/Users/andrea/_magisterarbeit/korpus/clean/testkorpus/1/file1.html NN ----------------------------------------- : Download NP a DT Postscript NN or CC PDF NN version NN of IN this DT paper NN . SENT Download RB all PDT the DT files NNS for IN this DT paper NN as IN a DT gzipped JJ tar NN archive NP . SENT Generate VV another DT one NN . SENT Back RB to TO the DT SCIgen NP homepage NN . SENT The DT Effect NN of IN Multimodal NP Technology NP on IN Operating NP Systems NPS The DT Effect NN of IN Multimodal NP Technology NP on IN Operating NP Systems NPS Marschner NP C NP . SENT Abstract JJ The DT synthesis NN of IN kernels NNS has VHZ deployed VVN flip JJ flop NN gates NNS , , and CC current JJ trends NNS suggest VVP that IN the DT evaluation NN of IN object NN oriented JJ languages NNS will MD soon RB emerge VV . SENT In IN fact NN , , few JJ steganographers NNS would MD disagree VV with IN the DT refinement NN of IN replication NN , , which WDT embodies VVZ the DT significant JJ principles NNS of IN electrical JJ engineering NN . SENT Our PP$ focus NN in IN this DT paper NN is VBZ not RB on IN whether IN forward JJ error NN correction NN can MD be VB made VVN wireless NN , , wearable NN , , and CC interposable JJ , , but CC rather RB on IN exploring VVG a DT reliable JJ tool NN for IN refining NN digital JJ to TO analog JJ converters NNS ALPHOL NN . SENT Table NN of IN Contents NNS 1 CD Introduction NN 2 CD Architecture NP 3 CD Bayesian NP Symmetries NNS 4 CD Evaluation NN and CC Performance NN Results NNS 4 CD . SENT 1 CD Hardware NN and CC Software NP Configuration NN 4 CD . SENT 2 CD Dogfooding NP ALPHOL NP 5 CD Related NP Work NN 6 CD Conclusions NNS 1 CD Introduction NN Self NP learning VVG models NNS and CC the DT lookaside NN buffer NN have VHP garnered VVN great JJ interest NN from IN both DT leading JJ analysts NNS and CC experts NNS in IN the DT last JJ several JJ years NNS . SENT To TO put VV this DT in IN perspective NN , , consider VVP the DT fact NN that IN foremost JJ researchers NNS entirely RB use VVP SCSI NP disks NNS to TO accomplish VV this DT ambition NN . SENT The DT impact NN on IN steganography NN of IN this DT discussion NN has VHZ been VBN well RB received VVN . SENT To TO what WP extent NN can MD model VV checking NN be VB investigated VVN to TO achieve VV this DT goal NN . SENT In IN this DT work NN we PP demonstrate VVP that IN consistent JJ hashing VVG and CC redundancy NN are VBP never RB incompatible JJ . SENT In IN addition NN , , indeed RB , , online JJ algorithms NNS 15 CD , , 12 CD and CC public JJ private JJ key JJ pairs NNS have VHP a DT long JJ history NN of IN interfering VVG in IN this DT manner NN . SENT For IN example NN , , many JJ systems NNS locate VVP web NN browsers NNS . SENT It PP should MD be VB noted VVN that IN our PP$ system NN requests NNS kernels NNS . SENT Therefore RB , , we PP see VVP no DT reason NN not RB to TO use VV replicated VVN algorithms NNS to TO explore VV concurrent JJ archetypes NNS . SENT Even RB though IN it PP might MD seem VV perverse JJ , , it PP is VBZ derived VVN from IN known JJ results NNS . SENT The DT rest NN of IN the DT paper NN proceeds NNS as RB follows VVZ . SENT To TO start VV off RP with IN , , we PP motivate VVP the DT need NN for IN public JJ private JJ key JJ pairs NNS . SENT Along IN these DT same JJ lines NNS , , to TO achieve VV this DT mission NN , , we PP demonstrate VVP not RB only RB that IN compilers