Computers & Industrial Engineering 190 (2024) 110021
A
0Contents lists available at ScienceDirect
Computers & Industrial Engineering
journal homepage: www.elsevier.com/locate/caie
Hyperledger sawtooth based supplychain traceability system for counterfeitdrugsAnum Nawaz a,b,∗, Liguan Wang b, Muhammad Irfan a, Tomi Westerlund a
a Turku Intelligent and Embedded Robotic Systems Lab, Faculty of Technology, University of Turku, Finlandb Shanghai Key Laboratory of Intelligent Information Processing, School of Computer Science, Fudan University, Shanghai, China
A R T I C L E I N F O
Keywords:BlockchainHyperledger sawtoothDrugs supplychainDrugs traceabilityQuantum secure communication
A B S T R A C T
Drug supply chains have been facing severe issues as counterfeit product cases are increasing exponentially.A supply chain management system that ensures transparency, reliability, and provenance of drugs wouldincrease the trustworthiness of the whole industry. One solution to all three needs is utilizing blockchain-based distributed ledger technologies (DLTs). Even though DLTs are emerging as an ideal infrastructurefor multi-stakeholder supply chain applications, they still need to be more mature to address the specificchallenges specific to each use case. In this article, we propose a distributed blockchain-based framework,PHTrack, leveraging hyperledger sawtooth as a drug supply chain traceability system. Hyperledger sawtoothaddresses scalability issues by offering a robust foundation to support large-scale drug supply chain operationsin a modular way for it’s each participating stakeholder. Furthermore, it simplifies the integration processwith existing systems, even those employing different technologies, thereby facilitating a smoother transitionto DLTs. The design of PHTrack is oriented towards minimizing resource consumption throughout theprocess, particularly within hyperledger sawtooth nodes. Additionally, it incorporates quantum secure off-chain communication for peer-to-peer (P2P) communication. A set of experiments was conducted to validatethe proposed framework. Experiments have shown that PHTrack provides reliable and comprehensive drugprovenance as well as real-time drug supply chain tracking.1. Introduction
Resilient and trustworthy supply chain systems are vital in modernsociety to answer sudden and unexpected changes in the demandfor drugs or other essential merchandise. One of the challenges isthat the current global supply chain management solution is a wildlycomplex, interlinked network of untrusted bodies (Hassija & Vikas,2020). Solely drug distribution has exploded in size and complexity,making it complicated to find loopholes in the systems. Therefore,building trust among the supply chain parties is essential. To underlinethe importance of the trustworthy supply chain, consider the processfrom drug discovery through development and regulatory approval topharmacy; the process is hazardous and takes several years at mini-mum (Dauvergne, 2022). Thus, having a supply chain system whichensures that a customer is receiving a genuine product developed bya legitimate manufacturer rather than a counterfeit one cannot beundermined.
∗ Corresponding author at: Turku Intelligent and Embedded Robotic Systems Lab, Faculty of Technology, University of Turku, Finland.E-mail addresses: anunaw@utu.fi, 18110720163@fudan.edu.cn (A. Nawaz), 17110240019@fudan.edu.cn (L. Wang), muhammad.m.irfan@utu.fi (M. Irfan),tovewe@utu.fi (T. Westerlund).URL: https://tiers.utu.fi/team (A. Nawaz).
There are several supply chain management platforms for the phar-maceutical industry. Yet, they are outdated, incapable of allowingmanufacturers and regulatory agencies to control drug distribution, andwithstand modern cyber-security threats (Melnyk, 2022). When a phar-maceutical company’s supply chain is breached, fraud and counterfeitdrugs are more likely to happen. Even a slight doubt is serious becausecounterfeit drugs may include harmful ingredients, incorrect propor-tions of ingredients, substandard composition, correct ingredients withwrong packaging, fake brand names, or expired products (Moosivand,Ghatari, & Rasekh, 2019). Even if the drugs are excellent, doubt can-not be removed because drug regulatory authorities do not superviseproduction and distribution.To give an insight into the size of the problem, the European UnionIntellectual Property Office’s (EUIPO) report estimates that counter-feit drugs are causing pharmaceutical companies to lose over EUR16.5 billion in sales and affect more than 80 thousand jobs in thevailable online 27 February 2024
360-8352/© 2024 The Author(s). Published by Elsevier Ltd. This is an open access a
https://doi.org/10.1016/j.cie.2024.110021Received 26 September 2023; Received in revised form 15 January 2024; Accepted
rticle under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
26 February 2024
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.
d2isttta
tmd2cblpaS
iOcmStedpharmaceuticals sector and other sectors that sell goods and servicesto it (Pharmaceutical sector - observatory, 2016). According to theWorld Health Organization (WHO), a trade worth 73 billion euros incounterfeit medicines is taking place annually (Kaiser-Kershaw, 2022).WHO also reports that counterfeit medical products increased by 47%from 2020 to 2021 (Ziavrou, Noguera, & Boumba, 2022). The caseof third-world countries is much more severe than that of developedcountries, and the news reflects its intensity regularly. Further, the re-port indicated that fake drugs causing pneumonia cause 72,000 deathsin children, and 69,000 people die of malaria each year, which isconsidered one of the significant causes of death.
1.1. Challenges and loopholes in existing systems
The presence of falsified drugs in the pharmaceutical supply chainis one of the fundamental threats to public health (Abdallah & Niza-muddin, 2023; Saraji, Rahbar, Chenarlogh, & Streimikiene, 2023).Developing and deploying an efficient traceability channel has becomea need of the hour around the globe. These tracing and trackingrequirements need an authenticity system throughout the life cycle ofa drug. It starts by tracking the origin of all pharmaceutical ingredientsthrough production, packaging, and transportation to wholesale dealersand pharmaceutical stores (Sylim, Liu, Marcelo, & Fontelo, 2018). In abigger picture, along with counterfeit and falsified drugs, the pharma-ceutical industry is facing several issues in conventional systems (Ali &Kannan, 2022; Dasaklis, Voutsinas, Tsoulfas, & Casino, 2022; Nguyen,Lamouri, Pellerin, Tamayo, & Lekens, 2022):
• Unavailability of the system for real-time verification of prove-nance for medicines.
• Artificially created market shortage of medicines in epidemicsituations.
• Uncontrollable situation of drug prices.
• Disruption in manufacturing and falsified usage details of importActive Pharmaceutical Ingredients (APIs).
• Wrong reporting of adverse drug reactions (ADR) for genuinemedicines caused by the consumption of counterfeit drugs.
Loopholes in centralized cloud systems turned the interest towardsecentralized and distributed blockchain systems (Malik & Alkhatib,021). The tracing capabilities of blockchain represent a novel man-festation of supply chain tracking solutions. It constitutes a temporallyequenced series of blocks systematically arranged by the computa-ional entities of diverse participants (Azevedo & Gomes, 2023). Withinhe blockchain framework, the immutability of blocks containing theransactional history among nodes in a P2P network is ensured bypplying hash functions (Altaf, 2023; Wang, Duan, & Zhu, 2018).Blockchain technology can be integrated as an embedded web layero facilitate various functionalities, including but not limited to pay-ent processing, currency exchange, token reception and distribution,igital asset transfers, and the execution of smart contracts (Wang,022). In particular, distributed ledger technologies provide improvedontrol of data by providing increased supply chain transparency usinglockchain Saberi, Kouhizadeh, Sarkis, and Shen (2019), privacy by uti-izing distributed technologies (Nawaz et al., 2019), ownership incor-oration by enabling blockchain-based solutions (Nawaz et al., 2020),nd leveraging blockchain as a security measure (Killer, Rodrigues, &tiller, 2019).Thus, Blockchain technologies can counter falsified drugs and allllegal activities through increased traceability of goods (Kordestani,ghazi, & Mostaghel, 2023). However, distributed-based solutions en-ounter challenges related to high variability in structural require-ents for each entity involved in supply chain solutions (Karuppiah,ankaranarayanan, & Ali, 2023). Scalability and reliability issues dueo complex requirements and large-scale implementations (Pandeyt al., 2023), integration with existing conventional solutions based on2
ispersed technologies and interoperability challenges.1.2. Our contribution
The main motivation of our proposed system is to answer the above-mentioned challenges in conventional and blockchain-based supplychain solutions. Based on the findings, we propose a modular dis-tributed blockchain-based solution utilizing hyperledger sawtooth framework‘‘PHTrack’’ to address the multifaceted above-mentioned challengesencountered in distributed supply chain solutions. Its highly modularstructure enhances supply chain operations while accommodating theunique requirements of each participating stakeholder.
• PHTrack addresses the solutions that stem from the substantialdivergence in structural and industrial requirements among thevarious entities participating in the drug supply chain traceabilitysystems.
• It addresses interoperability challenges by fostering a cohesiveand collaborative environment among the diverse stakeholders inthe drug supply chain ecosystem.
• It resolves scalability issues by providing a robust foundation ca-pable of supporting large-scale industrial supply chain operationsby utilizing a highly modular structure, hyperledger sawtooth.
• It eases the integration process with pre-existing systems, oftenemploying different technologies, enabling a smoother transitionto blockchain-based solutions.
• The design of PHTrack is geared towards minimizing resourceconsumption throughout the process, notably within the block-chain hyperledger sawtooth node, while ensuring the privacy andsecurity of participating entities.
• It incorporates quantum secure off-chain communication for P2Pcommunication.
To ensure the feasibility and usability of the proposed PHTrack system,We put it in a real-time testbed environment. To safeguard transac-tion execution against post-quantum attacks, we employ post-quantum-based data sharing and authentication schemes using TLS-based securecommunication. Based on the author’s understanding, the proposedframework represents one of the first comprehensive solutions utilizinghyperledger sawtooth, offering real-time provenance capabilities.
