CA2388577A1 - Vsb-moe pre-equalizer for 8-vsb dtv - Google Patents

Abstract

A vestigial sidehand (VSB) modulation minimum output (MOE) pro-equalizer operating on a received ATSC 8-VSB DTV signal includes one or more of an adaptive food-forward filter and an adaptive feedback filter each utilizing only reel adaptive coefficients, with the direct term for the overall filter constrained to unity to remove one degrees of filter parameterization freedom. Adaptation of the filter(s) is based on minimizing a blind energy cost function, and is independent of adaptation of a conventional adaptive channel equalizer. The pre-equalized signal is passed to the conventional adaptive equalizer for channel equalization utilizing DFE, IIR-CMA, etc., resulting in performance improvements including faster convergence and greater robustness with relatively small implementation costs,

Description

29.osaoo2 VSB-MOfi p:e-equa~lizsr for 8-VSB DTV
This application claims priority to Uruttod Stator provisional applications saiial nos. 60/295,864 and 60/297,497 filod Juao 5, 2001 and Junc 12, 2001, respectively, which are incorporated herein by reforeace.
The present invention is directed, in g~orat, to channel squal~atioa for wireless sisal recoption and, more specifically, to channel equali~ticm for,~ivers daurodulattiuy~, vCatixial sidtband modulated wireless signals.
Wird~s ooiuiuuuicatious suds as tLusa cuupluycd fur digital lxlt3vision (DT'V~
bxoadoaats typically require c~haoncl equalization for sucocssful demodulation and decoding of the received sisal. For a variety of reasons, such as failuro of sync basod tlaaing recovery achemeR under mulfiigttth channel interference conditions, the perFoimance of conventional adaptive oqualizers in aunetrk uaa iq not entirely satisfactory. Moreover, oextain ~orms of wireless ~:vmmunication such es vesti81a1 sidabaad modulation have specific propertie.R that rosy ba advantageously CXpIUItCd.
These is, therofnro, a Good in the art for augme~ntatlou of 2idupiivv e~h~tnnel equalization, and in particular augue~ntatioa oustamized to the porticular properties of the wireless coanmuaicatio~ modulstioa scheme etaployod~
To address the above-diacmaed cle~tciencies of the prior nit, it is a primary object of the present invention to provide, for use in television r~aceiver., a. vestigial sid:eband (V9B) modulatloa uiiuiruuus uu,~pu~ enemy (MOE) pro-oqualizsr oper~ing on a rccdved A'TSC 8-VSD DTV signal flint It~cludss uuC ur ruuro ur an adaptivtj food-fotvvard ~l~c and as adaptive feedbndC filter each uWiziag only rent eoeffidoats, with the direct tcm~. for tuo overall filter constrained to unifiy to remove one dograe of $ltcx paramctcriuttion firaiom.
Adaptation of the fiiter(g) is based on minimizing a~n energy cost function, stud ie iadepandeut of adaptation of a conventional adaptive channel equalizer. The actualized signal is passed to flay couvc~liuntsl r~idaptiv~e channel equalizer fox channel equalization utili~ng T)F'rr, 1'1R-CMA, etc., rcaaltiag in perfosmaucc impmve~uauts iucludis~ fisal~cx wnv~euce and robuetneos with relatively Small ixnplcmcntatioa costs.
The foregoing has outliaad sathar broadly the fi~atures sad technical sdvantaQes of the ttreaent 9nvesttioa so that those slcillod in the art may better widesetaad the lr7.VJ.bV Vlr deisiled doscrlption of rho iuvauli~ou that follows. Additional fes~ures cad advantages ofthe Invention wlh be doscxibcd h~eroinaftor tb~x force the subject of the elaiaas of the invention.
Those skilled iu the art will nppxcoiate that they rosy readily toss the conception and the specific cmbodimeat disclosed erg s basis fna~ maiifiying or deigning other stnua~s for S oarryi~ out the same purposes of the preescnt invention. ~hoae sklllect is Lho art will also realize that such y ivalent eon~tCdons dv nol. ~lcpart fcorn the spirit nerd scope of the iaveutlnn in its br08dCSt form.
Sefurc~ W'tukiag the dctailod description of the iaveatio~t below, it rosy he ~sdvtuita~cous so sat forth doi~tions of certain words or phra9cs used thmu~oui this pafeut docurae~nt: the t~ "include" and "conapris~ " as well es de~ivssiives thereof, menu Inclusion without lfmitatioa; this term "or" is inch»ive, meaning end/ot: the phtssos "ash with"
a»d "asaoGiated therewith," ss well as derivatives tl~toof, may mntuZ to i.tlcludo, be included within, intm~coanect rah, gain, be cxmtained within, ooaacct to err wig, couple to or with, ha communicable wish, cuupere~o with, intarleavc, juattapose, be p~t~oxlmate to, be bouad to or wit>a, travo, have a property oi' or the like; and t'1 team "controller"
uaesna any device, system or part thereof that controls at least one opcratian, whether such s devfee is imple~e~ed in hardware, flsmware, sofinwarre or same ecuabination of at least two of the same. It should be noted that the functionalfty assooiaud with any particular controller may be cenlrali~.ed. or distributed, locally or iamulnly, Deflaitiona for curtain words and pare provided lhtvuguwut this pat~t doau~mcstt, and those of ordiawy skill is ~ art will undcrataud drat such dofinitioaa apply in many, if not most, iastancae to psior as well as future uxa of such debased words and. phrs~a.