NNS can MD be VB made VVN interactive JJ , , optimal JJ , , and CC constant JJ time NN , , but CC that IN the DT same JJ is VBZ true JJ for IN B NP trees NNS . SENT Next RB , , we PP place VVP our PP$ work NN in IN context NN with IN the DT related JJ work NN in IN this DT area NN . SENT Finally RB , , we PP conclude VVP . SENT 2 CD Architecture NP The DT properties NNS of IN our PP$ methodology NN depend VV greatly RB on IN the DT assumptions NNS inherent JJ in IN our PP$ design NN . SENT in IN this DT section NN , , we PP outline VV those DT assumptions NNS . SENT Further RBR , , the DT methodology NN for IN ALPHOL NP consists VVZ of IN four CD independent JJ components NNS . SENT scalable JJ configurations NNS , , the DT structured JJ unification NN of IN XML NP and CC consistent JJ hashing VVG , , robots NNS , , and CC Internet NP QoS NP . SENT This DT may MD or CC may MD not RB actually RB hold VV in IN reality NN . SENT We PP estimate VVP that DT Byzantine JJ fault NN tolerance NN can MD manage VV stable JJ algorithms NNS without IN needing VVG to TO deploy VV extreme JJ programming NN . SENT Further RBR , , we PP assume VVP that IN each DT component NN of IN our PP$ heuristic JJ deploys VVZ thin JJ clients NNS , , independent JJ of IN all DT other JJ components NNS . SENT This DT seems VVZ to TO hold VV in IN most JJS cases NNS . SENT See VV our PP$ prior JJ technical JJ report NN 2 CD for IN details NNS . SENT Figure NN 1 CD . SENT The DT flowchart NN used VVN by IN our PP$ framework NN . SENT Reality NN aside RB , , we PP would MD like VV to TO evaluate VV an DT architecture NN for IN how WRB our PP$ methodology NN might MD behave VV in IN theory NN . SENT Even RB though IN steganographers NNS always RB assume VVP the DT exact JJ opposite NN , , ALPHOL NP depends VVZ on IN this DT property NN for IN correct JJ behavior NN . SENT Rather RB than IN managing VVG large JJ scale NN technology NN , , our PP$ application NN chooses VVZ to TO evaluate VV the DT study NN of IN XML NP . SENT Furthermore RB , , we PP consider VVP an DT application NN consisting VVG of IN n NN superblocks NNS . SENT Rather RB than IN locating VVG client NN server NN symmetries NNS , , ALPHOL NP chooses VVZ to TO store VV large JJ scale NN configurations NNS . SENT This DT may MD or CC may MD not RB actually RB hold VV in IN reality NN . SENT See VV our PP$ related JJ technical JJ report NN 2 CD for IN details NNS . SENT 3 CD Bayesian JJ Symmetries NNS Our PP$ implementation NN of IN our PP$ methodology NN is VBZ electronic JJ , , homogeneous JJ , , and CC stochastic JJ . SENT Next RB , , theorists NNS have VHP complete JJ control NN over IN the DT client NN side NN library NN , , which WDT of IN course NN is VBZ necessary JJ so RB that IN the DT partition NN table NN and CC IPv NP 6 CD can MD connect VV to TO accomplish VV this DT purpose NN . SENT System NN administrators NNS have VHP complete JJ control NN over IN the DT virtual JJ machine NN monitor NN , , which WDT of IN course NN is VBZ necessary JJ so RB that IN write VV back JJ caches NNS and CC the DT lookaside NN buffer NN can MD agree VV to TO accomplish VV this DT goal NN . SENT 4 CD Evaluation NN and CC Performance NN Results NNS Our PP$ evaluation NN represents VVZ a DT valuable JJ research NN contribution NN in IN and CC of IN itself PP . SENT Our PP$ overall JJ performance NN analysis NN seeks VVZ to TO prove VV three CD hypotheses NNS . SENT 1 CD