ADSL,… • Overview • Activities Lectures Course NotesLiterature HomeworksExercise sessions Project ExamDSPII Version 20092010 Lecture1 Introduction p. 3 Why study DSP ? • Analog Systems vs. Digital Systems translate analog (e.g. filter) design into digital going `digital’ allows to expand functionalityflexibility… (e.g. how would you do analog speech recognition ? analog audio compression ? …? ) IN OUT IN OUT AD DA 2 +2 =4DSPII Version 20092010 Lecture1 Introduction p. 4 DSP in applications : GSM Cellular mobile telephony (e.g. GSM) • Basic network architecture : country covered by a grid of cells each cell has a base station base station connected to land telephone network and communicates with mobiles via a radio interface digital communication formatDSPII Version 20092010 Lecture1 Introduction p. 5 DSP in applications : GSM • DSP for digital communications (`physical layer’ ) : – a common misunderstanding is that digital communications is `simple’…. – While in practice… Transmitter 1,0,1,1,0,… Channel x + a noise 1a x Receiver decision .99,.01,.96,.95,.07,… 1,0,1,1,0,…DSPII Version 20092010 Lecture1 Introduction p. 6 DSP in applications : GSM • DSP for digital communications (`physical layer’ ) : – In practice… – This calls for channel modeling + compensation (equalization) Transmitter 1,0,1,1,0,… + Receiver ?? 1,0,1,1,0,… noise `Multipath’ Channel .59,.41,.76,.05,.37,… DSPII Version 20092010 Lecture1 Introduction p. 7 DSP in applications : GSM • GSM Channel EstimationCompensation – Multipath channel is modeled with short (3…5 taps) FIR filter H(z)= a+b.z1+c.z2+d.z3+e.z4 (interpretation?) – Channel coefficients (cfr. a,b,c,d,e) are identified in receiver based on transmission of predefined training sequences, in between data bits (problem to be solved at receiver is: `given channel input and channel output, compute channel coefficients’). This leads to a leastsquares parameter estimation procedure (see Linear Algebra course). – Channel model is then used to design suitable equalizer (`channel inversion’), or (better) for reconstructing transmitted data bits based on Maximumlikelihood sequence estimation (`Viterbi decoding’). – Channel is highly timevarying (e.g. terminal speed 120 kmhr ) => All this is done at `burstrate’ (+ 100 times per sec). = SPECTACULAR DSPII Version 20092010 Lecture1 Introduction p. 8 DSP in applications : GSM • GSM Channel EstimationCompensation • GSM Speech Coding – Original `PCM’signal has 64kbitssec = 8 ksamplessec 8bitssample. – How to reduce this to 4000 512point FFTs per second ) • basic sampling rate is 2.21 MHz (=5124.3215k) 8.84 MHz AD or DA (multirate structure) • fixed HPLPBP frontend filtering for frequency duplex • adjustable timedomain equalization filter (TEQ) e.g. 32 taps 2.21 MHz filter initialization via leastsquareseigenvalue procedure • adaptive frequencydomain equalization filters (FEQ) VDSL specs • e.g. 4096point (I)FFT’s, etc…. = BOX FULL OF DSPMATHEMATICS DSPII Version 20092010 Lecture1 Introduction p. 16 DSP in applications : Other… • Speech Speech coding (GSM, DECT, ..), Speech synthesis (texttospeech), Speech recognition • Audio Signal Processing Audio Coding (MP3, AAC, ..), Audio synthesis Editing, Automatic transcription, DolbySurround, 3Daudio,. • ImageVideo • Digital Communications Wireline (xDSL,Powerline), Wireless (GSM, 3G, WiFi, WiMax CDMA, MIMOtransmission,..) • …DSPII Version 20092010 Lecture1 Introduction p. 17 DSP in applications Enabling Technology is • Signal Processing 1GSP: analog filters 2GSP: digital filters, FFT’s, etc. 3GSP: full of mathematics, linear algebra, statistics, etc... • VLSI • etc... SignalsSystems course (JVDW) DSPI (PW) DSPIIDSPII Version 20092010 Lecture1 Introduction p. 18 DSPII AimsScope • Basic signal processing theoryprinciples filter design, filter banks, optimal filters adaptive filters • Recentadvanced topics robust filter realization, perfect reconstruction filter banks, fast adaptive algorithms, ... • Often `bird’seye view’ skip many mathematical details (if possible… ☺ ) selection of topics (nonexhaustive)DSPII Version 20092010 Lecture1 Introduction p. 19 0 0 .5 1 1.5 2 2 .5 3 0 0.2 0.4 0.6 0.8 1 1.2 P assb and R ip ple S top band R ipple P assb and C uto ff > < S topb and C utoff Overview (I) • INTRO : Lecture1 Lecture2 : Signals and Systems Review • Part I : Filter Design Implementation Lecture3 : IIR FIR Filter Design Lecture4 : Filter Realization Lecture5 : Filter ImplementationDSPII Version 20092010 Lecture1 Introduction p. 20 Overview (II) • Part II : Filter Banks Subband Systems Lecture6 : Filter Banks IntroApplications (audio codingCDMA…) Lecture78 : Filter Banks Theory Lecture9 : Special Topics (Frequencydomain processing, Wavelets,…) . H1(z) 3 subband processing 3 G1(z) H2(z) 3 subband processing 3 G2(z) H3(z) 3 subband processing 3 G3(z) H4(z) 3 subband processing 3 G4(z) + IN OUTDSPII Version 20092010 Lecture1 Introduction p. 21 Overview (III) • Part III : Optimal Adaptive Filtering Lecture10 : OptimalWiener Filters Lecture11: Adaptive FiltersRecursive Least Squares Lecture12: Adaptive FiltersLMS Lecture13: `Fast’ Adaptive Filters Lecture14: Kalman Filters .DSPII Version 20092010 Lecture1 Introduction p. 22 Prerequisites `Systeemtheorie en Regeltechniek’ (JVDW) `Digitale Signaalverwerking I’ (PW) signaaltransformaties, bemonstering, multirate, DFT, … `Toegepaste Algebra en Analytische Meetkunde’ (JVDW)DSPII Version 20092010 Lecture1 Introduction p. 23 Literature Campus Library Arenberg • A. Oppenheim R. Schafer `Digital Signal Processing’ (Prentice Hall 1977) • L. Jackson `Digital Filters and Signal Processing’ (Kluwer 1986) • P.P. Vaidyanathan `Multirate Systems and Filter Banks’ (Prentice Hall 1993) • Simon Haykin `Adaptive Filter Theory’ (Prentice Hall 1996) • M. Bellanger `Digital Processing of Signals’ (Kluwer 1986) • etc... PartIII PartII PartIDSPII Version 20092010 Lecture1 Introduction p. 24 Literature DSPII Library • Collection of books is available to support course material • Listinforeservation via DSPII webpage • contact: beier.liesat (EC)DSPII Version 20092010 Lecture1 Introduction p. 25 Activities : Lectures Lectures: 14 2 hrs Course Material: • Part IIIIII : Slides (use version 20092010 ) ...download from DSPII webpage • Part III : `Introduction to Adaptive Signal Processing’, Marc Moonen Ian.K. Proudler = support material, not mandatory …(if needed) download from DSPII webpageDSPII Version 20092010 Lecture1 Introduction p. 26 Activities : HomeworksEx. Sessions • `Homeworks’ …to support course material • 6 MatlabSimulink Sessions …to support homeworks …come prepared • contact: amir.forouzanesat (English+Persian) beier.liesat (English+Chinese) prabin.kumarpandeyesat (English+Nepali) pepe.gilcachoesat (English+Spanish)DSPII Version 20092010 Lecture1 Introduction p. 27 Activities : Project • Discover DSP technology in presentday systems examples: 3Daudio, music synthesis, automatic transcription, speech codec, MP3, GSM, ADSL, … • Select topicpaper from list on DSP II webpage (submit 1st2nd choice by Oct.5 to pepe.gilcachoesat) • Study www surfing • Build demonstration model experiment in MatlabSimulink • Deliverable : – Project Plan: 1 page status report time plan (send to marc.moonenesat by Nov 1st) – Presentation: .ppt or similar, incl. MatlabSimulink