2 5 For a more complete underb-tandinK of Iha present ion, and the advantages 'lhai'eof" ~ i,~ now ~xde to the following deecriptioaa taken is oo~jvootion with lha acroviuiaanying drawings, wh~aia like numbers deeig~ like objects, and is which:
Fig. 1 dcpict3 a wireless cammunieatiotu receiver sy=fens including a pm-cquolizw for i~aaproved pe~rformauce aacosdi~ag to one dmbadimerrt of the print inve~on;
Figs. 2A tBrough 2D it1»ctrate a pre-equatizcT $ltc~r in a~c:c;ordcmcr~ with various emhc~d~ment of the p~s~t lnv~on; and Figs. 3A-3C3, 4A-4d, SA-Sl7 cad 6A-6tl relate to performance eimutatioas i~or w:uuveutioaal adaptivo oqualizGr with nerd without a pre-equsiizer aoc~rdiog to one orabodiment of tiers pros~t invention.

3 29.05.2002 rigs. 1 through GCI, aisouaacxl baluw, and thn various embodiments used to describe the priaoiploa of the precept invontion in ttzie patent docuiuaul. an by way of illustration only and should not be ooastrued. in any way to limit the ecopo of the invention.
Those skilled in the art will uaderstaad that the principles of the present invention may be implemented is any mifihly arranged device.
Fig. 1 depicts a rwvireless communications receiver system inciWing a pre-oqualuor fur improvdd performance according to one embodiment of the present invention.
R~ooo3vor ayabem 100 includes a roaaiver 1 O1, which is a digital telowision (D1'~ rrcesivGr in tho oxomplary e~nbodima~, including au input I02 for receiving wireless acigaals, optionally as output 103, and a demodulator 104.
Ia the example slsourn, seaeiwer 101 is intended to receive sad derxtadulate ve~.~ial sideband (VSB) sigaalc with eight dis~cte IevelR (R-'V'S>~) according to current 1 S Advaacxd Tnleviaion Systems Committee (AxSC) standards, However, receiver 1 O 1 may alvely be any typo of i~ivr~r fur a waununi~tiuna~ ayslam mquiriug cvnununlcations ehaaael equnli2~ioa end employing sync based timing recovery. Rcaciver l0I may ~eraforo be any audio and/or vfdoo oommunioakious reaeivet inoludiag a satellite, teaestrial or ca'blc broEtdcast rcceivQr sad/or television, a video cassette recorder (VCFt) or digital video recorder (DVR), or a digital versatile disk (nV0) player.
Those skilled is the arc will recognize that the full dxtails of the constcucdon sad operation for a complete wimlass coaniuuuicatioas r~ivcr tun nut dcpicixd in ihn dravviugs or dcaaribcd herein. Instead, for simplicity and clarity, only ac much of the eoastnLetioa and operation of a wireless aommunioatioas eystsm as is unique to the prosoat - 25 invention os necessary for an understandia~g of the precept invention is depicted sad described.