2. Background and related work
Regulatory bodies like the FDA have introduced rules like the DrugSupply Chain Security Act (DSCSA), a law to tackle the problem ofcounterfeit drugs (Commissioner, 2023). They aim to create a transpar-ent supply chain to prevent counterfeit drugs from entering the market.The DSCSA, enacted in 2013, was a response to a nationwide fungaloutbreak caused by contaminated steroidal injections. It mandates thatall entities involved in the pharmaceutical supply chain, includingmanufacturers, distributors, repackagers, and pharmacies, must providedetailed transaction histories and statements, known as T3 informa-tion, to the next party in the chain. This information helps verify theauthenticity of the product and ensures that it has been handled byauthorized trading partners. It must be retained for at least six yearsafter the product is received.
2.1. Non-distributed ledger technologies
Several non-distributed methodologies were presented in differentstudies to counter drug counterfeit problems. Authors in King andZhang (2007), Onieva et al. (2015), proposed RFID-based architectureto create a drug supply chain resistant to counterfeit drugs. Theyalso present a method for finding RFID events more efficiently anddiscuss the requirements needed to ensure the authenticity of a productusing RFID technology. The proposed components communicate usingtechnologies such as Bluetooth and Wi-Fi to provide real-time trackingand tracing, along with incident reporting. NFC tags-based systemswere presented by Alzahrani and Bulusu (2016) that can only be
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.
Table 1Perspective mapping of Blockchain as an Entity for supply chain solutions tracking and counterfeit detection.Sr. No Entity Relation Entity Perspective
1 Blockchain consists of Blocks Distributed Ledger2 Blocks contain Header Chain Records3 Blocks contain Body Transactional Records4 Blocks contain Reference Distributed Trust5 Blockchain maintains Historical Record Timestamped Blocks6 Historical Records preserves Irreversibility of Record Provenance7 Irreversibility of Record maintains Transparency Reliability8 Transparency uses Pseudonymity Anonymity9 Blockchain includes Shared Database Distributed Trust10 Shared Database secures P2P Transmission Validity11 P2P Transmission includes Encrypted Transmission Integrity
scanned once, thus preventing the duplication of these tags and theintroduction of counterfeit drugs. Authors in Meng, Liang, Xu, and Li(2022) combine NFC tags with a lightweight authentication protocol.This protocol allows for the updating of NFC tags. Once a tag isauthenticated by the anti-counterfeiting server, the user receives theactual drug data.Recently, authors in Valizadeh et al. (2023), present a solutionfor improving the vaccine supply chain to overcome obstacles in thepublic vaccination program based on government and organizationalconcerns. To achieve this, a strong two-level optimization model issuggested. Findings of the hybrid model discussed in Khan, Gupta,Gunasekaran, Mubarik, and Lawal (2023) can help decision-makers andmanagers in the healthcare sector develop strategies to improve theirsupply chain performance. Authors in Aytekin, Görçün, Ecer, Pamu-car, and Karamaşa (2023) present fermatean fuzzy sets-based optimalselection criteria of pharmaceutical supply chains by the healthcareindustry. Decision-makers can use the proposed model as a guide toevaluate their capabilities and improve their skills. The sensitivity anal-ysis results show that the proposed fuzzy fuzzy-weighted anticipatorySystem (FF-WASPAS) approach is a robust and practical framework.Despite the benefits of this semi-centralized or centralized architec-ture, which improved the efficiency of drug supply chains, there are stillchallenges to address. Integrating this with existing systems is difficult,and all trade partners must also subscribe to this system. Centralized au-thorities are more susceptible to security threats and data manipulation.With an increase in the number of requests, the system’s response timeis severely affected (Nguyen et al., 2022). Information fragmentation isalso a significant issue, as it does not provide transparency into thesupply chain, leading to many discrepancies. Auditing such systemsis also challenging, and the cost of deployment and maintenance ishigh. Therefore, more alternatives that take advantage of the latesttechnological advancements, such as distributed ledger technologies(DLTs), should be explored.
2.2. Distributed ledger technologies
The first successful DLT application was Bitcoin, which managesdigital assets to solve problems of double spending and anonymityissues (Nakamoto, 2008). Blockchain-enabled solutions leverage dif-ferent DLT frameworks, and some of them are designed for specificdomains. Some frameworks are good for applications requiring permis-sionless architectures, and a few are specifically designed for permis-sioned networks. Permissionless blockchain solutions are publicly avail-able for everyone to join, which creates scalability and performanceissues (Liu, Wu, & Xu, 2019). For example, the transaction rate of bit-coin is limited to 7 transactions per second, which makes it incompetentto handle high-frequency trades. Furthermore, in permissionless net-works, the transaction confirmation rate increases exponentially as thenetwork expands. Along with the performance, it creates transactioncost issues. Due to the above-mentioned issues, many observers believepermissionless networks are unsuitable for large-scale non-financial
3
applications such as supply chain solutions (Wamba & Queiroz, 2020).In contrast to permissionless networks, permissioned blockchainnetworks identify each node, and administrator nodes are capableof removing malicious nodes (Helliar, Crawford, Rocca, Teodori, &Veneziani, 2020). These network models improve performance through-put by using more adaptive consensus protocols like Practical ByzantineFault Tolerance (Xu et al., 2021), side chains (Singh et al., 2020),and blockchain-based edge-computing solutions (Nguyen Gia, Nawaz,Peña Querata, Tenhunen, & Westerlund, 2019). In recent years, re-searchers proposed various permissioned blockchain frameworks suchas ethereum (Buterin et al., 2013), EOS (Grigg, 2017), hyperledger (An-droulaki et al., 2018), and ripple (Benji & Sindhu, 2019). Permissionedblockchain networks are highly suitable for supply chain solutions andcan be customized according to business needs and requirements. Theysupport transactional-level privacy and network-level transparency. InTable 1 perspective mapping of blockchain as an entity for supplychain solution is presented. However, it is worth noting that eachframework may have its strengths and weaknesses according to drugsupply chain solutions. For instance, ethereum is focusing on Layer 2rollups, a technique that can support many transactions per second.In Musamih et al. (2021), ethereum blockchain is utilized along withoff-chain storage to make tracking products easier in the pharmaceu-tical supply chain. The proposed method ensures the origin of data,removes the need for middlemen, and provides a secure, unchangeablehistory of transactions for everyone involved. Dwivedi, Amin, andVollala (2020) proposed and developed a smart contract algorithmusing directed graphs with six states and six actions. In addition toperforming strong key management in smart contracts, it also achievesreasonable performance in terms of computation and communicationoverheads. The proposed protocol was robust and achieved reasonableperformance regarding smart contract performance. Authors in Agrawaland Angelis (2022) proposed a framework and smart contracts toensure data accuracy and authenticity within supply chains requiringhighly accurate and authentic data. The proposed framework has yetto be tested in an industrial setting. Transaction and maintenance costswere not analyzed on the blockchain network. In Abdellatif and Al-Marridi (2020), authors optimize data sharing among participatingstakeholders using restricted blockchain using edge computing concept.They proposed a blockchain-based architecture and enabled a flexi-ble configuration to securely share and access medical data betweenhealthcare organizations, enabling the detection of probable epidemics,remote monitoring of patients and quick response times.
2.3. Modular distributed ledger technologies
Wang, Ye, Meng, and Xu (2020) undertook an examination ofthe four predominant blockchain platforms, namely Ethereum, Fab-ric, Sawtooth, and Fisco-Bcos. Their findings indicated that in termsof performance metrics such as latency and throughput, hyperledgersawtooth exhibits superior performance compared to other platforms.The architectural attributes of hyperledger sawtooth were highlighted,emphasizing its simplicity, modularity, and considerable flexibility forcustomization.
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.Fig. 1. Drug distribution supply chain and its stakeholders.
Hyperledger frameworks, on the other hand, offer a modular design,allowing for extensive customization. Authors in Sharma and Rohilla(2023) present hyperledger fabric-based application for drug discovery.The proposed application allows organizations to upload, update, view,and verify their contributions. Each contribution is given a uniqueidentifier using a secure hash algorithm, and the design also allowsregulatory authorities to issue certificates confirming the ownershipof contributions. Uddin (2021) investigated the challenges faced bypharmaceutical supply chains and proposed hyperledger fabric-basedsolution, but their proposed solution lacks any kind of implementationtest-beds which leads to the feasibility of the proposed framework.Work of Vijay and Priya (2022) proposed hyperledger fabric-based agri-cultural supply chain to enhance the trust between the end consumerand the product they are purchasing, proposed solution ‘‘grainchain’’enables the tracking of the grain from the farmer to the retailer.Authors in Wisessing and Vichaidis (2022) present an IoT-basedseafood supply chain by leveraging hyperledger sawtooth framework.Consumers can track seafood and notify drivers through a seafoodsupply chain when the container temperature exceeds a specified level.The authors recorded the shipment log as a blockchain transaction byconnecting the application, the devices, and the blockchain database.Another study is proposed (Mohit, Kaur, & Singh, 2022) as genericownership and traceability of products using hyperledger sawtooth.The proposed system can prevent counterfeit goods from entering thesupply chain.Compared to other permissioned frameworks in Table 2, hyper-ledger sawtooth’s event system can potentially lead to more efficientand effective operations in a drug supply chain context. Broadcastingand relaying events across the network enables real-time updates andactions, which are crucial in supply chain management for timely andaccurate tracking of goods. Due to its modularity, each participatingstakeholder can define its own business logic and interact throughtransaction families. Transaction families work similarly, such as smartcontracts in ethereum. However, unlike ethereum and other distributednetworks, each node or application handles and defines its own transac-tion family rather than using the business logic of a complete system.Transaction states share updates between untrusted stakeholders andare regulated through consensus protocols.