demonstration (December, 20 mins per group) – Software • Groups of 2DSPII Version 20092010 Lecture1 Introduction p. 28 Activities : Project TopicsPapers • List available under DSPII web page • Other topics : subject to approval (email 12page description to pepe.gilcachoesat before Oct. 5) Tutoring 10 research assistantspostdocs All PPT presentations will be made available, for ref.DSPII Version 20092010 Lecture1 Introduction p. 29 Activities : Exam • Oral exam, with preparation time • Open book • Grading : 5 pts for question1 5 pts for question2 5 pts for question3 5 pts for project (softwarepresentation) ___ = 20 ptsDSPII Version 20092010 Lecture1 Introduction p. 30 homes.esat.kuleuven.be~pepedspII • Contact: pepe.gilcachoesat • Slides • Homeworks • Projects infoschedule • Exams • DSPII Library • FAQs (send questions to pepe.gilcachoesat or marc.moonenesat )
Digital Signal Processing II `Advanced Topics’ Marc Moonen Dept E.E./ESAT, K.U.Leuven marc.moonen@esat.kuleuven.be www.esat.kuleuven.be/scd/ Lecture-1 : Introduction • Aims/Scope Why study DSP ? DSP in applications : GSM, ADSL,… • Overview • Activities Lectures - Course Notes/Literature Homeworks/Exercise sessions Project Exam DSP-II Version 2009-2010 Lecture-1 Introduction p Why study DSP ? • Analog Systems IN OUT vs Digital Systems IN A/D +2 =4 OUT D/A - translate analog (e.g filter) design into digital - going `digital’ allows to expand functionality/flexibility/… (e.g how would you analog speech recognition ? analog audio compression ? …? ) DSP-II Version 2009-2010 Lecture-1 Introduction p DSP in applications : GSM Cellular mobile telephony (e.g GSM) • Basic network architecture : -country covered by a grid of cells -each cell has a base station -base station connected to land telephone network and communicates with mobiles via a radio interface -digital communication format DSP-II Version 2009-2010 Lecture-1 Introduction p DSP in applications : GSM • DSP for digital communications (`physical layer’ ) : – a common misunderstanding is that digital communications is `simple’… Transmitter 1,0,1,1,0,… Channel x a Receiver + noise x 1/a decision 99,.01,.96,.95,.07,… 1,0,1,1,0,… – While in practice… DSP-II Version 2009-2010 Lecture-1 Introduction p DSP in applications : GSM • DSP for digital communications (`physical layer’ ) : 59,.41,.76,.05,.37,… – In practice… Transmitter 1,0,1,1,0,… `Multipath’ Channel + !! Receiver ?? 1,0,1,1,0,… noise – This calls for channel modeling + compensation (equalization) DSP-II Version 2009-2010 Lecture-1 Introduction p DSP in applications : GSM • GSM Channel Estimation/Compensation – Multi-path channel is modeled with short (3…5 taps) FIR filter H(z)= a+b.z^-1+c.z^-2+d.z-3+e.z^-4 (interpretation?) – Channel coefficients (cfr a,b,c,d,e) are identified in receiver based on transmission of pre-defined training sequences, in between data bits (problem to be solved at receiver is: `given channel input and channel output, compute channel coefficients’) This leads to a least-squares parameter estimation procedure (see Linear Algebra course!) – Channel model is then used to design suitable equalizer (`channel inversion’), or (better) for reconstructing transmitted data bits based on Maximum-likelihood sequence estimation (`Viterbi decoding’) – Channel is highly time-varying (e.g terminal speed 120 km/hr !) => All this is done at `burst-rate’ (+- 100 times per sec) DSP-II Version 2009-2010 = SPECTACULAR !! Lecture-1 Introduction p DSP in applications : GSM • GSM Channel Estimation/Compensation • GSM Speech Coding – Original `PCM’-signal has 64kbits/sec = ksamples/sec * 8bits/sample – How to reduce this to