tpt naan memantine | wNote-2 PONG inc era ea od ay te 1p) DELTA MODULATION (pvt DELTA MOPOLATION owt PCM ystems Eramemiby oll Lie bits which ave used 50, Wak Ihe aignoling scoke omd oma ural axe aubte Jorge. te code a. sample » Avensmission chanel b Signaling eE Ce) = 74, Dono. f WIT /pwhse i 4. > Sompling 4req: L te 224.| Bomd widlkr = See “TS overcome his problem, we used Delta Modulation DY Ayomsmiti onky one. LAr pes nomgle. Heve the present Aamnpie, value Ws compared wii se a on value amd \Ris verwd wheallver See ot ke : 5 in » sed = Accremred % Tien ee he . = aryoxinated to sep Mg @ The difference eat re hee (tah Fa UE tngek nigh Ce) geik. er cate YE gna confined te boo Jevels jes, Fa emd — 4 EE re i de Smereoned, Ue ALB. ts Gove) , hen aggre omated vlageas © Bnet by one hep, i.e: ta’, 3b Wwe Ay. & negolive en : : Aur ! Gil as ras ca 7 ABisepSiee) 3 mv on Tey Sampling ‘ period . a oio!g! rele time >DELTA MODULATION: “TRANSMITTER a : DELTA MODULATION TRANSNUTTER edi Sommpled itp, te esr One bik bos) 2h) a ett mn ui) | cece: : Acciumulatos, — —! Principle of DUt The exvoy behocen Ire sampled Vole of 2.1L) amd Lark approdmaked sample to qiven by. eQ@u)= amt) - 4605) —>O where a(n) = ester peer aomple aulnre) = Aompled aignod of xO) Ay) = Look Aamgple agpronwation oflWe chokvease © -amefoon . ah we assume Lenk utnt) oo We presenk aemple agprocination of stalresre op biker , uso] = alms) —~@ = Last ramagle appronimalten of shabreant wane form. Lek us define a quantity Lew) Seach 0 eg | Wak Own) = Asgne(ns) + @ “thio means hak depending on Mie Aigo oh exrey ees) The Atapn of step hee A. decsded , dn ole ORD be can writebmw) = fre AE) KT) 2 4 4 NT) & awn) ® : Moo Ye BET) = d dhen a binary ‘YU Aponewllhed ama Uy BET) = — 4 han a Binary ‘0 otrammmited, Sexe Qe Somplings arttevel The summer sm le accummator adds quantizer ofp G4) WWW INS previous namie agprofimalion The glver prentnk © Aampe agprnimalon AO) = a Goat) + [24] oF UWS) = UL T+ bu) — 6 ; a previous sample: areroeimaten ul(n-i)x] o veobored’ by slaying one rangle peri Tr. he Aausplel ixp aignal x(vt) amd slatrcare approximated aged E(mo) are subkocted to get err aro ene), Thas , depending on \he Aign” of eet) one eit qonkizes genesakes am ofp of +4 ov — 4. ee Ai2e is 4.4 hen: binary “A Vs Avemsnitted ama oh nt se =A, en binary 00" ws tromemired . RECEIVER SECTION & Ie 9 Scaeaes Sena | ! ue A eeumulaloy ® Tr INZ weceiver seelion ile aceumulatsy ond Ler axe used. The accumulator genemates Une Aeiecaneapproximated xgnal ofp ands delayed by ome J) puree pested any 4 3% > Men ine! es ae Up alana. Sb Up Lima ‘Yen we meat Shep to previous ele Culhiely ddayed. Shp » binary ‘0! én one cep ‘A’ bp aubleacted {rom Wie delayed signal . Also, the Ler thos WE cu- of frequency equa Ke Adaghed- Frespenny— = a: “Thy LPF Amoolhens We abalycase sci Ko se combrucl origenoh menage Frama ate) ADVANTAGES OF DM 45 The DM Atamemle~ only one wit for one erie . ; an . Arronsmusston therefore re signaling rae ond 2 channed band widlh io quite small Ser compared ao Po, : 2 The Wh. 4 Re Arnglemmento von every murch romple. | efor Aalto. module. Disa DyANTRGES : ol iy Slope overload -distestty | BY Growler ov dhe move ee r : ee Granular noise Slope overload dishorhon : : J OE) ) a : Stoly-cose — approximation kr “Ct)re Overload at ‘b> Shope Varese eo cet When Ihe vale of wise of “yp Algnal x(t) b> +2-Righ Uk lke balveose rignah comne® approtimake SE arigmak 0), SKE step-alze wy becomes too Amall fox stoivense sigmod ule) to ofallaa (Ke shep segment of als) . Hence Were is oe large errr beboeen \Ke sloly care axpp coxvcmatre Aignak amd. SKE originak Bayle aignak 9 Ernie Lem oF noe Yo Rumen ov alope querlecd distortion. 