Research Article Temporal Index Scheme Of Hyperledger Fabric System In IoT

4Wireless Communications and Mobile ComputingTx1:[.]Tx2:[.]Tx3:[.].Transactionsk, 𝜃𝜀(k, 𝜃){value(1), value(2),., value(n)}kvalue(n).Block headerPayload:Tx1, Tx2, Tx3,.State nBlockfilek value(n 1)k, 𝜃𝜀(k, 𝜃){value(1), value(2),., value(n 1)}kvalue(n 1).State n 1Composite keyEmpty value.HistoryDBFigure 6: TISD construction process.Input: key k, time interval τOutput: list of events1: Events ϕ2: Get current θðkÞ3: for θðkÞ has next do4:if relationðθ, τÞ is Join or Inclusion then5:Append θ to oðθðkÞ, τÞ6:end if7: end for8: for j 0 ; j lengthðoðθðkÞ, τÞÞ ; j do9:res GHFKðoðθðkÞ, τÞ½j Þ10:for res has next do11:Timeporal Join with τ12:Delete event not in τ13:Add events in time interval τ to Events14:end for15: end for16: return EventsAlgorithm 2. TISD query ðk, τÞ.𝜃File name, file path, file stateFigure 7: TIF construction process.algorithms are mainly divided into three types: Proof ofWork (PoW [27]), Proof of Stake (PoS [27]), and traditionaldistributed consensus algorithm.On the block data structure, each blockchain platformuses the Merkle Tree to process transaction data [19]. MerkleTree is normally a full binary tree; as shown in Figure 2, thebottom of each leaf node represents a transaction data onthe blockchain. T 0 is the hash value of Tx0 ’s transaction data.T 1 is the hash value of Tx1 , H 1 is the hash value of T 0 and T 1 ,and T 2 is the hash value of Tx2 ’s transaction data. T 3 is thehash value of Tx3 , and H 2 is the hash value of T 2 and T 3 .After the hash computation of every layer, it can produce ahash and store in the block header.2.2. Hyperledger Fabric Framework. The Hyperledger projectwas established in 2015, led by the Linux Foundation, andattracted projects including IBM, Intel, and other companiescontributed several blockchain platforms like Fabric [28].Fabric is an open-source enterprise blockchain platform,which provides a highly modular and configurable architecture. It can be widely applied to multiple industries [29]. Atpresent, Fabric has been applied in many fields such as banking, financial institutions, and insurance.Fabric needs to be more flexible in the context of enterprise usage. So far, the following options are provided in different versions of Fabric: Solo, Kafka [30], SBFT [31], andRaft.Solo mode is the single-node communication mode. Inthis environment, there is only one sort of service node,and messages sent from Peer point are sorted by this node.The availability and scalability are limited, so it is not suitablefor production environment and is generally used for development and test environment.Kafka mode is the processing of sorting information provided by Kafka service. Kafka is an open-source project ofApache that mainly provides distributed message processingand distribution services. Each Kafka cluster is composed ofmultiple nodes and supports fault tolerance.SBFT is a simple Byzantine algorithm. Compared withKafka, it can tolerate fault nodes and a certain number ofmalicious nodes.In Fabric, there are four kinds of nodes participating inthe network: Peer node, Orderer node, Certificate Authoritynode, and Client node [32].The Peer node is divided into four nodes: accountingnode, endorsement node, master node, and anchor node.Each Peer node is a billing node in the channel, responsiblefor receiving and verifying data blocks sent from the sortingservice while maintaining a copy of the ledger.The Orderer node is received transactions containingendorsement signatures, packages them into blocks, andbroadcasts them to Peer nodes. It can ensure the logical orderof transactions and ensure that all nodes on the chain receivethe same message.Certificate Authority node is to issue identity informationto other nodes in the network, which is the certificate authority in Fabric.Client node is an entity directly operated by users. It participates in the communication of the blockchain network byconnecting to a Peer node or an Orderer node.

