An Analytic Hierarchy Process for Selection of Blockchain-Based Platform for Product Lifecycle Management

Abstract

Blockchain technology has disrupted traditional business processes and hence gained significant attention and popularity in recent years. Consequently, a number of blockchain-based platforms are available today that offer vast applications across multiple sectors and industries. Implementing these blockchain-based platforms as an alternative to traditional product lifecycle management systems (PLMs) is one of the applications. However, before any platform is adopted, its nature, functionalities, and adaptability need to be clearly defined, evaluated, and verified. In this context, the proposed work explores the available blockchain-based platforms that can be used for the purpose of product lifecycle management. We then apply one of the multi-criteria decision-making techniques, i.e., the analytic hierarchy process (AHP), to select the best possible blockchain-based platform for PLM. As transaction speed, data privacy, and scalability are our prime concerns in PLM, we only considered the permissioned (private) blockchain platforms as available alternatives in the final selection process. Results achieved on the basis of considered criteria show that Hyperledger Fabric is the top-ranked among available alternatives to be used for PLM. Furthermore, as blockchain is a new technology, a clear comparison of the available platforms based on the performance-based metrics and key performance indicators is not completely matured and is still in the development stage. However, our proposed approach can be considered an attempt to create a procedure for evaluating blockchain-based platform implementation in any sector.

1. Introduction

Production industries have been struggling a great deal for a long time to develop and support their products throughout their lifecycle. Throughout this duration, their prime focus in product development has been to optimize their product design and improve its intended functions. However, in today’s competitive environment, the aim is not just design optimization and function improvements, but the ability to connect and accordingly update their products to conform to the current market trends. This is only possible if the industry grabs all the opportunities coming along with breakthrough technologies and, hence, properly manages the complete lifecycle of its products. To this aim, several software systems have been implemented in industries. According to one of the recent surveys, a total of 135 such software are currently implemented in different sectors for managing the various aspects of the product lifecycle from conceptualization to its disposal stage [1]. This existing software provides both on-cloud and on-premises services and contains various capabilities and challenges.

Product lifecycle management (PLM) focuses on data creation, storage, and retrieval throughout the lifecycle of a product from its conceptualization all the way to its ultimate disposal or retirement [2]. Its purpose is to provide an integrated view of what is happening with the products across their lifecycle. In this strategy, efficient integration of the products, associated stakeholders, working processes, and data storage systems is required to obtain the best results in product data management. PLM evolution has played an essential role in high-tech industry leaders’ decisions to change their way of work to fulfil critical requirements due to high market changes [3]. The benefits obtained through the implementation of the PLM solution have encouraged industries to invest in PLMs [4]. J.G. Enríquez et al. developed a quality evaluation model for evaluating the different aspects of the PLM systems implemented in industries [5]. The introduction of PLM systems into the industry has both short-term benefits, i.e., reduction of time for each routine activity, reduction of overhead activities, and improvements to the way of work, as well as long-term benefits, i.e., improved competitive position, reduction of time to market, etc. [6].

The existing PLMs, i.e., Teamcenter, Windchill, Enovia, SAP, etc., are implemented based on a standalone and centralized strategy. They are mainly developed for in-house use (mostly manufacturing) and hence are incapable of accessing, processing, and analyzing data across other stages of the product lifecycle. For instance, an industry producing a product, for example, a production machine or equipment, and using the existing PLM can access and analyze the product data only in the manufacturing stage and is unable to access the data of that product when it goes to another stage, i.e., operations and maintenance where it performs its intended function. Hence, as the product information chain along the lifecycle spans enterprise boundaries, these PLMs are unable to meet all the data management requirements of the product [7]. Interoperability, data transparency and openness, decentralization, credibility, collaborative data provision, and data security are some of the broadly discussed major challenges faced by the currently implemented PLM systems in the industry [1,6,7,8,9]. To meet all these challenges and expectations of the production industry, novel blockchain-based frameworks and solutions have been proposed by researchers to manage the entire lifecycle of a product [1,4,10]. Nowadays, an increasing number of initiatives in the blockchain are opening new horizons for its implementation in various sectors. Although most of the current work on the blockchain in the real business environment is still in an early stage [11], it is strongly expected by experts that blockchain will considerably target every industry and significantly change existing practices [12,13,14].