R.oceivor 101 in the exemplary earbodimont indudaa a ualixer 105, a timing rocovary loop 106, euul an adaplivcs nqualixt~r 107 wttbin demodulator 104. Pro-equalizer 105 is thus iatondod to augrncat, but sot replace, a conventional adaptive equaliiccr 107. Yn its simplest form aaoordiag to the present iavontion, pro-oqualixcr 105 consists of a relatively ~ahnrt, transvessal vaeax alter, typically with 16-24 real eoef~leieuts (although ss little as a few coe~ciemts or a.5 many a.R several hundred could be employed), which ~ on complex sy~ubul ~pa~:nd input samples cad $enerares complex symbol spaced output samples.
The adaptation algorithm oiuployod is a wnslrainnd minimum output energy (M08) blind 4 29.05.2002 criterion similar to t1u co~ct ~taotioa ecx~tployed for blind equaIiu~on witlurui ~~
id~flcation, btrt uniquely expioittng the prap~ies of vaedgial sidcbaad modulatto~t.
In vefthlal sidebsnd mnduladon the lower sidebaad is missing and eoataiue an ai..t aigual tazrsay (i.o.. no si~cant lnfnrmat3on that can ~aprove perfoanan:aa), smote that the prinuiplc vquauZCr fssuts is to avoid adverse noise enhancemextt in this lower sideband that is, avoid high equalixor gains iu tlsis aidcbnnd since such gains would only ~plif~r aoise~ The solutloa adopted is the t inveutlwz is in cwnalxaiu the pro-t~ttalizer tilts 105 be being purely real, which aiay be regarded as a oomplcx filbor having au imaginary part of ~o. Because this corrstraiat ianpliet x aymmebrio filter m~ituda reaipoz~. the mod~ost gains that apply at any positive fi~equelnoy tire coa~rsined to also apply at uo~tivo frequ~awaos, Ltru f~lura which limits noise smpl~iaation_ Figs. 2A through 2D 111ustrato a p~qalter in accordance 'with various embodiment of the prafla~at iavontion. The strucrdue and operation of tlio luo-oqualit~r 105 tun apeafically tailored for vesti~lal eidebaad modulation traasa~i~iona is tbo point invGnnou.
I 5 The crperni'ion of pre.equalizer 105 is described hernia is 'the co~upleac bs8eband enp~n, vuhere the lower videbwt~d is i:acgaty ab9esJt_ In this conteact, complex baseband ra~~ea~ou moans the si~nels are complex sad at the sym~l rate (thus forminE
lVyc~nst sanrplos). As such, rtes prG-cduali~r 105 accepts a complex ai~rsl as input, convolves with coal oooffiaica~. sad outputs a eompler~c siganl.
Fig. 2A illus~ratae the input and output relatioa;>bipa o~tho real pre-equalizer ~t~ x~~~ deaot~ tl~ z ~f~ oftbs ~a ~-eq~~ ins aooo~a~ag to:
FI(t) -1r~ t h,z'' + hay +...
ho . h, . hs . efic. are real. lxt :(k) denote the complac input of the restl pre-equalizes filter 105 at tiso~e k, with rest and iiaagiaory oompoaauuts dnnotod:
rk(k) - Iis (r(k)j .
r, (k) r ~f r(k) ~ .
and let u(lE) dcazous Ilus comply output of the real pre-equatixer' fitter 1A5 at time lk , with real and ima~inaxy coutpon~eats denoted:
~Rt~) _ ~fu(~)~ .
ur(~) = lm{n(k)} .
Thcai the real p~~e.equalizsr filter coastraiae its a~atioa to the following i~aput-output selatioaship as illuin Fig. 2A:

29.05.2002 ua (k) s H (~')~'a (k) u~ (k) = X (q)r~ (k) wee X3(9) is tlLe dolay operator form of the flltar. 'with 2 -lra~naffo~m H(a) . Thus, in the goaerali~cd filter iuputruulput toladonehip i11118ttatod la Fi,g. 2A, the filter H(z) coef'hcieats rug trail vatlues and the Same for the upper and lower branches, while 'the iaptrt r(k) sad the ouxput rt(k) axe both complex.
The seal ~e-eqva~ file 105 is aigt~otvasly eimplor thaa a oomplox file, ~hiah mey be e~pxased is testers of ibur dfatiaat seal filters. For the real pee-$lbor 105, only'~wo filters. constrained to have identirsl rnetliaients, errs required. Whilc the x~oeeon for uslc~ twu iwa~~t coal 8ltars is to l~Cilitate inhibition of adverse not ve aahaacomont yvhcn equalizing vG~ii~~inl didcsband modulated signals rather than sitApliScation of the filter iraplomoatatioa, tho oonetratnt also results is implamartatiou ~w~liun~.
?ho coal pry equalizer filter 105 has a linear Z -traaaforaa ,~,I (z) , whicli may be rood is tho follawiag mauaar to doF~a~e a class of filters tatlvrod speciGw~lly fcu implame~tioa;
~ + d(a) 1 1 + B(a)(o" t e~(s)) ( ) where A(s) s ~.., °c.s~'"' sad 8(s) ~~ ~~ ~ bas'" , which ooaosponds to the atruoture is hig.
2B, where all Slur coe~lcientR are real valued hm input r(k) and output u(k) are both oomplcx. Tn the ermboditnent of Fig_ 2B, p~ce~equaliaar I05 is an auto-rsgrecaive moving average (ARMA) raalizatinn in which tl~e fxedbaalc pasameter 1~(x) iattoduaed by tiltar component 202 at afgaal adder 201 is based oa the feed forward paraazeta~ ao i A(s) iatr~oduc~ by ~r oompoeoat 203.
When 8(s) ~ 0, a purely transvoreal soalizatioa rosulta as illustrated in Tig.
2C, in wltioh the pmacamobcr B(z) of the pmo.equaliaa 105 is zoo and tl~
ac~aplivts z5 f~eod-forward pa~rauuctor A(z) in~rodercxd by $ltar component 203 at signal adder 201 is real, with unity lain by 8,lter eon0.ponent Z04. 'When A(z) a 0 , a purely recorsiwev realizzattioa results as illustrated is Fig. 21), in which the feed-forward parameter A(s) of the pso-eqvalizer 105 is ~m and the Adaptive feedback paraazater B(x) iatroducod by filfier component 202 at s~i~l sadder 201 is real, auoally produaag a unity duct tasa~
doer to 6 29.05.2002 the closed loop. Otherwise, the rurto~ewe waving average filter realist ills is Fig. 2B is achieved.
hi the present iave~adoa, the filter cc~efRcierta for tha real pie-equalize filter 105 aro pcrmittad to chaugc (uadAS adr~ptauar~l with a constraint moving one degree of froodosa, For cxamplc, if tha pTO-cquallzw filioc 105 is pucaly transvorssl:
II(s)~~a"z'".
",..o whey the a" ~e real coe~aier~, flan the simplest ooastraiat is no -1, wlriah implies, lxom equation (1) obovo with B(s) ~ 0 (sirroo tliara ie no irecursivo puruon) that II(s) j. l + A(s) .
Tba meal eoe~l(al, an, ...a", } ~rres free to ba adapted. 'fhia adaptive tsansversal real pre-equali'.er fi Iter system, is ille~strated in Fig. 2C, where .4(s) is adapted and io strictly proper (i.e,, has no throat tea) cad the direct term. of tbo ovoTall filter is oo:gained bo ut~iiy.
More generally, if X ~ Ia~aS..,avbb2...bN]' is tho total roal parametcs spaoc, then one dogreo of ~rcvdom may bo x~araoved by au arbitrary lieiear cvuatrain~
of thn form:
XL~-1, (2) where C is a aot~~zero constant vector of dimension M + N + 1, 8quaxioa (2) abovo reponesernts a hyperplaa~e is the total real parameter space, with t'he only impoitont feoturo being thet the hypcrplaae does not pass through the orlgia. For example, ~3at -2a,1 a 3br -1 is oa~e typiool linear conaisaia~t. Knowledge of hor~r to ~tdap~t under s mo~~a ~ liarx~r constraint of this type is comruoa general lonowledge to those worldag with adaptive ayateuis and is not described in detail hereaa.
La the puroly rcoursive case when A(s) . fl , the most appropriate constraint is no Q 1 which impliss:
H(s) a + B(z) as This alter arrattg~ema,t is i~h~saratea is Fig. 2D, where B(s) is adapted and is etriouy proper (i,e., has no direct te~rni) cad the direct term of the overztl filter ie ixaplicitly oonatroincd to amity.
Adaptntwu of the ramdiaiug parameters (arose not consn~~ined by removal of ono degree of finedOm) may be freely a~da~xl. In tire ps~nt invention, a cost Rmedon corresponding to blfad equalization bosod on energy cost is seleatod for controlling z9.os.2ooz adap~cm of the re~aaiaiag pau~era for sovoral mesons: (r) as au~~y cosrt dcpettds only on the second cmier sratiatias of the oh~ma~al, and will thorofora mchibit quick couvo~~oC:
(ii) tlrc nary cost iS convex end remains so under the imposed linear constraint, vvbicla implies w~oll-bohampd wnvtu'genice: (iii) the ccW: is phase blind, actfag only to oorreot ma,gnitudc distortions iu flea ubxuuel; (iv) the Oost itself is blind sn~i nt2~y therefore bs saaily implennaated without rolyiag on training data; acid (v) the cost ion ci~ectively deal a (only) ~uvith the mini~uum pbaso pordon of tho channel dlstor~iun.
The energ~r coat fimotson to be imized in tho prroacnt iavaotion is glveu by:
,1= ~S (1 x(k) ~ ~ ~ ~'~u~ (k) t u; (k)~ ~ (3) Voa this cost fwnatlon arc poaalblo, auoh as n court c,~umbinelioa of the real and imaginary c~napona~ which subsutsnos the far ri~;t hand portion of opuatiun (3) above ss a ~1 ease. Such vdo not sep~tt oant modifiaotend qualita~vely p~Crrm aimilatly.
In iilus~tra~nB the adaptedion, the simplest conatraiat ~ relented, where no ~
1.
1 S Und~ex the usual stoabaadc gradient deeosnt, the adaptation oqua~o~ booomc a~ (k + 1) - a" (k) - /i.Iu~ U)xr (k ' ~) ~ ~R (k)xR (k - ~)) ('4') sad ~ (k t 1) ~ b~ (k) - ~~ur (~)~r (~ - n) 't' ua (k~uR (1~ - n)1 (S) vYham x(k) is the output oftxra signal adder 201 within the pre-equalixea lU5 illustrated is I~ig. 7.K and xx(k) and x,(k) era the real and im~paary compons~, respoctively, of that output. For the recursive portion, a etandsud gradient approximation is pcxformcd, which aSSUmea certAin contributions from rea~usiv~e terms are zero. Such aspoata ore common g~ral lmowledge to those eidlled is adaptive ayetelmg, as are standard vax:iatioas uwag more couyutn~iunally dctnandin~ lilies rogressor forms that raay aitemativerly be employed.
Ia pcrfomce ro~rum, whcn ac:l3nlt in oo~mCdon with a following ooaventioaol odaptirro oqualinar 107, the prrroqualixor 105 vfibrs the following adVBntages:
(r) adaptstion of pre-eq~liaer 105 is iadep~endenfi of (doaouplcd from) a~dap~ion of coavm~l3onel adariive~ e~quali~r 107; (ii) pre-equalizer 105 ea~hi'bita fast convergence; (iii) pm-uqusliur I05 improves the ooaverger~ce d and robustinesx of the followlug convcntio~aal adaptive equaliser 10?; and (iv) pre-~ualizer 105 providra aigniflcant pe~ormanoo imprvvcm~euts wYith rclatlvely aru~slt impltunontation cost.
Sir~aulationa weze porFo~mod utilizing a co~t~ventionnl lufluiio impulsu~
ravponse ao»RUnt modular adaptive (>IR-CMA) algorithm equali~r, with and without a'V$B