that IN digital JJ to TO analog JJ converters NNS have VHP actually RB shown VVN improved JJ latency NN over IN time NN . SENT 2 CD that IN local JJ area NN networks NNS no RB longer RB influence VV NV NP RAM NN space NN . SENT and CC finally RB 3 CD that IN the DT LISP NN machine NN of IN yesteryear NN actually RB exhibits VVZ better RBR effective JJ clock NN speed NN than IN today's JJ hardware NN . SENT Our PP$ logic NN follows VVZ a DT new JJ model NN . SENT performance NN might MD cause VV us PP to TO lose VV sleep VV only RB as RB long JJ as IN performance NN constraints NNS take VVP a DT back JJ seat NN to TO clock NN speed NN . SENT The DT reason NN for IN this DT is VBZ that WDT studies NNS have VHP shown VVN that DT bandwidth NN is VBZ roughly RB 94 CD higher JJR than IN we PP might MD expect VV 14 CD . SENT Further RBR , , only RB with IN the DT benefit NN of IN our PP$ system's JJ hard JJ disk NN space NN might MD we PP optimize VV for IN scalability NN at IN the DT cost NN of IN complexity NN constraints NNS . SENT Our PP$ work NN in IN this DT regard NN is VBZ a DT novel NN contribution NN , , in IN and CC of IN itself PP . SENT 4 LS . SENT 1 CD Hardware NN and CC Software NP Configuration NN Figure NP 2 CD . SENT The DT average JJ distance NN of IN our PP$ method NN , , as IN a DT function NN of IN distance NN . SENT A DT well RB tuned VVN network NN setup NN holds VVZ the DT key NN to TO an DT useful JJ performance NN analysis NN . SENT We PP carried VVD out RP a DT deployment NN on IN CERN's NP system NN to TO quantify VV decentralized JJ modalities's NNS influence VVP on IN Dennis NP Ritchie's NP visualization NN of IN RAID NN in IN 1977 CD . SENT This DT configuration NN step NN was VBD time NN consuming NN but CC worth IN it PP in IN the DT end NN . SENT We PP removed VVD 8 CD FPUs NN from IN our PP$ mobile JJ telephones NNS . SENT Furthermore RB , , we PP removed VVD 25 CD 150 CD kB NN hard JJ disks NNS from IN the DT KGB's NP system NN . SENT Next RB , , we PP tripled VVD the DT mean JJ clock NN speed NN of IN CERN's NP desktop NN machines NNS to TO consider VV modalities NNS . SENT Finally RB , , we PP halved VVD the DT clock NN speed NN of IN the DT KGB's NP concurrent JJ testbed NN to TO probe NN algorithms NNS . SENT Figure NN 3 CD . SENT The DT effective JJ hit NN ratio NN of IN ALPHOL NP , , as IN a DT function NN of IN block NN size NN . SENT We PP ran VVD ALPHOL NN on IN commodity NN operating NN systems NNS , , such JJ as IN Ultrix NP Version NP 9 CD . SENT 7 CD , , Service NP Pack NP 7 CD and CC Ultrix NP . SENT All DT software NN components NNS were VBD hand RB assembled VVN using VVG Microsoft NP developer's NNS studio NN with IN the DT help NN of IN T NN . SENT Sasaki's NP libraries NNS for IN provably RB constructing VVG fuzzy JJ public JJ private JJ key JJ pairs NNS . SENT Such PDT a DT hypothesis NN might MD seem VV unexpected JJ but CC never RB conflicts NNS with IN the DT need NN to TO provide VV Boolean NP logic NN to TO biologists NNS . SENT We PP implemented VVD our PP$ simulated JJ annealing VVG server NN in IN B NP , , augmented VVN with IN collectively RB exhaustive JJ extensions NNS . SENT We PP made VVD all DT of IN our PP$ software NN is VBZ available JJ under IN an DT UCSD NP license NN . SENT Figure NN 4 CD . SENT The DT effective JJ energy NN of IN ALPHOL NP , , compared VVN with IN the DT other JJ methods NNS . SENT 4 LS . SENT 2 CD Dogfooding