Blockchain is a distributed and immutable ledger that facilitates the process of recording transactions and tracking assets in a business network [15]. The transactions are recorded in a data format called block and added to a shared database in chronological order to form a chain [12,16]. The transaction data in the block are validated by the pre-defined nodes in the network and then added and stored on the chain [13,17,18]. The uses of blockchain have been investigated for different purposes across various industries. Three basic themes, i.e., impact, functions, and configuration, of blockchain implementation have been discussed and expanded by various researchers and practitioners [19,20]. In parallel, a substantial amount of research is in progress on identifying blockchain implementation enablers and challenges. Samuel and Babak developed a fuzzy-based analytic model to identify the key blockchain adoption enablers and analyze their impact on supply chain performance [21]. The identification and evaluation of blockchain adoption challenges is not an easy task and, therefore, needs an integrated approach. In this context, Karuppiah et. al. used fuzzy the Delphi-assisted grey-DEMATEL analytic method to identify the eminent challenges in blockchain adoption [22].

Four generations of blockchain, i.e., blockchain 1.0 to 4.0, have been recognized since the inception of Bitcoin by Satoshi Nakamoto in 2008 [23,24,25].

Table 1. Evolution from Blockchain 1.0 to 4.0.

Generation	Theme	Meaning	Popular Platform	Source
Blockchain 1.0	Cryptocurrency	Using Blockchain for Cryptocurrency	Bitcoin	[24,25,30]
Blockchain 2.0	Smart Contracts	Using Blockchain for Smart Contracts	Ethereum	[24,25,30]
Blockchain 3.0	DApps	Running blockchain in a P2P network	BitMessage, IOTA	[24,25]
Blockchain 4.0	Industry 4.0	Blockchain application in Industry 4.0	Unibright, SEELE	[25,30]

illustrates this evolution of blockchain from the first generation, i.e., cryptocurrency, until the fourth generation, i.e., application in industry 4.0. In Blockchain 1.0, the focus is on finances concerning cryptocurrency [26]. Blockchain 2.0 supports the creation of smart contracts for many agreement terms through autonomous computer programs and commands that execute automatically [24,27]. Blockchain 3.0 emphasizes decentralized applications (DApps). It uses decentralized storage and communication, hence its backend code runs on a decentralized P2P network. Blockchain 4.0 focuses on applications and the impact of blockchain implementation in different industries, specifically in industry 4.0 [28]. In other words, Blockchain 4.0 refers to making blockchain 3.0 usable and satisfying industry 4.0 requirements by making blockchain promises true in the industry. At present, various challenges still exist, leading to the evolution of blockchain 5.0. In this generation, blockchain would be applied together with artificial intelligence, hyper-converged infrastructure, and other advanced data analytics and industry 4.0 technologies for high security, reliability, scalability, etc. [29].

To date, several blockchain-based platforms have been developed by various developers and organizations. Depending on different applications, all these platforms can be divided into three types of networks, i.e., public, private, and consortium blockchains. Public blockchains are permissionless in nature and open and anyone in the world can access, perform transactions, and participate in the consensus process [31]. Private blockchains are permissioned blockchains controlled by a single central authority. The central authority permits the nodes that can join and perform transactions. Consortium blockchains are also permissioned blockchain but instead of a single authority, it is controlled by a group of preselected nodes [32]. Blockchain-based platforms are technically mature with sufficient community support to make sure future maintenance [33]. All the permitted members of the network control each other, therefore, no intermediary or central point of the account is needed [34]. Furthermore, the use of blockchain technology is especially beneficial in the case of high-value products with low trading volume [35]. In general, the traditional PLMs manage the data from conceptual design to product release. However, blockchain-based platforms have the capability to manage the below-mentioned data through the entire product lifecycle, i.e., design to final disposal or recycling.