3. Hyperledger sawtooth based pharmaceutical supply chain
As discussed, drug transparency from the point of origin of activeingredients (supplier) to distributors and then to end-users (consumers)requires a complete network framework. In the next section, we in-troduce all the entities in our proposed system: pharmaceutical supply4chain stakeholders, the hyperledger sawtooth framework and its func-tionalities, REST APIs, and a consensus protocol to validate transac-tional blocks. Fig. 1 illustrates all stakeholders and their relationshipsin the pharmaceutical supply chain. and Table 3 describes the role ofeach participating stakeholder.
3.1. Hyperledger sawtooth framework
Under the broad spectrum of open-source hyperldeger frameworks,hyperledger sawtooth is a private permissioned network proposed andbuilt by the Linux foundation (Dhillon, Metcalf, & Hooper, 2017). Inteldesigned this distributed ledger specifically for highly modular businesslogic where governance bodies must customize rules and regulations ina run-time environment while maintaining immutability and privacy.Its modularity separates its core system and the application domain.Therefore, each participating entity can define a set of rules accordingto its requirements without knowing the underlying business logic ofthe core system.In sawtooth, business decisions take place on the transaction pro-cessing layer, where transaction families work as models to handlelow-level functions like sets of permissions, policies, and storing blockstates and logs. Transactions are explained by the ‘‘transaction family’’through which a transaction state changes. Corresponding transactionprocessors are bonded to the transaction’s execution by each entity.This modular structure can handle several consensus protocols such asPractical Byzantine Fault Tolerance (PBFT) (Xu et al., 2021), RAFT (On-garo & Ousterhout, 2015), Proof of Elapsed Time (PoET) and PoETsimulator (Corso, 2019). Core Modules of hyperledger sawtooth aredescribed in detail:
3.2. Consensus protocol
A procedure in which all participating nodes of a network endorseor reject some transactional state is known as a consensus proto-col. Several options with different attributes like throughput, finality,size, latency, threat model, censorship resistance, and failure modelare available. Distributed technologies build trust among untrustedmembers through these algorithms. After facing many drawbacks inProof-of-X protocols, PBFT, RAFT and PoET consensus protocols areproposed with hyperledger sawtooth. The PBFT consensus engine ispreferable to maintain fault tolerance and resolve issues in the originalchain. Raft is a leader-based consensus protocol that is preferable fora small number of participating nodes. Intel has presented Proof ofElapsed Time (PoET) (Corso, 2019), a promising, novel consensus pro-tocol, and their SGX hardware as a trusted environment. Nevertheless,PoET can also be used without Intel hardware support by using the
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.
3
tiTable 2Comparison of hyperledger sawtooth with hyperledger fabric, iroha, and burrow.Feature Hyperledger sawtooth Hyperledger fabric Hyperledger iroha Hyperledger burrow
Ledger Type Permissioned Permissioned Permissioned Permissioned
Smart ContractFunctionality Transaction Processors Chaincode Smart Contracts Ethereum VM
Consensus Algorithm Pluggable Framework Pluggable Framework YAC Raft
Governance Linux Foundation Linux Foundation Hyperledger Not specified
Key Features Parallel transaction processing,Compatible with Ethereum bySeth, Secure privatepermissioned network,Pluggable consensusmechanism
Modular framework, Pluggableconsensus mechanism User-friendly interfaces,Modular design, Custom smartcontract logic
Ethereum compatibleTable 3Drug supply chain stakeholders and their roles.Stakeholder Role in drug supply chains
Drug RegulatoryBodies: The World Health Organization (WHO) defines the drug regulatory bodies role as monitoring,protecting, and enhancing pharmaceutical products’ safety, efficacy, and consistency.Additionally, they supervise the production, delivery, and storage of prescription drugs toidentify and sanction illegal drug production and trafficking.
ActivePharmaceuticalIngredient (API)Supplier:
APIs supply raw materials and other active ingredients to the manufacturer. Variousinformation about the ingredients will be logged on the ledger so that, if a recall occurs,stakeholders can identify the source of raw materials, such as the name, unique code, quantitysupplied, and date of supply.
Manufacturer: To produce prescription medications, pharmaceutical suppliers provide traceable identifiers tosuppliers. The manufacturer is responsible for encoding the information related to the drugusing coding standards such as EAN/UCC-13, since the manufacturer owns the drug.
Wholesaledistributor: Wholesale distributors ensure that all stakeholders in the supply chain of drugs can purchasedrugs efficiently, transparently, reliably, and uninterruptedly. Besides offering wholesaledistribution of drugs, they also offer packing, repackaging of drugs, and online orderingservices.
Re-distributor: Distributors manage a complex supply chain, harnessing innovative technologies to ensure safe,secure, and efficient delivery. The next step is initiating the distribution process after wholesaledistributors. The distributor will pack and transfer the drug lots to the concerned retailers.
Retailers: Pharmacies work as retailers, and according to estimates, pharmacies monitor about 70% ofthe prescription drug industry. They buy drugs in bulk from wholesalers and resell them toend-users (patients). The relationship between pharmacies and patients is close. Based on thepoint-of-sale requirements, these pharmacies may be self-governing or franchise-based.
Consumers: Consumers are also included in this supply chain as they buy drugs from pharmacies.Consumers are those actors in the proposed solution who have the right to query the ledger.Consumers can scan QR codes from their mobile phones and see the complete history oftransactions and events for the specified drug.PoET simulator with hyperledger sawtooth. This protocol stands out formany reasons. It is considered one of the most robust implementationsof the Proof-of-X protocol and comes as a part of the sawtooth project.Furthermore, it is highly parameterizable, unlike the bitcoin protocol.It is also a currency-independent protocol, which makes it best suitedfor use cases without financial transactions. As it comes as a suit witha sawtooth, its working environment follows the sawtooth structure:
• Validator node requests for a waiting time from an enclave(trusted module).
• Enclave assigns waiting time randomly to each validator.
• Leader is elected by checking the validator with the shortest waittime ‘‘CreateTimer’’ function creates a timer which guarantees thecreation of transaction blocks by an enclave.
• Validator can claim leadership after finishing the allocated wait-ing time.
.3. Data model
The architectural design of the data model incorporates a sequen-ially arranged collection of transactions, logs for transactional activ-5
ties, and a distributed framework for data storage to maintain theresultant states. The management of transaction serialization is facil-itated by deploying a Radix Merkle Tree structure (de Ocáriz Borde,2022). Each participating node is equipped with a transaction processorand is allocated a discrete namespace for implementing its proprietarybusiness logic. Within a sawtooth ecosystem, a transaction processorfunctions analogously to smart contracts within the ethereum platform.The serialization schema offers considerable adaptability and can lever-age decentralized storage mechanisms, both on-chain and off-chain, topreserve batches of committed transactions chronologically. Each databatch comprises multiple transactions, a timestamp, the index hash ofthe previous batch, and a Merkle root to authenticate the integrityof the batch data. The aggregation of transactions into a single databatch occurs at arbitrary intervals. Upon the amalgamation of severaldata batches, a data block is constructed, with the timing parametersbeing determined by the business logic. Each participating entity isfree to select any decentralized storage system, such as Firebase (Paul,2023), Filecoin (Vakilinia, Wang, & Xin, 2023), or the InterplanetaryFile System (IPFS), based on their specific business requirements.Every stakeholder has the option to integrate their individual off-chain storage solution. Sawtooth’s utilization of off-chain storage sys-tems offers distinct benefits, provides cost-effective scalability andefficient data processing for less sensitive information. This bifurcation
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.Fig. 2. Layered structure of sawtooth based Supplychain traceability framework PHTrack.allows for the secure submission and validation of requests, main-taining the confidentiality of transactional data while catering to thespecific needs of various stakeholders. Moreover, the approach aids inadhering to sawtooth’s business rules functionality, ensuring a robust,scalable and flexible solution.
3.4. Execution model
The highly flexible structure of hyperledger sawtooth supports therunning and compilation of code as docker images. Docker images aresimply chain codes which can interact hyperledger’s back-end via pre-defined interfaces. It supports parallel executions of chain codes andtransactions in each node. The transaction processor must verify eachstate and transaction before adding it to the data batch. Moreover,each participating node can connect to other miners, validators, andlight nodes via some RPC-like mechanisms. Third-party interfaces andapplications are built on top of the chain and can connect the ap-plication domain to the core-level module and vice versa. The RESTAPI allows the client to communicate with the core system module‘‘Validator’’ using HTTP/JSON standards. This pragmatic RESTful APIprovides a simple, language-neutral interface to submit transactionsand read/write requests. It works as a lightweight layer on top ofsawtooth’s internal ZMQ communication, so it does not require anyauthentication and passes the message requests to the validator forsignature verification. It uses the validator component as a black box,which can only send requests and get the required result withoutknowing the internal system logic.
4. PHTrack: Hyperledger sawtooth based pharmaceutical supplychain
In this section, we described our proposed system, PHTrack, indetail by highlighting its architecture, system flow, and deployment toenable telemetry of the pharmaceutical supply chain and provenanceof drugs.