8 29.05.2002 minimum output e»,erglr (MOE) pr~o.equal~xer of the type illustrated is Figs.
2A t'hmugh 2D
and de~ibdd above. The «mv~ioasl adaptive equalizer employed is aapablo of operating in both CMA end decfsfon dixeCted (Db) modex, a~rti~ng in C.MA mode end awltG6ing to DD
mode whoa the mean sqtiaro attar drops below 27% but switching book to CMA, mode if the S moms squ~ro ca~or axeocds 30f6. Beth fioe~d-rurwrard (Fly and ttedbaok (fB) filters were employed iu the VSB-MOE pre-oquolixot.
The aimulai3on was perfononed using the following p~m~etars:
MOE FTi fl Iter length ~ 24 FF Slmr length ~ 64 F8 Ghvr lar~th ~' 300 . 5 IIm-CMA FF filter step s~ ~ 0.0004 (CMA utodo), 0.003 (DD awde) ~t-CMA FS filter atop size = 0.00005 (CMA mode), 0.001 (Db mode) Figs. 3 A, thrrn~gh 3G eo~poad to the simulsti~ with the V~SB-MOE pa~e-equalizer, while Figs. 4A tl~OUgh 4C coaespond to tare cimulatinn without the V~8 MOE p~-equalizer.
Figs. 3 A end 4A depict thd real part of the conventional 1'~'t ~f:M A, equalizar output for a sagn~aut (832 syuibula) priur to (ha qu~antiznr, while Figs. 38 acid 48 depict me ooastcllexion (equalixor aye) diagram of the equalizer output fur fuur sesgmeu~. Figs. 3C and 4C dapioe the beeeband input power speatnm~ to, roapoativcly, the pre-equali~r (for FiB. 3C) 7t) or conventional adaptive ~ (for Fig. 4C), while Figs. 3D and 4D dspiat the baseband output power m fnnm the conventionsl Adaptive equalizer. From thane powe~c spac;trums, the VSB-MOE pre-equalizer can be sin as having whitened the ai~,al.
Fib. 3E anal 4E depict the mean square error (MSE) for the equalizer output oolaulatod for oevoh acgmcat from the last 832 symbols, with tl~ w~sctiwal axial Wing min aquase et'sor pera~sto~go (M88%) sad the horf zoatal aycis rcproscniing tba sogturut anmber. An the ahs~l b~gi~aa ooavargia.6 after 20 segmaata with the Vt98-MOE
pra.equali~r (F'lg. 3h~, but not until approxinoatdy 70 aewithout the VS8-MOB
pra-nyua~z~ (Fig. 4E).
Figs. 3F and 4F illustrate the fea~d-forward ~aguittlde for the two simulations, while Figs. 3G end 4f1 illuatrnxo the foalback filter co~ciunt~.
A coeond aimuiaxfoa was pe~Ormed using the peramotcsa:
MOE FB f lter length ~ 16 FF filter lemgtb ~ 64 ~'H filicr lcagth ~ 300 9 7.y,05.2002 MOE FB flitcr seep sin ~ o.aoos Int-CMA FF filter step siao - 0.000 (CMA niodo), 0,003 (DD mode) IIR-CMA F$ filter sip size - 0.00005 (CMA modo), 0.001 (DD moda) Figs. 5A through SCr Go~rrespo~nd to the simulafiioa with tho VSB-MOE
pro-cqvaliz~et, rwrhiic S Figst. 6A mrough t5G eon~espcmd tn the simulation ~rithout the VSB-MOE
pre~.equatizor. The ooniout of Figs. SA-5Q and 6A-6G coa~esponds to the con4.mt rtf s3nnilarly deaigaabed plots withia Figs. 3A-3O and 4A~. In Ihix ximulation, the oonventional adaptive culaaliu~
errteared DD triode a~ I2 sogonarta when th,e ptn-aguntliur waa~ employed (F1$. 38). v~hile 34 ae~ts wars r~quired beforo tho switch without the pro-equalizor (Fix. 68).
It ~ important to note that while tha prescat f nvos~tion has been doscsibed in the CO~oxt of a fully funeticms L systean, those s~lalled is the art Will apprsoiato tk~t at lenat portions of t'he ateChanism of the present invesnfi~nn era capable of being distributed is the form of a ma~cbluc usable medium vontaininS insnuctioos in a variety of fnnn~, arid that the prosout inventioa applies equally i~agaWlaas of the particular type of signal beating medium t~iliud to aad~ally carry out tlto distrib~n. Lxamplos of maohina usavld uao3ium:c inu:luda:
nonvolatile, hs~rd-aodsd type mediums such as road only momorics (rtOMs) or crasabla, electrtcalty progiammable te~rd only mea (EfiPROMs), neoordsble type modivmau~
scab as floppy diRke, hard digl~ drives and compact disc rrad only mmmorl~t (CD-ROMa) or digital var~atilt~ discs (,D''V'ne), and ~ansmiseion type mediums such ac digital an,d analog communicatloa licks.
Although the pre~t iavv~tion her been des~ib~, in Jeteil. sldllal in the azt will undesctand that various a~hanges, subetitutioxis, variations, cnhanca~aats, auanccs, ~radstio~, lessee foams, char, tewisione, impmvetneoxs and kaodc oz's of tha inv~ioa disclosed herein may be made without departing from the spirit and Scope of the invontioa is 2S ftg 'luniulcat, hum,