Wireless Communications and Mobile n)},{k:value(n 1)}State nk value(n 1)k, 𝜃.kvalue(n 1).𝜃 fileState n 1𝜃{key1:value(1)},{key2:value(2)},. {keyn:value(n)},{k:value(n 1)}MemoryFigure 8: TIF construction process.Input: transaction key and valueOutput: index state1: if current θ still valid then2:if cache full then3:Write cache data to θ file4:Update state ðθ, f ileÞ5:Index state write key value to cache6:else7:Index state write key value to cache8:end if9: else10:Generate new θ11:Update θ12:Index state Generate index based on file ðkey, valueÞ13:end if14:return Index stateAlgorithm 3. TIF generate ðkey, valueÞ.Fabric’s data is stored in block file and database, respectively. Block file mainly stores transaction information,including hash value and block packaging time. For the database, from the perspective of database type, it can be dividedinto Couch DB [33] and LevelDB [34], which can be specifiedthrough the configuration file, and LevelDB is selected bydefault. From the perspective of database usage, there is a ledger index for storing account ID, a block index for storingblock, a state database for the current latest state of transactions on the storage chain, and a history database for storingvalid parts of the transaction information [35].2.3. Level DB. LevelDB is a key-value storage engine developed by Google. LevelDB, like most KV systems, also has asimple operation interface and basic operations that includewriting record, reading record, and deleting record [36].LevelDB is widely used in various projects. The general architecture of LevelDB is shown in Figure 3.Memtable is a storage mode in memory, with the datastructure of skip table as the default mode [37]. When per-forming a data write, it writes to Memtable first, and immutable Memtable changes to read-only Memtable when its datasize reaches the threshold; and then, compression threads inthe background change immutable Memtable to an SSTablefile for storage on disk.SSTable is a data persistence file, which is an orderlyarrangement of keys within the file [38]. The first part ofthe file is the data, followed by the metadata for the index.MANIFEST file contains the key range and the hierarchy.The version of MANIFEST file is changed when the databaseis reopened. If the database is opened, a new file will be generated each time.CURRENT file is the file name of MANIFEST files [39].CURRENT file can be used to locate the MANIFEST file torestore the database in the event of a failure, such as a poweroutage.LOCK file is used to control multiprocess access to thedatabase [37]. When a process opens the database, an exclusive file lock will be added to the file. If the process finishes,the lock will be automatically released.

6Wireless Communications and Mobile ComputingInput: key k, time interval τOutput: list of events1: Events ϕ2: Get θ from StateDB3: for θ has next do4:if relationðθ, τÞ is Join of Inclusion then5:Append θ to oðθ, τÞ6:end if7: end for8: for j 0 ; j lengthðoðθ, τÞÞ ; j do9:s GetState ðθÞ10:if s! nil then11:Read file to memory12:Parse file13:Events temporal14:else15:Read memory16:Events temporal join17:end if18: end for19: for i 0 ; j lengthðEventsÞ ; i doif Events ½i not belong to k then20:21:Remove Events ½i from Events22:end if23: end for24: return EventsAlgorithm 4. TIF query ðk, τÞ.2.4. IoT System. The IoT system connects the Internet andinformation-sensing equipment to achieve intelligent controland management of all objects. It can be divided into threelayers: Perception Layer [40], Network Layer, and Application Layer. Perceptual Layer is responsible for object collection and data processing. Network Layer is responsible fortransmitting the collected data. Application Layer providesIoT-based applications [41].The IoT system usually includes embedded devices withresource constraints. As an important part of IoT, sensorswill face some technical challenges in practical applications[42]. First, due to the limitations of low cost, low power consumption, and small size of sensors, these limit sensors’ computing, storage, and communication capabilities, therebyaffecting the expansion capabilities of the network [43]. Second, most sensors are deployed in unattended areas, so sensor security issues are particularly prominent. Third, sincethe monitoring and recording capabilities of sensors are oftenopaque, when sensors share data information, they will causedata information leakage. The IoT system is being integratedwith various technologies to solve various problems, such asLiang W who proposed a fast defogging image recognitionalgorithm based on bilateral hybrid filtering to solve IoTimage issues in foggy weather [44].Blockchain has significant cryptographic security andimmutability [8]. With its characteristics of decentralization,consensus mechanism, and nontamper ability, IoT can usethe advantages of blockchain to solve its security and scalability issues [45]. The decentralization of blockchain enablesIoT devices to directly obtain information without a centralserver, which reduces the problem of high operating costsfor central institutions. But the IoT system uses the traditional temporal indexes that cannot meet IoT’s needs. Wepropose two index building methods to solve this issue.3. Efficient Temporal IndexFabric distinguishes between the historical state and the current state of the data [46]. The data of transactions are storedin LevelDB in the form of key-value pairs. For each key in thetransaction, there are multiple states corresponding to it. Thelatest state is called the current state, and other states arecalled historical state [47].The state database stores the current state of the key,while the history database stores only the index location ofthe historical data on the block file, which is stored on disk[48]. The historical data is inserted in chronological orderand distributed in different block files, so the efficiency ofprocessing the historical data is relatively low.In Fabric, there is no interface for temporal queries in thesystem. When we need to analyse the historical data, we callthe GHFK function to get an iterator. Through the dataobtained by iterator that is connected with the correspondingtime interval in memory, we realize the temporal query.GHFK function can return all the history data for the committed valid k. The calling process is shown in Figure 4.For example, there is many trading information in thetime interval ½0, T 1 . If we want to query the transactioninformation in the time interval ½0, T 2 ðT 2 T 1 Þ, we needto call GHFK function to access large amounts of data, deserialize the block file, and connect the data to the time interval.If we want to query the transaction information in the timeinterval ½T 2 T 1 , we should connect the data to the timeinterval ½0, T 1 . In other words, the data is useless in the timeinterval ½0, T 2 . It is a waste to read and deserialize the data inthe time interval ½0, T 2 .As time goes on, the block file that needs to be deserialized is increasing by transaction volume and data volume[49]. It will inevitably lead to the decline of query efficiencywith the change of query interval.In this part, we propose two methods which are temporalindexes based on state database and file in Fabric to solve thisproblem. It can improve the efficiency of temporal query inHyperledger Fabric.3.1. Temporal Index Based on State Databases. The reason forthe low efficiency of Fabric temporal query is greatly relatedto the number of reads and writes to the database and thenumber of accesses to the file; especially, when the query volume is relatively large, the efficiency will significantlydecrease. In our method, we divide the transaction into corresponding intervals according to such criteria as time intervals and add indexes to the intervals to speed up the temporalquery. In this section, we will introduce how to create a temporal index (TISD: temporal index based on state databases).In the process of constructing a temporal index, let t 1 bethe starting time of the temporal index and t 2 be the startingtime of the next build process. For each entity k in the transaction, we divide ½t 1 , t 2 into many time frames θðkÞ fθ1 ,