64.1. PHTrack: System architecture
PHTrack leverages hyperldeger sawtooth technology to provide on-chain governance to the drug regulatory authorities. It provides cen-tral authoritative controls and offers better scalability, transaction ef-ficiency, immutability, interoperability, highly modular, fine-grainedtraceability, and run-time upgradation of the consensus protocol. Itprovides on-chain governance by using dynamic consensus protocolsand provides permissioning features. Each stakeholder possesses itstransaction processor and can run its business logic via chain codes,side-chain decentralized storage, and consensus protocol. The PHTrackframework prefers the PoET consensus protocol. It offers the solution tothe Byzantine Generals’ Problem by using a trusted environment. PoETis unlike the traditional lottery-based protocol, where the chances ofwinning are proportional to the amount of computational work. Insteadof spinning the computational load, it uses the element of randomnessto control the block commitment. PoET elects the individual nodes toexecute requests at a prescribed target rate. It is similar to Proof ofWork but replaces huge computation with a cheap random wait.To create a trusted network, we use a peer-to-peer network of nodesas a validator, which provides journal block management and identitymanagement system services. It manages User_ID and authenticatesall participating nodes (modules) by issuing registration_certificate. Thehigh-level architectural diagram of the proposed framework PHTrack isdepicted in Fig. 4 and layered structure of proposed system is defined inFig. 2, highlighting all the major system modules involved. Each noderepresents one complete module (stakeholder) of a system. The numberof compulsory and optional components involved in the constitution ofeach node. Compulsory items are P2P sawtooth environment, transac-tion processor, consensus engine, validator, and REST API services tointeract with the system. Optional items include side chains, storagesystems, client applications and their inter-communication protocols.A highly modular structure consists of all stakeholders involved in thepharmaceutical supply chain, a sawtooth core system module, an appli-cation domain, a distributed storage system, and clients. The REST API
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.Fig. 3. Process to Track backward using dynamic QR code.Fig. 4. A high-level architecture for the proposed PHTRACK sawtooth-based system for Drug supply chain.
develops communication between core system modules and applicationdomains. Fig. 3 shows how clients use it to submit transactions andread/write requests to the validator and its associated components.Each stakeholder node represents some sort of node category:
• Validator node: Each stakeholder is represented as a validatornode. It is responsible for authorizing transaction requests relatedto it after getting approval from the transaction processor. Itdoes not need to participate in other network transactions (otherstakeholders’ transactions).
• Leader node: The leader node is responsible for committing abatch of authorized transactions after random time interval t.
• Client node: Clients can submit read requests. Customers can trackdrug information from the origin to the point of sale by scanningdynamic QR codes.7Chain codes handle and deploy regulations, business logic, and trans-actions. These are the central and essential components for handlingpeer-to-peer networks and managing the complete system flow. More-over, accessibility roles are different for each stakeholder, defined andexecuted via chain codes. Dynamic QR codes are used to track druglots. When creating a QR code for each drug lot, all of the drug lot’stransactional history is saved in the QR code, which updates all ofthe event information to their side chain and broadcasts the hashand metadata of the saved block in the network. Each participatingstakeholder can request access to the information saved in the block.The drug regulatory bodies and the owner of the data block own thevalidating rights.
4.1.1. Core system moduleThe core system module consists of a validator node, which includesa single validator, a REST API service, a consensus engine, a state
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.
Mrspactoa
4
bvnulaslispdatcipbca
4
ihiewctttoowwolt
4
rtfri
attitPotw
4
dtTviwabtdla
5
otobdec
𝑆
+
w
𝑡database, a peer-to-peer network, and one or more transaction pro-cessors. The transaction handling module is inside the validator node,which handles messages, transaction batches and blocks, validation,and publishing after getting approval from the specified transactionprocessor. One data batch consists of a random number of transactions,and one datablock consists of databatches created at a specified time 𝑇 .ultiple transactions work as an atomic unit in a batch. The validatorejects the complete transaction batch by rolling it back to the previoustate, even if a single transaction failed to complete the validationrocess. Transaction batches are wrapped in batch lists. It allows par-llel transaction execution and validation. Each pharmaceutical supplyhain stakeholder works as a network’s validator node. Furthermore,he validator is also responsible for peer-to-peer communication withther validator nodes (stakeholders) on the network. REST API servicesre used to communicate with transaction processors and clients.
.1.2. Application domainThe client is responsible for creating and signing transactions, com-ining them into batches and blocks, and submitting them to thealidator through the REST-API module. The client application usesode.js based application, which provides a web user interface for theser to interact with. Transaction families are used to write businessogic and policies. Transaction families separate the transaction rulesnd content of core functionalities. A transaction family can be a simpleet of rules to change the state of a system or a complete set ofogic defined by the stakeholder. These are smart contracts writtenn any supported programming language, used to automate the entireystem. We use python3 and Go to write transactions. The transactionrocessor and client use the same data model, serialization, and ad-ressing scheme. Setting transaction families enables participants togree on prescribed network policies. Transaction families authorizehe transaction requests by other validator nodes and the clients. Alient creates a transaction and submits it to the validator. The val-dator then authorizes it after getting approval from the transactionrocessor. Each node is assigned one authorized key, which is usedy the validator to authorize each requested transaction. After gettingonsensus approval, any validator node can act as a leader to approvetransaction batch.
.2. PHTrack: System flow
To track and trace counterfeit drugs, our proposed system flows described in two ways: track forward and backwards. Each stake-older works as a validator node, which defines its working policiesndependently without leaning on the core system. To join a network,very stakeholder must implement online system requirements andrite transaction policies based on their business logic. A stakeholderan start their validator node and associated components after get-ing approval from the drug regulatory bodies (Leader). To achieveraceability, it is essential to have two separate and asynchronousransaction flows. Transaction flow describes the physical movementf the original drug and the associated information flow going fromne stakeholder to another. ’Track backward’ is the primary workflow,hich allows tracing the origin of active ingredients, manufacturer,holesale distributor, and retailer, as well as the location of the drugsn the PHTrack. Drug tracking refers to tracking the drug’s currentocation in the supply chain and its related information in the currentransaction process until the drug reaches its end-user (patient).
.2.1. Track backwardUsing Track backward flow, participants can access the transactionecord at any time, ensuring the process’s accuracy, integrity, andransparency. To track backwards, rules and regulations are definedor each stakeholder, depending on the proportionality of informationevealed. Access control transaction rules are implemented to regular-ze them throughout the system. Decentralized applications (DApps)8re deployed at point-of-sale units. Consumers can scan QR codes torack back all the information about the required drug. The completeracking process is depicted in Fig. 3. In this process, Sources of activengredients (supplier), manufacturers (owners of drugs), a chain ofransfers (transfer activities), wholesale distributors, re-distributers andOS units are included in the transparency of details. The sold statusf the drug appears as the QR code scanned at retail pharmacies at theime of the transaction, and the transaction details are uploaded alongith the location data.
.2.2. Track forwardTrack forward-protocol is implemented to track all the transactionetails of the drug by all stakeholders. Access control methods definehe attributes of drug transparency for each stakeholder in a system.ransaction families are used to implement access control methodsia smart contracts managed by leader nodes. Transaction rules aremplemented in a modular approach based on the nature of tasks theyill handle and initialize, such as registration, transactions, tracing,nd tracking. Procurement of dynamic QR codes is available on a timelyasis for each drug manufacturer. End-users or regulatory bodies canrackback any drug they need. An end-user can scan a QR code toisplay all the transactions performed on this drug during its completeife cycle. Specific access control methods, defined for each stakeholder,dd additional layers of privacy and security to a system.
1. Drug Regulatory Bodies: Drug regulatory bodies work as ad-ministrator nodes, which can govern the whole system.2. Active Pharmaceutical Ingredient Suppliers (APIs): Each APIsupplier has its validator node and associated components suchas transaction processors. It can only track its orders, man-age transactions with the manufacturers, and request additionaldetails from the drug regulatory bodies. It can use off-chaindistributed storage for additional storage requirements.3. Manufacturers: Manufacturers work as independent validatornodes. They can access the statistics of APIs and associateddistributors by defining the set of transaction families. They needauthorization from other collaborative nodes.4. Wholesale distributors: Wholesale distributors can only accessthe drug details of drugs they will distribute and those alreadydistributed.5. Re-distributors: Re-distributors can track the ongoing deliver-ies booked by them and their transaction status.6. Point-of-sale Unit: Point-of-sale units have client applicationswith access to writing a sale transaction. They are required toscan a QR code to update the status of the drug as sold, alongwith the details of the POS unit, date, and time.7. End-Users: End-users can check a drug’s status, including itsdate of manufacturing, expiration date, and manufacturer, bysimply scanning its QR code.
. Implementation and performance analysis
In this section, we introduce our research problem in the contextf service optimization. PHTrack, represents a drug traceability sys-em which comprises 𝛼𝑥 individual participating nodes. The numberf parallel transactions generated by each participating node will belocked together in databatches (𝑑𝑏), and the size of databatch isenoted as 𝑆𝑡𝑥𝑛 (𝑑𝑏). Each transaction within a blockchain databatchncompasses multiple attributes that collectively determine its size andan be expressed as follows:
txn (𝑑𝑏) = 𝛾flag + 𝛴𝑡elapsed𝑡=𝑡initial (𝑐𝑛𝑡in + 𝑐𝑛𝑡out )
𝑡elapsed + 𝛼here 𝛼 is network id of registered entity node, 𝛾flag is flags data,is initial time, cnt is initial count of input values, cntinitial in out
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.
is output count list, and 𝑡elapsed is total elapsed time. If the numberof transactions in a databatch 𝑑𝑏 are 𝑛𝑡𝑥𝑛 (𝑑𝑏), then the total size oftransaction data in a databatch can be given as:
𝑆𝑡𝑥𝑛 (𝑑𝑏) = 𝑛𝑡𝑥𝑛 (𝑑𝑏) ⋅ 𝑆𝑡𝑥𝑛 (𝑑𝑏)Each databatch, in addition to the transaction data, also carries somemetadata. This metadata includes elements like a databatch header, thenetwork id of the registered entity node, a random wait time, elapsedtime, merkle tree, and a time stamp. As such, the size of a databatchcan be determined based on these factors.
