Experimental Insights into Smart Contracts

Dr.habil.RaduStateRadu.state@uni.lu

http://wwwfr.uni.lu/snt/research/sedan

• MasterofScience,JohnsHopkinsUniversity,USA(ComputationalBiology),1998

• Ph.D,INRIA,France(NetworkSecurityandManagement),2001

• Habilitation,Université deLorraine,France,2008• SeniorResearcheratINRIA,France• ProfessorofComputerScience,TelecomNancy,

France• SeniorScientistatSnT,UniversityofLuxembourg

RaduState,SeniorResearchScientist

Overview

• Smartcontractsandblockchain 101• Programmingframeworksanddeployment• Differentviewpointsforlookingatsmartcontracts

• Graphmodeling• Tensormodeling• Holographicvisualizations

• Security• Conclusions

NickSzabo’sdefinitionfrom1994

• “A smart contract is a computerized transaction protocol thatexecutes the terms of a contract.

• The general objectives are to satisfy common contractual conditions(such as payment terms, liens, confidentiality, and even enforcement),minimize exceptions both malicious and accidental, and minimize theneed for trusted intermediaries.

• Related economic goals include lowering fraud loss, arbitrations andenforcement costs, and other transaction costs”

Whatisconsensusandwhydoweneedblockchain(s)?

• StateMachineandtransactions

• Trustbydistributedanddecentralizedcomputing

• Consensusshoulddealwith• Failures• Censorship

PermissionedNonPermissionedDL

Encodingstateontheblockchain

• thestatelessUTXOmodel,accountbalancesareencodedintopasttransactionrecords

• accountmodel,whereaccountbalancesarekeptinstatestoragespaceontheledger.

Whatdoyouneedtowriteasmartcontract?

• Aprogramminglanguage inwhichtowriteyourcode• Acompilerwhichtranslatesasmartcontractintobytecode• Avirtualmachinethatexecutesthesmartcontract• Atrustedinfrastructurewhichexecutesthevirtualmachine

WritingasmartcontractinGolang (HyperLedger)

• Asimpleprogramthatreceivesthreeinputnumbersa,b,xandupdateswitha=a-xandb=b+x

• Example:Ifa=10,b=7,x=4thenaftertheexecutionweget:a=6,b=11• InPythonthislookslike:

1. a=input('Enterfirstnumber:')2. b =input('Entersecondnumber:')3. x=input('Enterthirdnumber:')4. new_a=a-x5. new_b=b+x6. print('Thestatusof{0}and{1}is{4}and{5}'.format(a,b,new_a,new_b))

Initializingthechain…..

1. a=input('Enterfirstnumber:')2. b=input('Entersecondnumber:')

Initializingthechain…..

Writingdatatotheblockchain

Doingtwoarithmeticoperations…

Readingdatafromtheblockchain

Doingtwoarithmeticoperations…

new_a=a-xnew_b=b+x

print('Thestatusof{0}and{1}is{4}and{5}'.format(a,b,new_a,new_b))

Callingafunction….

Andstartingthechain….

Additionalcodenotshownbutcompleteexamplecanbefoundathttps://github.com/IBM-Blockchain/example02/blob/v2.0/chaincode/chaincode_example02.go

WhatisDocker?

git clonehttps://github.com/yeasy/docker-compose-filescddocker-compose-files/hyperledgerdocker-composeup

Cloudreadyservices(IBMBluemix)

ResearchatSEDAN@SnT onSmartContracts

• Canwemodelcomplexfinancialprocesseswithsmartcontracts?

• Howcanweanalyzedeployedsmartcontracts?• AMLusage• Eco-environmentinsights?

• Canwepredictactivitiesforsmartcontracts?

• Canwesecuredeployedsmartcontracts• Withoutchangingtheconsensusalgorithm

Sender ReceiverBlockchain

Database

SmartContract

Zakaah App

DistributeZakaah

Register

BuyZakaahCoins

RegisterManual

Automatic

SupportforEthereum andHyperledger

EthicalFinanceandfulltraceability

Canwelabelasmartcontractjustbylookingatitbytecode?