Wireless Communications and Mobile Computing71016091408120Execution time (s)Execution time (s)76541008060340220100010203040506070Query time interval (ks)FIFE8090100010203040506070Query time interval (ks)8090100FabricDIM1Figure 9: Dataset UD, network structure T1, and TISD query time.θ2 , , θm g, and every time frame θi is adjacent. We buildtemporal index for every k in every time frame θi .In Figure 5, for entity k, let εðk, θÞ represent a set of eventsthat relate to k and occur in time frame θ. Let the key-value ðk, θÞ, εðk, θÞ insert the state database, where ðk, θÞrespresents the composite value and εðk, θÞ represents thevalue. It is useful to speed up the temporal index.When the data is submitted to the state database, webuild an index of the time state of the data. The constructionprocess of the index is as shown in Algorithm 1. First, youneed to determine the current time interval θn to verify andthe current time interval is valid. If θn is valid, we use GetStateð k, θn Þ to obtain εðk, θn Þ and update the currentstatus.As shown in Figure 6, the status of k, θn is updatedfrom n to n 1; it also means the current event valuen 1 isadded to εðk, θn Þ. At the same time, the value of k is updatedfrom valuen to valuen 1 . If θn satisfies the separation condition, it will generate a new time interval θn 1 . If θn is invalid,the status of k, θn will update to k, θn , }} and generatea new event ðk, θn 1 Þ, εðk, θn 1 Þ .Algorithm 1: TISD generate ðk, eventÞ.Input: key k andvalue eventOutput: index state1: if current θ still valid then2:Get current state of εðk, θÞ from StateDB3:Append event to εðk, θÞ4:Update εðk, θÞ to StateDB8:Generate new θ9:Append event to εðk, θÞ10:end if11:return stateDuring index building, we need to divide the time intosome time intervals θ, the time intervals are continuous,and all data events are included. We use three times intervalsegmentation methods: FI (fixed time interval), FE (fixedevent number), and DI (dynamic interval).FI: In the total length of time, it builds a temporal indexin the fixed time interval. Each entity may match a differentnumber of events in each time interval by different tradingsituations. Therefore, the efficiency may be erratic.FE: In this model, time intervals θ are determined only bya fixed event number. The length of temporal index may havea big gap in each entity by different trading situations. Forfrequently updated entity values, the index is also frequentlyupdated. There may be multiple time intervals in a shortperiod of time, and the database may be accessed multipletimes during the query.DI: The time interval is determined by calculating thetime and measuring the number of events. A time intervalor several events are fixed, and the time interval must bedetermined in one of two ways. One is that the number ofevents must equal or exceed the specified number of eventswhen the time interval is equal to a fixed value. Another isthat the interval must be at least fixed when the number ofevents is equal to a fixed value, so it is able to avoid thatevents occur too much or too little in a time interval.5: else6:Write hðk, θÞ, εðk, θÞi to block file7:Update εðk, θÞ to HistoryDB3.2. TISD Query Process Analysis. To illustrate the query process, we define four relations: temporal joined relation, temporal connected relation, temporal included relation, and