NP ALPHOL NP Figure NP 5 CD . SENT The DT effective JJ complexity NN of IN ALPHOL NP , , as IN a DT function NN of IN block NN size NN . SENT Our PP$ hardware NN and CC software NN modficiations NNS make VVP manifest JJ that IN emulating VVG ALPHOL NN is VBZ one CD thing NN , , but CC deploying VVG it PP in IN a DT chaotic JJ spatio NN temporal JJ environment NN is VBZ a DT completely RB different JJ story NN . SENT With IN these DT considerations NNS in IN mind NN , , we PP ran VVD four CD novel NN experiments NNS . SENT 1 LS we PP measured VVN hard JJ disk NN space NN as IN a DT function NN of IN flash NN memory NN space NN on IN an DT Atari NP 2600 CD . SENT 2 CD we PP measured VVN NV NP RAM NN speed NN as IN a DT function NN of IN ROM NP speed NN on IN a DT Nintendo NP Gameboy NP . SENT 3 LS we PP deployed VVN 61 CD Macintosh NP SEs NNS across IN the DT 100 CD node NN network NN , , and CC tested VVD our PP$ hash NN tables NNS accordingly RB . SENT and CC 4 CD we PP compared VVN power NN on IN the DT ErOS NP , , Sprite NN and CC Microsoft NP Windows NNS NT JJ operating NN systems NNS . SENT We PP discarded VVD the DT results NNS of IN some DT earlier JJR experiments NNS , , notably RB when WRB we PP dogfooded VVD our PP$ algorithm NN on IN our PP$ own JJ desktop NN machines NNS , , paying VVG particular JJ attention NN to TO RAM VV space NN . SENT Now RB for IN the DT climactic JJ analysis NN of IN experiments NNS 1 CD and CC 4 CD enumerated VVN above IN . SENT Note VV the DT heavy JJ tail NN on IN the DT CDF NP in IN Figure NP 2 CD , , exhibiting VVG exaggerated JJ response NN time NN . SENT Though IN this DT might MD seem VV unexpected JJ , , it PP fell VVD in IN line NN with IN our PP$ expectations NNS . SENT Along IN these DT same JJ lines NNS , , operator NN error NN alone RB cannot JJ account NN for IN these DT results NNS . SENT On IN a DT similar JJ note NN , , the DT data NN in IN Figure NP 4 CD , , in IN particular JJ , , proves VVZ that IN four CD years NNS of IN hard JJ work NN were VBD wasted VVN on IN this DT project NN . SENT Shown VVN in IN Figure NP 3 CD , , all RB four CD experiments NNS call VVP attention NN to TO ALPHOL's NP signal NN to TO noise NN ratio NN . SENT The DT curve NN in IN Figure NP 4 CD should MD look VV familiar JJ . SENT it PP is VBZ better RBR known VVN as IN G NP 1 CD n NN n NN n NN . SENT Operator NN error NN alone RB cannot JJ account NN for IN these DT results NNS . SENT Of IN course NN , , all DT sensitive JJ data NNS was VBD anonymized VVN during IN our PP$ middleware NN simulation NN . SENT Lastly RB , , we PP discuss VVP experiments NNS 1 CD and CC 4 CD enumerated VVN above IN . SENT The DT key NN to TO Figure NP 2 CD is VBZ closing VVG the DT feedback NN loop NN . SENT Figure NN 2 CD shows NNS how WRB ALPHOL's NP effective JJ RAM NN space NN does VVZ not RB converge VV otherwise RB . SENT Of IN course NN , , all DT sensitive JJ data NNS was VBD anonymized VVN during IN our PP$ software NN deployment NN . SENT Along IN these DT same JJ lines NNS , , Gaussian JJ electromagnetic JJ disturbances NNS in IN our PP$ network NN caused VVD unstable JJ experimental JJ results NNS . SENT 5 CD Related JJ Work NN In IN this DT section NN , , we PP consider VVP alternative JJ systems NNS as RB well RB as IN existing VVG work NN . SENT The DT much RB touted VVN methodology NN 11 CD does VVZ not RB analyze VV the DT exploration NN of IN vacuum NN tubes NNS as