Claims (21)

CLAIMS:

1. A pre-equalizer 105 comprising:
- a signal adder 201;
- at least onto adaptive filter 202, 203 coupled to the signal adder 201, an input to the pre-equalizer 105, and an output from the pre-equalizer 105, the at least one adaptive filter 202, 203, 204 operating on one of an input signal r(k) to the pre-equalizer 105 or an output signal u(k) from the pre-equalizer 105 and generating a signal which is added by the signal adder 201 to tho input signal r(k), - wherein adaption of the at least one adaptive filter 202, 203, 204 is derived by minimizing as output energy cost function defined for the pre-equalizer 105, and - wherein the pre-equalizer 105 generates an equalized output signal u(k) from the input r(k) for transmission to an equalizer 107 performing channel equalization.

2. The pre-equalizer 105 according to claim 1, wherein the at least one adaptive filter 202, 203 further comprises:
- a food-forward filter 203 coupled between the signal 201 signal adder and the output from the pre-equalizer 105; and - a feedback filter 202 coupled within a feedback loop between the output from the pre-equalizer 105 and the signal adder 201, - wherein the pre-equalizer 105 is an auto-regressive moving average filter.

3. The pre-equalizer 105 according to claim 1, wherein the at least one adaptive filter 202, 203 further comprises:
- a feed forward filter 203 coupled between the input to the pre-equalizer 105 and the signal adder 201; and - a unity gain filter 204 coupled between the input to the pre-equalizer 105 and the signal adder 201 in parallel with the feed-forward filter 203, - wherein the pre-equalizer 105 is a transversal filter.

4. ~The pre-equalizer 105 according to claim 1, wherein the at least one adaptive filter 202, 203 further comprises a feedback filter 203 coupled within a feedback loop between the output from the pre-equalizer 105 and the signal adder 201, wherein the pre-equalizer 105 is a recursive filter.

5. ~The pre-equalizer 105 according to claim 1, wherein adaptation of the at least one adaptive filters 202, 203 employs a minimum output energy blind criteria algorithm.

6. ~The pre-equalizer 105 according to claim 1, wherein the at least one adaptive filter 202, 203 employs only real coefficients.

7. ~The pre-equalizer 105 according to claim 1, wherein the at least one adaptive filter 202, 203 is a linear filter employing real coefficients and acting on complex symbol spaced input samples to generate complex symbol spaced output samples.