𝑆hdr (𝑑𝑏) = ℎ (𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠𝑏𝑙𝑜𝑐𝑘) + ℎ (𝛼) + ℎ (𝑚𝑒𝑟𝑘𝑙𝑒𝑡𝑟𝑒𝑒)
+ℎ (𝑡𝑜𝑡𝑎𝑙𝑒𝑙𝑎𝑝𝑠𝑒𝑑𝑡𝑖𝑚𝑒) + ℎ (𝑟𝑎𝑛𝑑𝑜𝑚𝑤𝑎𝑖𝑡𝑡𝑖𝑚𝑒) + ℎ (𝑡𝑖𝑚𝑒𝑠𝑡𝑎𝑚𝑝)
In this context, ℎ symbolizes the hash function employed in the hy-perledger sawtooth framework. The term ’previous databatch’ per-tains to the hash value of the databatch preceding the current onein the network. ’Merkle tree’ signifies the hash value encapsulatingall the transactions in the databatch. The ’target formula’ is appliedto compute the target value required for mining a fresh databatch.‘Nonce’ is an acronym for ’Number Only Used Once’, while ‘timestamp’marks the precise time the databatch was generated. Consequently, thecomprehensive size of the metadata for databatch 𝑑𝑏 can be defined as:
𝑆𝑚𝑑 (𝑑𝑏) = 𝑆ℎ𝑑𝑟 (𝑑𝑏) + 𝑆𝑐𝑡𝑟 (𝑑𝑏) ,where 𝑆𝑐𝑛𝑡 (𝑑𝑏) signifies the size of transaction counts. Every databatchin a network encompasses both transaction data and its correspondingmetadata. Hence, by utilizing the previous four equations, the cumu-lative size of the databatch 𝑑𝑏 for the network 𝛽𝑖 can be expressedas:
𝑆
(
𝑑𝑏, 𝛽𝑖
)
= 𝑆𝑡𝑥𝑛 (𝑑𝑏) + 𝑆𝑚𝑑 (𝑑𝑏) ,As previously explained, a databatch is composed of numerous trans-actions. The duration required for a network to achieve consensus ona specific databatch is contingent on the complexity of the consensusalong with network latencies. Where a databatch 𝑑𝑏 is published,it undergoes a propagation period within the 𝛽𝑖. Consequently, thepropagation delay for transmitting a databatch 𝑑𝑏 on network 𝛽𝑖 canbe characterized by:
𝑡𝑝
(
𝑑𝑏, 𝛽i
)
= 𝑡𝑐
(
𝑑𝑏, 𝛽i
)
+ 𝑡𝑝𝑟
(
𝑑𝑏, 𝛽i
)
+ 𝑡𝑞
(
𝑑𝑏, 𝛽i
)
Where, 𝑡𝑐 (𝑑𝑏, 𝛽𝑖) = network delay, 𝑡𝑝𝑟 (𝑑𝑏, 𝛽𝑖) = execution delay, and
𝑡𝑞
(
𝑑𝑏, 𝛽𝑖
)
= wait delay. In addition to the turnaround delay, the totalelapsed time taken for a databatch to complete includes propagationdelay time, contributing to the overall delay before confirmation. Thisturnaround delay, 𝑡𝑠 (𝑑𝑏, 𝛽i), can be represented by:
𝑡𝑠
(
𝑑𝑏, 𝛽i
)
= 𝑡𝑐𝑠
(
𝑑𝑏, 𝛽i
)
+ 𝑡𝑜𝑡
(
𝑑𝑏, 𝛽i
)
.
Where 𝑡𝑐𝑠 (𝑑𝑏, 𝛽𝑖) denotes the encryption plus consensus processingtime, and 𝑡𝑜𝑡 (𝑑𝑏, 𝛽𝑖) represents other potential delays, such as synchro-nization delay or network delay. Consequently, employing the equa-tions mentioned above, the comprehensive network latency 𝑇 (𝑑𝑏, 𝛽𝑖)required for the generation of a data batch can be formulated asfollows:Where, 𝑡𝑐𝑠 (𝑑𝑏, 𝛽i) denotes the encryption plus consensus processingtime, and 𝑡𝑜𝑡 (𝑑𝑏, 𝐵i) represents other potential delays, such as synchro-nization delay or network delay. Consequently, employing the equa-tions mentioned above, the comprehensive network latency 𝑇 (𝑑𝑏, 𝛽𝑖),for a databatch to be generated can be given by:
𝑇
(
𝑑𝑏, 𝛽i
)
= 𝑡𝑝
(
𝑑𝑏, 𝛽i
)
+ 𝑡𝑠
(
𝑑𝑏, 𝛽i
)
.
Consequently, to attain optimal performance within a network, themathematical representation of the optimization problem can be ex-pressed as follows:
min 𝑇
(
𝑑𝑏, 𝛽
)
&max𝑆
(
𝑑𝑏, 𝛽
)
9
𝑖 𝑖Table 4Hardware and software specifications.Component Description
Framework Hyperledger SawtoothOperating System Ubuntu 18.04.6 LTS x64Server Specification PC Core i5-6300U @ 2.5 GHzAWS Server specs EC2 instance: t2.LargeClient Application Nodejs ServerRuntime Environment Docker-compose v.1.29.2
Fig. 5. CPU Utilization in terms of CPU percentage.
Thus, this optimization problem provides a mathematical foundationfor determining the most efficient combination of databatch size, trans-actions per second (tps), resource utilization and network resourceswhile ensuring the reliability of transaction commitment. To evalu-ate the platform performance, two separate testnets of the proposedPHTrack framework were designed and deployed with the specifica-tions written in Table 4. Permissionless PoET consensus engine in devmodes is used. A Client App was used to implement the applicationlayer. Nodejs was used as the client-side execution engine and dockercontainers for the server side.In our testnets, the complete hyperledger sawtooth system was in-stalled and initiated on a Linux-Ubuntu operation system as a local hostcommunication. An EC2 instance of AWS was used for separate servercommunications. Each component of sawtooth, including transactionprocessors, events, chain codes, and different nodes, was launched asdocker containers. Sawtooth is implemented using Python embedded indocker images. Each node is only responsible for validating transactionsrelated to it and committing them in batches after validation by thevalidator. In addition to the REST API, transaction processor, validator,and data from blocks and states, every validator node has severalcontainers.Each stakeholder application has a different interface, which de-pends on the rights it owns, and leader nodes can access every moduleand update the entire system. Each validator node can prescribe itsown set of rules and define roles for its sub-system modules and off-chain settings. Interfaces can add transactions, accept inputs, retrievetransactions, and log information. Transaction processors are server-side programs that store operations and process transaction submissionsaccording to business logic. Validators run all the validation processesthat happen in a node. They manage everything from business logicvalidation by the transaction processor to consensus validation by theconsensus manager. A genesis batch, the first block of the chain, wascreated by the first validator node. Upcoming data batches appendthis chain by adding new batches. Any validator node can start thegenesis node if it does not acquire any extra authority. All authoritativeprotocols, terms, and conditions can be added later at any time duringthe running system via chain codes.
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.Fig. 6. Impact of blocksize and transaction rate on throughput.Fig. 7. Incoming and outgoing Network traffic Packets.
To determine the business logic for the Traceability System, twotransaction processors (TPs) were used, namely stakeholder-registration-TP and transaction-request-TP, and the PoET development modeconsensus was used. Successful transaction requests are combined intodata batches as shown in Fig. 10. Multiple transaction batches arecombined to make one block.
5.1. Impact of TPS and blocksize on latency
Firstly, we evaluated the impact of blocksize by varying the trans-action rate (transaction per second, tps) in terms of transaction latency.
10Transaction rate refers to the number of transactions processed by thesawtooth network within a specific time period 𝑇 . Latency is defined asthe time gap between the transaction submission and completion. Thelatency 𝐿𝑡 is calculated as follows.
𝐿𝑡 = 1
𝑛
𝑛∑
𝑖=1
(
𝑡𝑥𝑖 − 𝑡𝑦𝑖
) 1
𝑇𝑖Here, 𝑡𝑥𝑖 and 𝑡𝑦𝑖 represent the transaction completion and submissiontimes, respectively, of the 𝑖th experiment and 𝑇 𝑖 denotes the totalnumber of transactions submitted for the 𝑖th experiment. It measuresthe rate at which transactions are submitted to and executed by the net-work. Transaction latency decreased linearly by decreasing the parallelnumber of transactions, which can be seen clearly in Fig. 8.With a low transaction rate, it is easier for the network to processand confirm transactions quickly. Latency tends to be lower in suchcases, as there is less competition for block space, and transactions canbe included in blocks sooner. However, a high transaction rate meansmany transactions are being submitted simultaneously. This can leadto increased competition for block space, potentially causing delays intransaction confirmation. Higher transaction rates may result in higherlatency due to the time it takes to accumulate and validate transactionsfor block inclusion.Proper network optimization and scaling can help mitigate thelatency impact of varying block sizes and transaction rates. Techniquessuch as load balancing, parallel processing, and optimized consensusalgorithms can help maintain lower latency even as the blockchainnetwork faces increased traffic. The availability of network resources,such as computing power, memory, and bandwidth, plays a role indetermining how well a sawtooth network can handle different blocksizes and transaction rates. Networks with abundant resources canhandle larger blocks and higher transaction rates more efficiently,resulting in lower latency. We need dynamic adjustment of block sizeand other network parameters. This adaptability can help balance thetrade-off between throughput, block size and latency in response tochanging network conditions.