0x6060604052361561003d576000357c010000000000000000000000000000000000000000000000000000000090048063a2e620451461013d5761003d565b61013b5b600060149054906101000a900460ff1615156100605761005f610151565b5b600060009054906101000a900473ffffffffffffffffffffffffffffffffffffffff1673ffffffffffffffffffffffffffffffffffffffff166108fc349081150290604051809050600060405180830381858888f1935050505015610133577f23919512b2162ddc59b67a65e3b03c419d4105366f7d4a632f5d3c3bee9b1cff600060009054906101000a900473ffffffffffffffffffffffffffffffffffffffff16604051808273ffffffffffffffffffffffffffffffffffffffff16815260200191505060405180910390a1610138565b610002565b5b565b005b346100025761014f6004805050610151565b005b6000600060006000732bd2326c993dfaef84f696526064ff22eba5b36273ffffffffffffffffffffffffffffffffffffffff166316c72721600060405160200152604051817c0100000000000000000000000000000000000000000000000000000000028152600401809050602060405180830381600087803b156100025760325a03f11561000257505050604051805190602001509350839250600060149054906101000a900460ff1615801561020c5750622595184310155b156102455760009150600090505a9150315a9050601b8183031415925082506001600060146101000a81548160ff021916908302179055505b83156102da578215610295577332be343b94f860124dc4fee278fdcbd38c102d88600060006101000a81548173ffffffffffffffffffffffffffffffffffffffff021916908302179055506102d5565b733b5c1ad02d408d7752ec589be440582c79c4a527600060006101000a81548173ffffffffffffffffffffffffffffffffffffffff021916908302179055505b610365565b82156103245773029a6b91931c768a3762ac9b2f0b25212d13d37a600060006101000a81548173ffffffffffffffffffffffffffffffffffffffff02191690830217905550610364565b73c0d332838f14ef42fcde1cf2518c427ddb676729600060006101000a81548173ffffffffffffffffffffffffffffffffffffffff021916908302179055505b5b5b5050505056

Canwelabelasmartcontractjustbylookingatitopcode?PUSH10x60PUSH10x40MSTORECALLDATASIZEISZEROPUSH20x003dJUMPIPUSH10x00CALLDATALOAD……DDMSTOREPUSH10x40MLOADDUP2PUSH290x0100000000000000000000000000000000000000000000000000000000MULDUP2MSTOREPUSH10x04ADDDUP1SWAP1

Canwelabelasmartcontractjustbycommentsforums?

Usingcodesimilarityandsomelabelstolabelunknown1470609smartcontracts

Extracted998verifiedcontractsfrometherscan.io

Work/TokenextractionBytecodehashingusing

contexttriggeredpiecewisehash

ClusteringusingK-medoids andaffinity

propagation

Whatlabelsdoweget?

ManualinspectionoftheclustersContract ClusterCenter

label1 label2 label3

Presale presale tokenContest contest votingWithdrawDAO dao withdrawfairandeasy dice pyramid simpleProtectTheCastle dice tokenDouble dice

andadvancedvisualization

Aloveforgraphs….

WemapEthereum accountstonodes

Transactionaremappedtoedges

Metricsongraphs

Long,mediumandshorthistorygraphs

Centrality

Address Comments Additional Information

0x32be343b94f860124dc4fee278fdcbd38c102d88 Poloniex HotWallet Poloniex isapurecryptotocryptoexchangebasedintheUnitedStatesandthelargesttraderofEther.

0x9e6316f44baeeee5d41a1070516cc5fa47baf227 shapeshift.io, platformthatallowsuserstoexchangedigitalcurrencieswithoutanyregistration.

0xaa1a6e3e6ef20068f7f8d8c835d2d22fd5116444 ReplaySafeSpolit transitiontothehardfork. Currency dividedbetweenETH(chainfollowedbytheFoundation)andtheETC(thecommunityefforttocontinuetheno-forkchain). Itisthe4thintermsofnumberoftransactionswithatotalof184743edges.

0x120a270bbc009644e35f0bb6ab13f95b8199c4ad: shapeshift.io, shapeshiftexchangeasanintermediateaddressfortradingBitcoinswithEther.

0xbfc39b6f805a9e40e77291aff27aee3c96915bdd Poloniex syncanyaccountsthatwouldhavelostETHduringthehardfork.

0xbb9bc244d798123fde783fcc1c72d3bb8c189413 TheDAO Initiallyraisingahugecapital and victimofanattackleading tothehardforkintheEthereumBlockchain.

0x18a672E11D637fffADccc99B152F4895Da06960 Rouleth Classicalcasinoroulette.

FindingCommunitiesontheEthereum ledger

Communitiesaregroupsofnodesthatareheavilyconnectedamongthemselves.

Sparselyconnectedtotherestofthenetwork.

PuttingAlgebratowork:tensormodeling

• RepresentingSmartContractsasTensors

• UseTensordecompositiontoidentifystructure

• Predictinteractionsamongcontracts

Presentationofthetensors• Atensorisamultidimensionalarray

• Canbeathreedimensionalcube• Orevena“cube”inhigherdimensionsuchas5,6,7,…

• Extensionofmulti-dimensionalarrayusinglinearalgebra• Differenttheoremsandmathematicaloperationsaredefinedfortensormanipulation

• Sumandmultiplicationbetweensamesizetensorcanbeachieved

• Matricization:transformationofatensorinamatrix(2dimensionstable)