8. ~A receiver 101 comprising:
- as input 102 for receiving an input signal;
- a channel equalizer 107; and - a pre-equalizer 105 coupled between the input 102 and the channel equalizer 107, the pre-equalizer 105 comprising:
- a signal adder 201;
- at least one adaptive filter 202, 203 coupled to the signal adder 201, an input to the pre-equalizer 105, and an output from the pre-equalizer 105, the at least one adaptive filter 202, 203 operating on one of an input signal r(k) to the pre-equalizer 105 or an output signal u(k) from the pre-equalizer 105 end generating a signal which is added by the signal adder 201 to the input signal r(k), - wherein adaptation of the at least one adaptive filter 202, 203 is derived by minimizing an output energy cost function defined for the pro-equalizer 105, and - wherein the pre-equalizer 105 generates an equalized output signal u(k) from the input signal r(k) for transmission to the channel equalizer 107.

9. The receiver 101 according to claim 8, wherein the at least one adaptive filter 202, 203 further comprises:
- a feed-forward filter 203 coupled between the signal adder 201 and the output from the pre-equalizer 105; and - a feedback filter 202 coupled within a feedback loop between the output from the pre-equalizer 105 and the signal adder 201, - wherein the pre-equalizer 105 is an auto-regressive moving average filter.

10. The receiver 101 according to claim 8, wherein the at least one adaptive filter 202, 203 further comprises:
- a feed-forward filter 203 coupled between the input to the pre-equalizer 105 and the signal adder 201; and - a unity gain filter 204 coupled between the input in the pre-equalizer 105 and the signal adder 201 in parallel with the feed-forward filter 203, - wherein the pre-equalizer 105 is a transversal filter.

11. The receiver 101 according to claim 8, wherein the at least one adaptive filter 202, 203 further comprises a feedback filter 202 coupled within a feedback loop between the output from the pre-equalizer 105 and the signal adder 201, wherein the pre-equalizer 105 is a recursive filter.

12. The receiver 101 according to claim 8, wherein adaptation of the at least one adaptive filter 202, 203 employs a minimum output energy blind criterion algorithm.

13. The receiver 101 according to claim 8, wherein the at least one adaptive filter 202, 203 employs only real coefficients.

14. The receiver 101 according to claim 8, wherein the at least one adaptive filter 202, 203 is a linear filter employing real coefficients and acting on complex symbol spaced input samples to generate complex symbol spaced output samples.

15. The method of pre-equalizing a received signal comprising:

- adaptively filtering one of an input signal r(k), an output signal u(k) or both to generate an equalized output signal u(k), wherein the filtering is adapted by minimizing an output energy cost function defined for tho equalized output signal u(k); and - transmitting the equalized output signal u(k) to as equalizer 107 performing channel equalization.

16. The method according to claim 15, wherein the step of adaptively filtering one of an input signal r(k), an output signal u(k) or both to generate an equalized output signal u(k) further comprises:
- an employing a feed-forward filter 203 coupled between a signal adder 201 and an output for the output signal u(k):and - employing a feedback filter 202 coupled within a feedback loop between the output and the signal adder 201, - wherein the feed-forward and feedback filter 202, 203 form an auto-regressive moving average filter.

17. The method according to claim 15, wherein the step of adaptively filtering one of an input signal r(k), an output signal u(k) or both to generate an equalized output signal u(k) further comprises:

- employing a feed-forward filter 203 coupled between an input for the input signal r(k) and a signal adder 201; and - employing a unity gain filter 204 coupled between the input and the signal adder 201 in parallel with feed-forward filter 203, - wherein the feed-forward and unity gain filters 203, 201 form a transversal filter.

18. The method according to claim 15, wherein the step of adaptively filtering one of an input signal r(k), an output signal u(k) or both to generate an equalized output signal u(k) further comprises employing a feedback filter 202 coupled within a feedback loop between the output for the output signal u(k) and a signal adder 201, wherein the feedback filter 202 forms a recursive filter.

19. The method according to claim 15, wherein adaptation of the filtering employs a minimum output energy blind criterion algorithm.

20. The method according to claim 15, wherein tho adaptive filtering employs only real coefficients.

21. An equalized output signal generated by the method as set forth is claim 15.