5.2. Impact of transaction rate on throughput
The transaction rate directly and significantly impacts throughput insawtooth network. Throughput measures the system’s ability to processa certain number of transactions within a specified time period. Tovalidate our proposed model, we create three categories of blocksizesand run parallel transactions per second. Fig. 6 depicts that transactionthroughput increases rapidly by increasing the transaction rate untilit reaches 120–130 tps while using blocksize-20, at which point itstops increasing significantly, and only a slight difference can be seen.These results indicate that bigger blocksizes show better performancethroughput until they reach their maximum throughput per block size.As an average, transaction throughput is higher in blocksize-20 thanin 5 and 50, which shows that the more often the consensus algorithm
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.Fig. 8. Transaction latency (sec) increases linearly by increasing transactions persecond (tps).
runs, the more often it decreases the overall efficiency of the systembut we need to set the limit dynamically according to the availabil-ity of the resources. The impact on throughput also depends on theavailable network resources, such as bandwidth. If limited, increasingthe transaction rate may not significantly increase throughput. Thetransaction rate increases significantly when we move towards 5 to 100transactions per second and then increases in a non-significant style.Latency can also be decreased by increasing the number of transactionsper block. In the case of blocksize-5, the average transaction rate was306.89 ms, which increased to 270.47 ms by increasing the blocksizeto 20 transactions per block. The transaction latency decreases furtherby increasing the size to 50 transactions per block, the transaction ratecomes to 198.68 ms average. The experimental results of
5.3. Network reliability analysis
More than 1000 parallel statements were initialized and tracked totest the system’s reliability; as shown in Fig. 9, the system could handle82 percent of the statements and miss only 179 requests with limitedcomputational resources. These sets of statements include differentkinds of commands and events, but missed statements are network-bound only.Node.js based client was responsible for initiating and deploying adifferent number of transactions by using Int-key transaction processor.These transactions are then sent to the validator through the REST-API module. Peer-to-peer nodes, as well as client applications, useRESt-API to communicate between them. RESt-API uses google protocolbuffers to make communication easier, and curl sends data from HTTPto RESt-API. Int-key transaction processor was also used to sign eachtransaction and combine these transactions into batches. The numberof batches is combined to publish one block on the network. CPUand network utilization performance test marks were recorded andanalyzed for Amazon AWS instances and personal computer systems.Hardware specifications are described in detail in Table 4. To collectthe matrices of PC, cAdvisor (Container Advisor) is used, and a runningdaemon is used to collect the analytics against each running containerseparately. 11 virtual cores were used to start the network.To track the performance metrics of AWS cloud instances, Cloud-Watch was used. We aggregate these performance parameters of run-ning containers and their impact on CPU and network consumption incharts. Fig. 5 describes the CPU utilization in terms of CPU percentage,which shows sudden spikes over time during the parallel number oftransaction commits. sawtooth is mainly a network-bound protocol,Fig. 7(a) and 7(b) show the network packet count used during trans-action requests and event handling using AWS EC2 instance. Suddenspikes show the network usage during transactions and event han-dling. The following analysis demonstrates that the suggested system
11Fig. 9. Total number of successful commits.
effectively maintains privacy, without introducing any disturbancesin the areas of traceability and ownership. Moreover, the integra-tion of privacy measures does not significantly influence the system’sperformance.
5.4. On and off-chain secure communication
The serialization model of sawtooth demonstrates remarkable flex-ibility, as it can seamlessly leverage both on-chain and off-chain de-centralized storage systems. For on-chain communication, sawtooth’sinternal ZMQ based communication model does not require any authen-tication and passes the message requests to the validator for signatureverification. In the context of drug supply chain traceability, wherenumerous industrial entities participate on a large scale, there arisesa need for direct communication between two or more entities withoutinvolving the entire network. For instance, some participating entitiesmay seek to enhance their functionality by gaining additional insightsfrom others, such as real-time data-driven recommendation systems.These entities have the capability to independently send and receivedata within the network through the use of chain codes. Moreover, par-ticipating entities can autonomously update their terms and conditionswithout requiring the involvement of other network members.To facilitate this off-chain communication within the network, wepropose the implementation of a secure communication infrastructurebased on Transport Layer Security (TLS) for encrypted data sharing.This approach incorporates post-quantum essential encapsulation meth-ods to ensure the security of private keys and employs digital signaturealgorithms for authentication purposes. Kyber 768 is utilized as a keyencapsulation method for sharing crypto keys over the network anddilithium3 for digital signatures. One of the NIST finalists, Kyber isa family of post-quantum key exchange algorithms. It is designed toestablish secure communication channels by exchanging cryptographickeys that are resistant to attacks by quantum computers. Kyber stan-dardized three security levels of 515,768, and 1024. Kyber768 provideNIST security level 3, which is equal to AES 192. This network utilizeskyber768 for key encapsulation before the off-chain key-sharing pro-cess. Dilithium signatures help guarantee data integrity by providinga way to detect any unauthorized modifications to the content of amessage. This is particularly important in critical applications whereeven small changes to data can have serious consequences. In industrialand legal contexts, digital signatures are also required to meet regula-tory compliance and legal standards. Dilithium provides a secure andlegally recognized means of signing digital documents and communi-cations, making it valuable in these settings. Dilithium is employed incommunication to ensure messages’ confidentiality, authenticity, and
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.Fig. 10. Data batches created after transaction.Fig. 11. Latency in (ms) for kyber76 and dilithium3 process.
integrity, especially in a world where the threat of quantum computinglooms. Its post-quantum security properties make it a critical choice tosecure sensitive and important communications against future quantumattacks. Fig. 11, shows the latencies (ms) we got using kyber678 andDilithium3 using TLS-based secure communication over wi-fi. In theoff-chain communication paradigm, the involved parties will assumeresponsibility for confirming a transaction (the actual execution) andcertifying the agreement. This chain of communication can be executedinstantly in comparison to on-chain transactions.
5.5. Security and privacy analysis
Sawtooth’s security framework inherits the utilization of policiesand roles within the identity transaction family. Policies, representedby sets of key rules, are assessed sequentially, while roles grant specificauthorizations for operations and data access. Transaction key permis-sions further enhance granularity, with roles controlling client accessbased on the signing public key, extending permissions to specifictransaction families (TFs) or entire batches. Additionally, Sawtooth em-ploys challenge-response authorization, requiring nodes to prove theiridentity through signed nonces, and supports encryption for variousaspects, securing data in transit through ZeroMQ with CurveZMQ.For off-chain communication and storage in distributed places, Ky-ber key encapsulation algorithm (KEM) with the Kyber public-key
12encryption algorithm is implemented. 32-octet cpaseed is used as inputfor the key generation process. 𝑐𝑝𝑎𝑃𝑢𝑏𝑙𝑖𝑐𝐾𝑒𝑦, 𝑐𝑝𝑎𝑃 𝑟𝑖𝑣𝑎𝑡𝑒𝐾𝑒𝑦, ℎ, and
𝑧 are variables representing the components needed to derive theprivate key. The publicKey is derived directly from 𝑐𝑝𝑎𝑃𝑢𝑏𝑙𝑖𝑐𝐾𝑒𝑦,while the privateKey is derived from a combination of 𝑐𝑝𝑎𝑃 𝑟𝑖𝑣𝑎𝑡𝑒𝐾𝑒𝑦,
𝑐𝑝𝑎𝑃𝑢𝑏𝑙𝑖𝑐𝐾𝑒𝑦, ℎ, and 𝑧. In terms of resource utilization, the perfor-mance of post-quantum Key Encapsulation Mechanisms (KEMs) canbe compared against elliptic curves implemented in the 𝑂𝑝𝑒𝑛𝑆𝑆𝐿cryptography library. The results show that, in general, the ellipticcurves perform similarly to the post-quantum KEMs, indicating theneed for further optimizations to make them more suitable for deviceswith low resources. However, the performance study can be extendedto include all currently competing KEMs, including those with highersecurity settings and code-based KEMs. Various type of attacks such asmulti-certificate attacks, double spending, 51% Attacks, phising, sybiland quantum Computing Threats have been eliminated by leveragingsawtooth transaction processors, PoET consensus and post-quantumcryptography.On the privacy front, PoET is utilized to ensure permissioningcapabilities enable the creation of private networks with distinct accesscontrols, safeguarding sensitive transaction patterns and confidentialinformation from unauthorized exposure. Core transaction families,including identity transaction family, facilitate identity managementand on-chain permissioning. The platform’s modular design enablesthe configuration and extension of these security and privacy fea-tures, providing a robust foundation for secure permissioned blockchainnetworks tailored to enterprise needs.
6. Discussion
The practice of counterfeiting drugs is widespread, which under-mines the confidence of end-users in the drug industry. In additionto being time-consuming, costly, and cumbersome, legacy drug ver-ification systems pose many security risks. In this article, we inves-tigated the challenge of drug traceability within drug supply chainsand demonstrated its importance in preventing counterfeit drugs. Totrack and trace drugs in a decentralized way, we have developedand evaluated PHTrack, a distributed ledger-based solution leveraginghyperledger sawtooth for the drug supply chain. By using crypto-graphic principles embedded in permissioned blockchain technology,we achieve tamper-proof logs of events within the supply chain and
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.