RB well RB as IN our PP$ method NN . SENT Without IN using VVG ubiquitous JJ epistemologies NNS , , it PP is VBZ hard JJ to TO imagine VV that IN the DT memory NN bus NN and CC robots NNS can MD connect VV to TO overcome VV this DT issue NN . SENT Similarly RB , , J NP . SENT H NP . SENT Wilkinson NP et NP al NP . SENT 15 CD , , 11 CD suggested VVD a DT scheme NN for IN evaluating VVG read VVP write VV communication NN , , but CC did VVD not RB fully RB realize VV the DT implications NNS of IN write VV back RB caches NNS at IN the DT time NN 8 CD . SENT This DT is VBZ arguably RB fair JJ . SENT Unlike IN many JJ prior JJ methods NNS 9 CD , , we PP do VVP not RB attempt VV to TO synthesize VV or CC synthesize VV the DT analysis NN of IN scatter NN gather VV I PP O NP 7 CD , , 5 CD . SENT Nevertheless RB , , these DT approaches NNS are VBP entirely RB orthogonal JJ to TO our PP$ efforts NNS . SENT The DT deployment NN of IN the DT emulation NN of IN the DT Turing NP machine NN has VHZ been VBN widely RB studied VVN 11 CD . SENT On IN the DT other JJ hand NN , , the DT complexity NN of IN their PP$ solution NN grows VVZ quadratically RB as IN the DT construction NN of IN write VV back JJ caches NNS grows VVZ . SENT Along IN these DT same JJ lines NNS , , Thompson NP introduced VVD several JJ trainable JJ solutions NNS , , and CC reported VVD that IN they PP have VHP minimal JJ inability NN to TO effect VV A DT search NN 6 CD , , 13 CD . SENT Similarly RB , , Thompson NP et CC al NP . SENT originally RB articulated VVN the DT need NN for IN the DT unfortunate JJ unification NN of IN 802 CD . SENT 11 CD b SYM and CC IPv NP 7 CD 3 CD , , 10 CD . SENT A DT recent JJ unpublished JJ undergraduate NN dissertation NN constructed VVN a DT similar JJ idea NN for IN Lamport NP clocks NNS 1 CD , , 14 CD . SENT Unlike IN many JJ prior JJ methods NNS 4 CD , , we PP do VVP not RB attempt VV to TO manage VV or CC measure VV multi NNS processors NNS 3 CD , , 8 CD . SENT Unfortunately RB , , without IN concrete JJ evidence NN , , there EX is VBZ no DT reason NN to TO believe VV these DT claims NNS . SENT We PP plan VVP to TO adopt VV many JJ of IN the DT ideas NNS from IN this DT prior JJ work NN in IN future JJ versions NNS of IN our PP$ heuristic JJ . SENT 6 CD Conclusions NNS In IN our PP$ research NN we PP validated VVD that DT simulated JJ annealing VVG and CC the DT transistor NN are VBP largely RB incompatible JJ . SENT Similarly RB , , in IN fact NN , , the DT main JJ contribution NN of IN our PP$ work NN is VBZ that IN we PP used VVD self NN learning VVG symmetries NNS to TO disprove VV that IN the DT acclaimed JJ read NN write VV algorithm NN for IN the DT study NN of IN fiber NN optic NN cables NNS by IN W NP . SENT Smith NP et NP al NP . SENT follows VVZ a DT Zipf NP like IN distribution NN . SENT We PP explored VVD a DT collaborative JJ tool NN for IN constructing VVG sensor NN networks NNS ALPHOL NP , , which WDT we PP used VVD to TO verify VV that IN the DT location NN identity NN split NN can MD be VB made VVN linear JJ time NN , , introspective JJ , , and CC stochastic JJ . SENT We PP proposed VVD an DT analysis NN of IN forward JJ error NN correction NN ALPHOL NN , , validating VVG that IN I PP O NN automata NN can MD be VB made VVN classical JJ , , virtual JJ , , and CC wireless NN . SENT Our PP$ application NN is VBZ not RB able JJ to TO successfully RB request VV many JJ B NN trees NNS at IN once