70 veduce Hilo erry She step ste wut be Amewenved shen lope gral tay St Riah- ‘Stree \We hep ate of DM yemains fred AT Raxcmum ory wont ho \ , eee ee eee Medulatow , Cordon to avoid sopeoverlond Sisteriion to A : => mex an) | D> Granular ov-Idle Noise + Granular ov Sdhe nove occust when Whe ahep do to loge compared. te smolh variation’ tn We Yp ALynal.. Wh means MRak for very aah vomalions om Ue Up yn, Ihe slaiveare nignak to changed bg Se taryt camounk a’ because of Loge. shepwee. When lhe Hp aye almost flok , Mire shobv eave Dagrrak uley Keepd ov os¢ Waking, bys tA oxound Whe Pranoak “The exer beeen We fp amd approximated rignah vs called gronubor move. The olution to Wis problem % to make ale size amolh. WN———— VY Band Width of DH Barnes Nid Neg BY Nowbes of Bib fomple = 7 LIS] see Nov of Somples/sec = fe samples [Jee A Number of WHT/See = Numbere| BUT/sample x Mombersh iano enfse = vxd, MK /see . Total wumber of BI /sec bs coed aaynallny vob coat te ‘ : ryxok, where 4 724, , Banduidilr (@wy = Y= rhe yh” Z im cave of vey vod “OW DA = se SNR FOR S\W19S01DAL_ MODDLATION Comdex a nigna : We) = ho Co* (andy) : the maximum Aspe of Lhe agra ay) is given by i man Se = 208 he x overload es no rhage cvertaad ee Hove 08 A k EE ae mux| \ cr ZV Bw, ho : =f AEE oe oo Sod ghegenah 4 2 MS ohhes oe v scaible value of Ue ele 4 es Hence lke moxemr perniss \e vi aagpede power equols Prox = aoe - oy ~ When Lee ts mo cloge-overtond (Ke . maximum qpantzaon exsor 2 £4 We Samet | amtume kok See quoukizing eres unifeeny— | dotibuted | So, lhe variance of quambizalion error bn DM oe f kegte [eu > quraaton ever] a | 3 | Vowerer, WE ~trecever Aide contains (ok UF ofp end) a le pom ~ pitger whose badd b ack” equal yee mensaae bandadllie | rer higher postible, fre5- componente of Hee menonge alayrok) + denoted an W, We note Wak 4,8 “Moreover, Ws | | amalh axed 4 he Aan png woke: Yous, ence | anwuming, Hak ‘We average poet of Me auontizahion [| exter b urd forvnly Asbribubed over a Freqreneyy | dnkerval eoctending, 4rom -+ +0 & we ger ke | veoh ‘Ss - \ Average ofp nase peer aN (ee ) | So,lke. masiwim ENR) = cae Vee mon WT (2% “a G%) eget featTo overcame HE quankizalon exrors due $7 slope overload ; ond qrocnula more the step ahre (A) » mode adaptive to variations in \ke supek sagna eX) Porioularly. a Mhe sheep segment of Were saan att), Whe alep - maze ts mereared. Alo, U4 the Supe verging show Ue ckep alee vs veduced Thin, Vela welled Known ao Adaptive Delta Modelabion Av). The adapkive della wodolateys com Yoke conkinows changes » ahep size ox Avcrele changes i) wt arep-aite ree eat for e4t time T es LA quonkred, Aagrol- igac wave from Sor ADM | “Ik. Secon of ADM * > opt Put Avarible ze pcesmuaby (Fy -2)Re. Sechion of Adtt ma B® E Togic a Atte contro a. en lhe Yan smith ex CRg2) Acckion \Ke Vogl c oe ia eae combed ic added The sheprize Meveaser oy decreasen accovding to a ae Led tule al . a 4 : pend on one bir gponkiver oe. S$ ome bh. quomkizer fe to Saige (ere 0] Wen ley alze may be doubled for nexk sample , 4h one bit quantizer ele & Sars , Men prep rte ‘way be weduced ba one akep . Ary. A aAncorlhe shetvcare woveborms of adoptve delta modulotor, omd We reapence of bils te be Womsited @ am. 3m WE veceiver of adapttye Ae\ba modaboker hewn figs _ Were exe +wo postions me ' Seer portion produces \e pep ote Soromench ‘ ee eee eoekn Ve some proces ve Falla ead, an \Kok on Yeonsm her . The preset sapele decides (ke ave ahze , SY te Wen applied ¥e an accumulatox alich builds we Alaxrcace Rave form ~The \suo-pars filter tken sruothens owe Mee Avalirens é wove Lorn, to scecanahuet— i Ike origina Dhyana 2 i @Advantages ot ADM. ML y The SNR becomes better lKein oxinory AeVre-- mokubakor becaune of \WE seduckion in slope overload Ais terktow omd idle noise. : i Because of voisialb\e shex- are \Ke Agnormie TOMAS ef ADM \ wider ham Ample “DY. Wi) URUizalion of BW vo beter WHE DM. Problems. Given a ane wove of Arequenca 4. hwgitbude Am applied too Aedke meddator avy skep alse “A; Sha Gok Vee Saye pueriead Molortron will eceoy Tyo bempngeaiod A Am 