WIt
Dautomate the recording of events in the hyperledger sawtooth environ-ment for access by all participating stakeholders. The highly modularsawtooth structure enables each stakeholder to define their systemspecifications based on their requirements. PHTrack provides systemtransparency, immutability, and reliability, and enables trust in multi-ple stakeholder systems by eliminating third-party services. Enablingtraceability and tracking all transactions in the drug supply chainlifecycle can provide a complete picture of each transaction, includingthe date, time, location, and stakeholders involved in the operation.
6.1. Theoretical contributions
Integrating Hyperledger sawtooth in the drug supply chain bringsabout several theoretical implications. Firstly, the decentralized natureof hyperledger sawtooth ensures that no single entity controls the entiresupply chain, fostering increased trust among participants. This decen-tralization aligns with theoretical perspectives on decentralized systemsand trust within supply chain management. Using the unique PoETconsensus algorithm in sawtooth enhances the security and efficiency oftransaction validation, a critical aspect in maintaining the integrity oftransactions within the drug supply chain. Additionally, the immutableledger of sawtooth ensures transparent and traceable records of alltransactions, crucial for verifying the authenticity of pharmaceuticalproducts and preventing the infiltration of counterfeit goods into thesupply chain.Furthermore, smart contracts within sawtooth enable automatingand enforcing predefined business rules, aligning with theories em-phasizing automation and efficiency in supply chain management. Theblockchain technology underlying sawtooth also contributes to secu-rity and immutability, creating a tamper-resistant system that reducesthe risk of counterfeit drugs. The theoretical underpinning of a DLTsystem involves ensuring data consistency and synchronization acrossmultiple participants, reducing discrepancies and potential errors in thepharmaceutical supply chain. These theoretical implications highlightthe potential for enhanced transparency, security, and efficiency in thepharmaceutical supply chain by adopting hyperledger sawtooth.
6.2. Managerial implications
Adopting blockchain technology, specifically hyperledger sawtooth,has significant managerial implications for each participating stake-holder in drug supply chain. Utilizing hyperledger sawtooth can leadto substantial cost savings for consumers by removing intermediaries,thus reducing administrative expenses. Faster payment processes andminimized errors also contribute to economic efficiency at variouslevels, including API suppliers, manufacturers, re-distributers, and re-tailers. Operational efficiency can be greatly improved by automatingtransaction recording and verification, freeing staff to focus on strategictasks. The transparency and traceability provide detailed recordingand tracking of transactions, which enhances risk management byfacilitating the early detection of irregularities for regulatory bodies.Hyperledger sawtooth’s ability to provide an immutable record oftransactions is particularly beneficial for industries with stringent reg-ulations, like pharmaceuticals, aiding compliance efforts. Managerially,the reduction in time and costs associated with traditional supplychain management, such as paperwork processing, is significant. Thesystem’s automated verification processes ensure that pharmaceuticalbatches meet necessary conditions before moving through the supplychain, increasing efficiency and ensuring adherence to quality stan-dards and regulatory requirements. Real-time monitoring capabilitiesof hyperledger sawtooth allow for better tracking of drug movementfrom manufacturing to delivery, improving inventory management and
13
allowing for prompt responses to potential issues.6.3. Conclusion and future work
Our proposed system consists of a private permissioned hyperledgersawtooth framework which provides the complete system flow. Theproposed network is divided into core system modules and applicationmodules. Stakeholders play a key role in the sawtooth network asparticipants, and their roles within the supply chain determine theirroles. Additionally, they can access on-chain resources such as loggingand history information to track transactions. Furthermore, they canaccess decentralized data such as images and complete informationlogs. In addition, it offers low-cost off-chain storage to ensure scalabilityand immutability by storing supply chain transactions in hashed form.It is important to keep the integrity of data, so for each uploadedfile, the server hashes it and stores the hashes on the blockchain.Chain codes can be used to get the hashes from the blockchain. Weemploy smart contracts and chaincodes to streamline the verificationand validation of PHTrack transactions, covering the entire journeyfrom the initial acquisition to the delivery process. This guaranteesthe seamless exchange of transaction information among all partici-pating stakeholders, adhering to the protocols established within theHyperledger Sawtooth framework.We demonstrated the performance of the proposed framework bytesting various significant parameters, which include the impact ofblocksize on latency by varying the transaction rate per second andthe impact of blocksize and transaction rate on throughput. Exper-imental results show that transaction latency decreases linearly byincreasing the blocksize and transaction rate (tps). Parallel transactionsincrease the system’s overall performance metrics by decreasing latencyand increasing transaction throughput. After reaching its maximumtransaction throughput, it does not show a significant rise and thenstarts decreasing gradually. Network bandwidth and CPU consumptionper core were also observed by using cAdvisor and AWS cloudwatch.CPU utilization metrics show that there is not much change duringtransaction handling or in idle conditions. However, sudden spikes innetwork packet count have been observed during transaction process-ing, and idle state systems do not require significant bandwidth, makingsawtooth network protocol network-bound rather than CPU bound. Tomake quantum secure off-chain communication, we utilized kyber768and dilithium3 algorithms.The experimental results demonstrated that PHTrack offers a reli-able, all-encompassing drug provenance system and real-time supplychain traceability capabilities. Our proposed system provides profoundtheoretical and managerial implications for drug supply chains, withbenefits ranging from improved transparency and regulatory compli-ance to cost savings, increased operational efficiency, and enhancedrisk management.We gathered information on our model’s performance through thebenchmarking tools, which can be used as a basis for future discussions.As part of our ongoing efforts to enhance the efficiency of drug supplychains, we will incorporate case study-based research on generating in-creased value for stakeholders by incorporating automated transactionprocessors for each core module.
CRediT authorship contribution statement
Anum Nawaz: Conceptualization, Data curation, Methodology,riting – original draft. Liguan Wang: Methodology. Muhammadrfan: Writing – review & editing. Tomi Westerlund: Conceptualiza-ion, Supervision, Writing – review & editing.
ata availability
Data will be made available on request.
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.
A
A
A
A
A
A
A
B
M
M
M
N
N
N
N
N
O
O
P
P(
S
S
S
S
S
U
V
V
V
W
W
WReferences
Abdallah, S., & Nizamuddin, N. (2023). Blockchain-based solution for pharma supplychain industry. Computers & Industrial Engineering, 177, Article 108997.Abdellatif, A. A., & Al-Marridi (2020). Sshealth: Toward secure, blockchain-enabledhealthcare systems. IEEE Network, 34(4), 312–319. http://dx.doi.org/10.1109/MNET.011.1900553.grawal, T. K., & Angelis (2022). Demonstration of a blockchain-based framework usingsmart contracts for supply chain collaboration. International Journal of ProductionResearch, 1–20.li, I., & Kannan, D. (2022). Mapping research on healthcare operations and supplychain management: a topic modelling-based literature review. Annals of OperationsResearch, 315(1), 29–55.ltaf, A. (2023). A survey of blockchain technology: Architecture, applied domains,platforms, and security threats. Social Science Computer Review, 41(5), 1941–1962.lzahrani, N., & Bulusu, N. (2016). Securing pharmaceutical and high-value productsagainst tag reapplication attacks using nfc tags. In 2016 IEEE international conferenceon smart computing (pp. 1–6). IEEE.ndroulaki, E., Barger, A., Bortnikov, V., Cachin, C., Christidis, K., De Caro, A., etal. (2018). Hyperledger fabric: a distributed operating system for permissionedblockchains. In Proceedings of the thirteenth euroSys conference (pp. 1–15).ytekin, A., Görçün, Ö. F., Ecer, F., Pamucar, D., & Karamaşa, Ç. (2023). Evaluation ofthe pharmaceutical distribution and warehousing companies through an integratedFermatean fuzzy entropy-WASPAS approach. Kybernetes, 52(11), 5561–5592.zevedo, & Gomes (2023). Supply chain traceability using blockchain. OperationsManagement Research, [ISSN: 1936-9743] 16(3), 1359–1381. http://dx.doi.org/10.1007/s12063-023-00359-y.enji, M., & Sindhu, M. (2019). A study on the corda and ripple blockchain platforms.In Advances in big data and cloud computing (pp. 179–187). Springer.Buterin, V., et al. (2013). Ethereum white paper. GitHub Repository, 1, 22–23.Commissioner (2023). FDA protects patients from harmful drugs through theDSCSA, URL https://www.fda.gov/news-events/fda-voices/fda-protects-patients-harmful-drugs-through-drug-supply-chain-security-act.Corso, A. (2019). Performance