RB . SENT References NNS 1 CD Bose NP , , I PP . SENT , , and CC Li NP , , T NN . SENT DedeIckle NP . SENT Improvement NN of IN operating VVG systems NNS . SENT In IN POT NN OOPSLA NP June NP 1996 CD . SENT 2 CD Daubechies NNS , , I PP . SENT Hob NP . SENT A DT methodology NN for IN the DT emulation NN of IN Lamport NP clocks NNS . SENT Journal NP of IN Event NP Driven VVN , , Secure VV , , Self NP Learning NP Models NNS 57 CD Jan NP . SENT 1992 CD , , 77 CD 84 CD . SENT 3 CD Harris NP , , O NP . SENT Investigation NN of IN von NP Neumann NP machines NNS . SENT In IN POT NN the DT Symposium NN on IN Metamorphic JJ , , Heterogeneous JJ Information NP Nov NP . SENT 1999 CD . SENT 4 CD Ito NP , , C NP . SENT , , Raman NP , , W NP . SENT , , Nygaard NP , , K NP . SENT , , Watanabe NP , , X NN . SENT , , Lamport NP , , L NP . SENT , , and CC Einstein NP , , A NP . SENT The DT impact NN of IN decentralized JJ information NN on IN complexity NN theory NN . SENT Journal NP of IN Adaptive JJ Communication NP 84 CD July NP 2002 CD , , 1 CD 15 CD . SENT 5 CD Kobayashi NP , , R NN . SENT , , Maruyama NP , , a DT . SENT , , and CC Chomsky NP , , N NP . SENT Cize NP . SENT Low JJ energy NN , , modular JJ theory NN . SENT Journal NP of IN Secure JJ , , Cacheable JJ Models NNS 8 CD Apr NN . SENT 1990 CD , , 20 CD 24 CD . SENT 6 CD Kumar NP , , F NN . SENT , , Floyd NP , , R NN . SENT , , Muthukrishnan NP , , L NP . SENT , , Thompson NP , , O NN . SENT , , and CC Davis NP , , B NP . SENT NobOrgy NP . SENT Symbiotic JJ symmetries NNS . SENT Journal NP of IN Symbiotic JJ Algorithms NNS 418 CD Sept NP . SENT 2003 CD , , 75 CD 97 CD . SENT 7 CD Lamport NP , , L NP . SENT An DT analysis NN of IN RAID NN . SENT In IN POT NN the DT Symposium NN on IN Interposable JJ , , Efficient JJ Communication NP May NP 2003 CD . SENT 8 CD Lee NP , , N NP . SENT , , and CC Wilkes NP , , M NP . SENT V NN . SENT A DT case NN for IN superpages NNS . SENT In IN POT NN SOSP NP Nov NP . SENT 1996 CD . SENT 9 CD Martin NP , , W NP . SENT M NP . SENT A DT methodology NN for IN the DT development NN of IN B NP Trees NP . SENT Journal NP of IN Automated VVN Reasoning NN 1 CD Nov NP . SENT 1998 CD , , 59 CD 66 CD . SENT 10 CD Nygaard NP , , K NP . SENT , , and CC Jones NP , , X NP . SENT Comparing VVG the DT lookaside NN buffer NN and CC model NN checking NN using VVG EtheNog NP . SENT In IN POT NN PLDI NP Apr NP . SENT 2003 CD . SENT 11 CD Smith NP , , Y NP . SENT Comparing VVG 16 CD bit NN architectures NNS and CC e NN commerce NN using VVG EBB NN . SENT In IN POT NN the DT USENIX NP Security NP Conference NP Oct NP . SENT 1999 CD . SENT 12 CD Tarjan NP , , R NP . SENT Deconstructing NP Internet NP QoS NP using VVG Trinket NN . SENT Journal NP of IN Virtual JJ Archetypes NNS 54 CD Apr NN . SENT 1998 CD , , 51 CD 66 CD . SENT 13 CD Wilkinson NP , , J NP . SENT The DT location NN identity NN split NN considered VVD harmful JJ . SENT In IN POT NN the DT Conference NN on IN Adaptive JJ Archetypes NNS Apr NN . SENT 1994 CD . SENT 14 CD Wirth NP , , N NP . SENT Exploring VVG superblocks NNS using VVG encrypted VVN theory NN . SENT In IN POT NN IPTPS NP Oct NP . SENT 2001 CD . SENT 15 CD Wirth NP , , N NP . SENT , , and CC Zheng NP , , I NP . SENT Visualizing VVG flip JJ flop NN gates NNS using VVG pervasive JJ technology NN . SENT Journal NP of IN Multimodal NP , , Embedded VVN Methodologies NNS 0 CD Mar NP . SENT 1995 CD , , 84 CD 104 CD . SENT