2 Sach Sdwhons- consider a sine wove : nn aU) = hyn Stn 2 BAe Pom onk- Wke slope of xCt) wil) be mourn when. devivalive -2 ats) ilk venpect Inte? Sh be mart , Step- Size The mouimuam lope ab Me ay ale Sampling pore’ We know tok | slope ever lead Aoler on Sih Yolk place uf slope oh Ame Save & greater Whitin ahoge, ef del\ha moduler Le. : moe | S32 > So hnteln) Se » i)1 Eromgles A delta malubator pogo © Bye’ Memes Une ny qndet woke for on dignak having 2B wok to Seth. Cohouhake WAZ rmoxtwunm amplitude of a 2KMa ww Aimuncid for ubich \We delta modubater doer snot door slope over lnad , Given Wok’ A= 250mv . bs designed to opevahe. a Sdutinr- Lek We angnek bo aU = ASin aah t —+O We macimun alepe of MRE otymal S rman | 228 | = A2wt,, “Tims Alege overload uth nel occur Uy x x wae |S | ‘here ‘ : A> Shep BES be Sampling terval i Aampllng Grey, = Stimen nyquiok valle” ? = 5K G Kh, = Fo ke — SF . eomp a taeruak. Ts 4 >in ee Given Wok 42 25ewV = 0°2SV 4 by = DME 9 LETT bo a 0-25 aes 2 ay T xO? At 4 SA, Lintene PREDICTION THEORY Comider am anakog orga ale) which hove denial Ae a aoe evdedaiok Sime te: Oring Tagiox nexies expansion doe »s Ne cam enprers mlcrte) 7 gives bq Joes m(tat) = md) Te MUD) + we)a =o IR mua TCE) foe aww Te Hhe above equation shoos Hak from Annoledge af he signal ond if derivatives ue cam predick Lie folare value of L2 signal of Hime (t+) - The Rxak devolve ts giver Be wo QR) = | a mngen) = Ge : bak S e Subiuking ya Ger) 3m ire aleve eqpotum mCunyt = MOE) + Ger) GTS Sam UT) - MACIGS mieayut = am 08) a MKNTY dy Or > Conbtiiene—. * Men) = ay mle) +a Lay y - Ly predicted vole, Why to Lhe equ op he 2nd over yreMeby The 247 for WW over predicey FOB) = ay mCG) agra VAhIE G+ ~~ -—ravmfQn nnd The equ? given ba fH Med) < arm WI ta Renman oo MOK order predichy, since Ike valun of Nat. 4b Ie vols of oi} then we call Uli Int order predic ty to be nothing bub ow deley element Ne 9fiers So Ditbeventiok Pune Code Hodvbabunn Goren) 7 - to WE Aromsmi scion of, Inntorrmakin axe gene roked : Y) BM AY When a. voice ‘ox vicleo signal ia sampled of a rate sbigh Higher Lian Me nyquist vale Cis Pom) Mie veoulling Aawple’ Aignal is found +> exhikit o figh covelobion bekicen adjacent sampler. The meaning, of this Righ correlohon ia Thak in am average sence , LE aignah deenrs\— change rapidly from one sample to the nex” itr Whe vesult LKak Lhe difference bebween adjacent amples Rao eo variance LKak to omolhor Dharm We varionce of ke nugnod ahh . when the Ad correlated samples are encoded Ure yeaubing ‘encoded Mgnak conteine xedundeni aaa malton . Lee. agmbels Bak orem abackidady ersovriak By Yemoving this vedundeney we obloin a more ebticlent : ode Modulation. esded aignah ive. Differentioh Puve Cod ) 0 aknt) Redindank dnfrnadiem of eMTiansmilles Seon of Dect | 3 aan I Somapled i calgitiss, g prem Signa ys) ihe von Yorke eocHie principle, of prectinten. he ae aan present dample 1s predicted. Prom lhe pork amples The predialion ariay fat be. quantizalion error. 1g ike predickiom Biter Ye GLny)- myn) = Ca.) + % (nt) ——(S) A bo) predicded snmylereles Substitahing, Ihe vole, of eg (n) From eqokton 2) wm equ 3) ve get, (nt) = XB) +eGh)s 405) —+@ a La a[serene ng ne ne cna Using equilion @ + Q) we gor Hy Ov) = A) + 2007) - 294) +15) 2 Hence , SE quamteed version of ihe aignoktetvt) w \Ke sum of exiginah sample volua and quo zation SMV gta) The quankzslion evry can lb eo negokve ov gostlive The equ © dees nok depend on tke predichon filler choroctevyLhea - Rx. Sechion of deem: Rx. Seclton of deen: DPCH €) Gee . filter + macht) “Te decoder Sint veconsrucl Mie quantized 2xTOY Atanok frm smcomi: binowy cago, he prediction Rlber ofp omd gquoinkized ever ney axe summed up Ie aive IKE quantized. version oF MWe ovginok pigneh. Tivo Ue aignnh ok IKE Yeeewer Lifer: From achunk mrsgnoh— al quake akin exon 4s), ess Str duced porronely In Sho ve combnrcked otro