analysis of proof-of-elapsed-time (poet) consensus in thesawtooth blockchain framework (Ph.D. thesis), University of Oregon.Dasaklis, T. K., Voutsinas, T. G., Tsoulfas, G. T., & Casino, F. (2022). A systematicliterature review of blockchain-enabled supply chain traceability implementations.Sustainability, 14(4), 2439.Dauvergne, P. (2022). Is artificial intelligence greening global supply chains? Exposingthe political economy of environmental costs. Review of International PoliticalEconomy, 29(3), 696–718.de Ocáriz Borde, H. S. (2022). An overview of trees in blockchain technology: Merkletrees and Merkle patricia tries.Dhillon, V., Metcalf, D., & Hooper, M. (2017). The hyperledger project. In Blockchainenabled applications (pp. 139–149). Springer.Dwivedi, S. K., Amin, R., & Vollala, S. (2020). Blockchain based secured informa-tion sharing protocol in supply chain management system with key distributionmechanism. Journal of Information Security and Applications, [ISSN: 2214-2126] 54,Article 102554. http://dx.doi.org/10.1016/j.jisa.2020.102554, URL https://www.sciencedirect.com/science/article/pii/S2214212620301484.Grigg, I. (2017). Eos-an introduction. White paper, https://whitepaperdatabase.com/eos-whitepaper.Hassija, & Vikas (2020). A survey on supply chain security: Application areas, securitythreats, and solution architectures. IEEE Internet of Things Journal, 8(8), 6222–6246.Helliar, C. V., Crawford, L., Rocca, L., Teodori, C., & Veneziani, M. (2020). Permis-sionless and permissioned blockchain diffusion. International Journal of InformationManagement, 54, Article 102136.Kaiser-Kershaw, S. (2022). Helping pharma help people: New NFC Tags Boost Safetyand adherence, URL https://www.nxp.com/company/blog/helping-pharma-help-people-new-nfc-tags-boost-safety-adherence:BL-HELPING-PHARMA.Karuppiah, K., Sankaranarayanan, B., & Ali, S. M. (2023). A decision-aid model forevaluating challenges to blockchain adoption in supply chains. International Journalof Logistics Research and Applications, 26(3), 257–278.Khan, S. A., Gupta, H., Gunasekaran, A., Mubarik, M. S., & Lawal, J. (2023). Ahybrid multi-criteria decision-making approach to evaluate interrelationships andimpacts of supply chain performance factors on pharmaceutical industry. Journalof Multi-Criteria Decision Analysis, 30(1–2), 62–90.Killer, C., Rodrigues, B., & Stiller, B. (2019). Security management and visualization ina blockchain-based collaborative defense. In 2019 IEEE international conference onblockchain and cryptocurrency (pp. 108–111). IEEE.King, B., & Zhang, X. (2007). Securing the pharmaceutical supply chain using RFID. In2007 international conference on multimedia and ubiquitous engineering (pp. 23–28).IEEE.Kordestani, A., Oghazi, P., & Mostaghel, R. (2023). Smart contract diffusion in thepharmaceutical blockchain: the battle of counterfeit drugs. Journal of BusinessResearch, 158, Article 113646.Liu, M., Wu, K., & Xu, J. J. (2019). How will blockchain technology impact auditingand accounting: Permissionless versus permissioned blockchain. Current Issues inAuditing, 13(2), A19–A29.14Malik, N., & Alkhatib (2021). A comprehensive review of blockchain applications inindustrial Internet of Things and supply chain systems. Applied Stochastic Models inBusiness and Industry, 37(3), 391–412.Melnyk, S. (2022). New challenges in supply chain management: cybersecurity acrossthe supply chain. International Journal of Production Research, 60(1), 162–183.Meng, X., Liang, W., Xu, Z., & Li, K.-C. (2022). A lightweight authentication protocolfor NFC-enabled drug anti-counterfeiting system. In 2022 IEEE 24th int conf on highperformance computing & communications; 8th int conf on data science & systems; 20thint conf on smart city; 8th int conf on dependability in sensor, cloud & big data systems& application (HPCC/DSS/smartCity/dependSys) (pp. 516–522). IEEE.ohit, M., Kaur, S., & Singh, M. (2022). Design and implementation of transactionprivacy by virtue of ownership and traceability in blockchain based supply chain.Cluster Computing, 25(3), 2223–2240.oosivand, A., Ghatari, A. R., & Rasekh, H. R. (2019). Supply chain challengesin pharmaceutical manufacturing companies: using qualitative system dynamicsmethodology. Iranian Journal of Pharmaceutical Research: IJPR, 18(2), 1103.usamih, A., Salah, K., Jayaraman, R., Arshad, J., Debe, M., Al-Hammadi, Y., et al.(2021). A blockchain-based approach for drug traceability in healthcare supplychain. IEEE Access, 9, 9728–9743.akamoto, S. (2008). Bitcoin: A peer-to-peer electronic cash system. DecentralizedBusiness Review.awaz, et al. (2019). Edge AI and blockchain for privacy-critical and data-sensitive ap-plications. In 2019 twelfth international conference on mobile computing and ubiquitousnetwork (pp. 1–2). http://dx.doi.org/10.23919/ICMU48249.2019.9006635.awaz, et al. (2020). Edge computing to secure iot data ownership and trade with theethereum blockchain. Sensors, 20(14), 3965.guyen, A., Lamouri, S., Pellerin, R., Tamayo, S., & Lekens, B. (2022). Data analyticsin pharmaceutical supply chains: state of the art, opportunities, and challenges.International Journal of Production Research, 60(22), 6888–6907.guyen Gia, T., Nawaz, A., Peña Querata, J., Tenhunen, H., & Westerlund, T. (2019).Artificial intelligence at the edge in the blockchain of things. In Internationalconference on wireless mobile communication and healthcare (pp. 267–280). Springer.ngaro, D., & Ousterhout, J. (2015). The raft consensus algorithm. Lecture Notes CS,190, 2022.nieva, E., Osaba, E., Angulo, I., Moreno, A., Bahillo, A., & Perallos, A. (2015).Improvement of drug delivery routes through the adoption of multi-operator evo-lutionary algorithms and intelligent vans capable of reporting real-time incidents.IEEE Transactions on Automation Science and Engineering, 14(2), 1009–1019.andey, M., Velmurugan, M., Sathi, G., Abbas, A. R., Zebo, N., & Sathish, T. (2023).Blockchain technology: Applications and challenges in computer science. Vol. 399,In E3S web of conferences (p. 04035). EDP Sciences.aul, M. B. (2023). Concept of firebase. Intelligent Electrical Systems, 44.2016). Pharmaceutical sector - observatory, URL https://euipo.europa.eu/ohimportal/en/web/observatory/ipr-infringement-pharmaceutical-sector.aberi, S., Kouhizadeh, M., Sarkis, J., & Shen, L. (2019). Blockchain technology andits relationships to sustainable supply chain management. International Journal ofProduction Research, 57(7), 2117–2135.araji, M. K., Rahbar, E., Chenarlogh, A. G., & Streimikiene, D. (2023). A sphericalfuzzy assessment framework for evaluating the challenges to LARG supply chainadoption in pharmaceutical companies. Journal of Cleaner Production, 409, Article137260.harma, N., & Rohilla, R. (2023). A novel Hyperledger blockchain-enabled decen-tralized application for drug discovery chain management. Computers & IndustrialEngineering, 183, Article 109501.ingh, A., Click, K., Parizi, R. M., Zhang, Q., Dehghantanha, A., & Choo, K.-K. R. (2020).Sidechain technologies in blockchain networks: An examination and state-of-the-artreview. Journal of Network and Computer Applications, 149, Article 102471.ylim, P., Liu, F., Marcelo, A., & Fontelo, P. (2018). Blockchain technology fordetecting falsified and substandard drugs in distribution: pharmaceutical supplychain intervention. JMIR Research Protocols, 7(9), Article e10163.ddin, M. (2021). Blockchain Medledger: Hyperledger fabric enabled drug traceabilitysystem for counterfeit drugs in pharmaceutical industry. International Journal ofPharmaceutics, 597, Article 120235.akilinia, I., Wang, W., & Xin, J. (2023). An incentive-compatible mechanism fordecentralized storage network. IEEE Transactions on Network Science and Engineering.alizadeh, J., Boloukifar, S., Soltani, S., Hookerd, E. J., Fouladi, F., Rushchtc, A. A., etal. (2023). Designing an optimization model for the vaccine supply chain duringthe COVID-19 pandemic. Expert Systems with Applications, 214, Article 119009.ijay, & Priya (2022). Grainchain–agricultural supply chain traceability and man-agement technique for farmers sustainability using blockchain hyper ledger.International Journal of Intelligent Systems and Applications in Engineering, 10(3),141–146.amba, S. F., & Queiroz, M. M. (2020). Blockchain in the operations and supply chainmanagement: Benefits, challenges and future research opportunities. InternationalJournal of Information Management, 52, Article 102064.ang, Q. (2022). Exploring web3 from the view of blockchain. arXiv preprint arXiv:2206.08821.ang, M., Duan, M., & Zhu, J. (2018). Research on the security criteria of hashfunctions in the blockchain. In Proceedings of the 2nd ACM workshop on blockchains,cryptocurrencies, and contracts (pp. 47–55).
Computers & Industrial Engineering 190 (2024) 110021A. Nawaz et al.
Wang, R., Ye, K., Meng, T., & Xu, C.-Z. (2020). Performance evaluation on blockchainsystems: a case study on Ethereum, Fabric, Sawtooth and Fisco-Bcos. In Servicescomputing–SCC 2020: 17th international conference, held as part of the services con-ference federation, SCF 2020, honolulu, HI, USA, September 18–20, 2020, proceedings17 (pp. 120–134). Springer.Wisessing, K., & Vichaidis, N. (2022). IoT based cold chain logistics with blockchainfor food monitoring application. In 2022 7th international conference on business andindustrial research (pp. 359–363). IEEE.15Xu, X., Zhu, D., Yang, X., Wang, S., Qi, L., & Dou, W. (2021). Concurrent practicalbyzantine fault tolerance for integration of blockchain and supply chain. ACMTransactions on Internet Technology (TOIT), 21(1), 1–17.Ziavrou, K. S., Noguera, S., & Boumba, V. A. (2022). Trends in counterfeit drugsand pharmaceuticals before and during COVID-19 pandemic. Forensic ScienceInternational, Article 111382.