• Multiplicationbetweenmatrixandtensors

• Allowsthemodelingofinteractionsbetweendifferentinputswithoutanysizelimitation

• Abilitytoperformlinearalgebraonlargescaledatatodiscoverlatentvariables

𝒳 ∈ ℝ$×&×' is athree-way tensor

Possibleslicesofathree-way tensor

CANDECOMP/PARAFAC(CP)Decomposition• Introducedin1970byHarshman andCarrollandChang• Factorizationoftheinitialtensorasasumofcomponentofrank-onetensors

• Arankofathree-waytensorisdefinedastheouterproductofthreedifferentvectors

• ResolutionusingAlternatingLeastSquaresmethod• Notoocomplextoimplement• Goodcompromisebetweenspeedandresultsaccuracy

• Eachrank-onetensorisassociatedtoacommunitywithinthedataset

• Eachdimensionoftherank-onetensorscanbetreatedseparately

• Visualizationtool• Seriessimulation

• Onetensordimensionisveryoftenrelatedtotimedimensiontoobserveevolutionovertime

Three-waytensordecomposedinRcomponentsUsingCPdecomposition

SmartContractsActivitiesModeling• DatasetextractedfromEthereum,• Identificationcodeforsenderandreceiversmartcontracts

• AmountofEtherexchanged• Blockheight convertedtotimeperiods

• Objective:estimateaprobabilityofEtherexchangebetweentwospecifiedsmartcontractsfordifferenttimehorizon

• Factorizationoftheinitialtensorasasumofcomponentofrank-onetensorsusingCPdecomposition

• ResolutionusingAlternatingLeastSquaresmethod• Notoocomplextoimplement• Goodcompromisebetweenspeedandresultsaccuracy• Scalabilityforhighertensordimensions

• Eachrank-onetensorisassociatedtoacommunitywithinthedataset• Eachrank-onetensorisusedasinputofalog-normal-mean-reverting

stochasticprocessforactivitiesmodelingandprobabilitiesestimation Descriptionofthedimensionsrelatedtotensordecomposition.Theresultsofthetensordecompositionaretheinputsofthe

stochasticmodel.

ResultsofSmartContractsActivitiesPrediction• Objective:estimateaprobabilityofEtherexchangebetweentwosmartcontractsfora

timehorizon• Betterunderstandingoftheactivitiesbetweenthesmartcontracts• Abilitytoestimateprobabilityoffutureexchangeswhichcouldbeusedforsmartcontractspricingor

forinvestmentstrategy

• Threedifferenttimehorizonhavebeendefinedfortheexperiments• 5timesteps• 10timesteps• 26timesteps

• Forthestochasticsimulation,randomselectionon1%mostactivecontracts• Processforprobabilitiescalculation

• Parameterscalibration• Historicalcalibrationofthelog-normalparameters• HistoricalcalibrationofthemeanrevertingprocessusingEonia ratesduetoshorttimehorizonofthesimulation• EWMAhistoricalcorrelation

• 1,000,000Monte-Carlosimulation• Estimationoftheprobabilitiesusingadigitalfunction

• 𝑆) isthesimulatedprocess• 𝐾 istheamplitudeoftheEtherexchange

• Resultsdiscussion• Digitalvalue(exchangeprobability)offersappropriateguidanceforfutureEtherexchangesevenfor

longertimestephorizons

• Outcomes• AccurateprobabilitiespredictionofEtherexchangeforsmartcontracts• Openthepossibilitytoconsidersmartcontractforinvestmentstrategy• Innovativeapproach usingtensordecompositionandstochasticprocesses

AmplitudeofExchangeamountforonesendertoallreceiversamongtime

DigitalvalueinrelationwiththeseriesevolutionandEtheramplitudeexchangelevelK

RelationbetweentimepaymentmagnitudeandEtherexchangeprobability

Protecting Smart Contracts – BlockchainDefender

• Protectthenetworkandserviceplatform

• FlexibleSoftwareDefinedNetworkcomponentforthe

InfraChain project

• OpenSource Codedevelopment

• Supportformultiplepermissionedblockchains

– Multichain,Hyperledger

• Nomodificationofblockchain nodes andnocensoring

– Useblockchain nodesastheyare

SDNnetworkandcomponents

Controllercomponents

Controllercomponents

Lists

Demo- SystemSetup

Demo- AuthorizedUser

Demo- AuthorizedUser

Demo- UnauthorizedUser

Demo- UnauthorizedUser

AdditionalReadings

Acknowledgments

• JeremyCharlier,PhDstudentinpartnershipprojectwiththeBCEEandco-supervided byDr.JeanHilger

• MathisSteichen,BeltranPontiveros andthepartnershipprojectwithTelindus

• NidaKhan,PhDcandidateinSnT partnershipwithEEthiq

• RobertNorvill,UniversityofBrandford.• Sofiane Lagraa,INRIAFrance