8Wireless Communications and Mobile Computing1016091408120Execution time (s)Execution time (s)7654100806034022010010203040506070Query time interval (ks)809001000DIM1FIFE102030405060Query time interval (ks)708090100708090100FabricFigure 10: Dataset UD, network structure T2, and TISD query time.2201009020080Execution time (s)Execution time y time interval (ks)70809010002030405060Query time interval (ks)FabricDIM1FIFE10Figure 11: Dataset ZD, network structure T1, and TISD query time.no temporal relation [50]. Let8Join Connection relation θi , θ j Inclusion :Noneθi θi ϕ θm θi θm ϕ, θm θi ϕθi θiotherwise,ð1Þwhere Join represents temporal joined relation, Inclusionrepresents temporal included relation, Connection representstemporal connected relation, and None represents no relation between two intervals.The query process is shown in Algorithm 2, εðk, τÞ is allevents about k in the time interval τ. The algorithm queriesk and time interval θðkÞ by the iterator returned of GetStateByRange and obtains oðθðkÞ, τÞ, where oðθðkÞ, τÞ representsan interval that has temporal joined relation or temporalincluded relation between θðkÞ and the target query interval.There is a temporal connected relation between the first θand the last θ in oðθðkÞ, τÞ.Algorithm 2: TISD query ðk, τÞ.Input: key k, time interval τOutput: list of events1:Events ϕ2:Get current θðkÞ3:forθðkÞ has next do4:ifrelationðθ, τÞ is Join or Inclusion then

Wireless Communications and Mobile Computing92201009020080Execution time (s)Execution time ery time interval (ks)8090100DIM1FIFE010203040506070Query time interval (ks)8090100FabricFigure 12: Dataset ZD, network structure T2, and TISD query time.Table 1: TISD data insertion time comparison 255.63DatasetUDZDFabric227.26228.59Table 2: TISD data insertion space comparison 80191.78DatasetUDZD5:6:Fabric103.5103.5Append θ to oðθðkÞ, τÞend if7: end for8: forj 0 ; j lengthðoðθðkÞ, τÞÞ ; j do9:10:res GH FKðoðθðkÞ, τÞ½j Þfor res has next do11:Timeporal Join with τ12:Delete event not in τ13:Add events in time interval τ to Events14:end for15:end for16:return EventsFor each θ in oðθðkÞ, τÞ, GHFKð k, θ Þ is performed.The block file is parsed through the returned iterator. Forthe interval with temporal included relation, the data parsedby the iterator is appended to the result sets. For the intervalwith temporal joined relation, the data returned by the iterator is traversed to remove the data not in the interval τ.In the temporal query process, the efficiency of the queryis mainly measured by the query time, which depends on therate of events, the size of the index interval, and the size of thequery interval. Therefore, how to choose the appropriateindex interval becomes a key problem.For the temporal query performed by Fabric, T qkτ is usedto represent the query time of the historical state of the keyword k in the time interval τðt s , t e Þ, which can be expressed asT qkτ numðEðk, τÞÞ T it ,ð2Þwhere EðkÞ represents events about the keyword k and numðÞ represents the number of events in the query time.In the normal process of data insertion, we use V k to represent the insertion rate of events related to the keyword k,which can be expressed asVk numðEðk, T ÞÞ,Tð3Þwhere T represents the total length of time the event inserted.Since the first transaction in the system needs to be iterated in the Fabric query process, T can be expressed as aperiod of time, starting from the initial transaction point t 0and ending at a larger time point in the target query intervalτ. SoT qkτ V k T T it V k ðmax ðt s , t e Þ t 0 Þ T it :ð4ÞFor the FI method, the index construction is determinedby a fixed time length. For the total time length T and the

10Wireless Communications and Mobile Computing1016091408120Execution time (s)Execution time (s)76541008060340220100010203040506070Query time interval (ks)80901000DIM1FIFE10203040506070Query time interval (ks)8090100FabricFigure 13: Dataset UD, network structure T1, and TIF query time.210807220064Execution time (s)Execution time ry time interval (ks)8090100DIM1FIFE010203040506070Query time interval (ks)8090100FabricFigure 14: Dataset ZD, network structure T1, and TIF query time.given fixed time length T f , letnumðθ, T Þ total length of time, soT,Tfð5Þwhere θ represents the time interval divided, numðÞ represents the number of partitions for all time intervals.The

2.2. Hyperledger Fabric Framework. The Hyperledger project was established in 2015, led by the Linux Foundation, and attracted projects including IBM, Intel, and other companies contributed several blockchain platforms like Fabric [28]. Fabric is an open-source enterprise blockchain platform, which provides a highly modular and configurable .