Provenance Management: Provenance management refers to the tracking of information about people, processes, and methodologies involved in producing data from design to disposal or recycling. It simply shows where, when, how, and by whom the data are generated. The purpose of provenance management is to provide greater visibility and better efficiency by creating records in the network [36].

Bill of Materials (BOM) Management: BOM management refers to the capturing, configuring, and management of data, i.e., raw materials, sub-components, constituent parts, and quantities of each required for the end products. Effective BOM management is critical for the success of any manufacturer irrespective of the complexity of their products.

Manufacturing data and Process Management: Securely managing and validating the data generated through the entire manufacturing and post-manufacturing as well as material handling. It also accurately connects process planning to production to ensure the transfer of correct data to and from the production setup.

Collaborative Document Management: Collaborative document management refers to capturing, storing, and making the documents available throughout the product lifecycle in a controlled and secure network. It allows for linking the related documents to products, processes, or any other document for easy and fast reference and retrieval.

Suppliers and Materials data Management: It enables the procurement teams to track and communicate with their suppliers in various matters, i.e., quality of raw material, selection and purchasing of parts, revision of already placed orders, and so on.

Change Management: Due to changes in customer demand and availability of raw materials, the most frequent changes occur in design and production. Change management ensures that these changes are clearly defined, documented, and managed so that the other stakeholders can access the updated data.

Quality Management: Critical quality documents such as PCP, PFD, fishbone diagram, fault tree analysis, etc., can be managed across the organization. It helps in ensuring that the quality of the products is as per the organizational goals and regulatory needs.

Requirements Management: Requirements management involves collecting, analyzing, and managing what the different stockholders want from the product or service. It also includes requirements traceability, product configuration to requirements, collaboration with different stakeholders, and approval of requirements.

Operational and maintenance data management: Collecting and managing the real-time operational data, i.e., product operational conditions, running time, etc., and maintenance history. It helps in ensuring better decision-making within and even after the end of the product lifecycle, i.e., disposal or recycling.

Product Distribution data Management: Product distribution data management includes packaging, transportation, and delivering the product to the customer or facility where it is intended to be used. It helps in answering the question of when, how, and under what conditions the product is delivered.

In this paper, the analytic hierarchy process (AHP) has been applied for selecting the best possible blockchain-based platform for PLM. As blockchain is a new technology, a clear comparison of the available platforms based on performance-based metrics and key performance indicators is still not possible. This is the reason for using an AHP-based approach for the selection of a blockchain-based platform. AHP is one of the multi-criteria decisions making (MCDM) tools broadly accepted because of its provision for converting complex problems into a form of hierarchy [37]. It involves matrix algebra for measurements based on an expert’s judgment on pairwise comparison. In AHP, both qualitative and quantitative criteria can be compared using informed judgments to derive weights and priorities. AHP is commonly used in selection, prioritization, and forecasting where it is assumed that the decision-makers implicitly or explicitly know the objectives and the associated alternatives [38]. The research question that this paper tries to address are:

• What are the available blockchain-based platforms for PLM?
• How can AHP be applied to the selection of a blockchain-based platform?

The rest of the paper is organized as follows: The available and top blockchain-based platforms are explored in Section 2. Section 3 provides a detailed framework and uses an analytic hierarchy process-based selection of platforms. Results are presented and discussed in Section 4. Finally, this work is summarized and future directions are presented in Section 5.

2. Available Blockchain-Based Platforms

A systematic procedure has been followed to explore the available blockchain-based platforms for managing the entire product lifecycle. In the literature, the existing research publications consider a limited number of available platforms, therefore relying only on research publications would not be a wise approach. In this context, parallel to published research articles, different open search engines and research blogs published by well-known publishers, i.e., Capterra, LeewayHertz, G2, Tech Target, HFS, etc., could be considered as a more sophisticated method, and hence is adopted in this exploration process. Furthermore, as this study focuses on the selection of a platform for managing the entire lifecycle of the product, several important points that have been considered in this identification and exploration process can be pointed out as follows:

Maturity Status: With the rapid increase in demand, the development of blockchain-based platforms by different organizations is also increasing day by day. Some of the well-known platforms are now fully developed and matured while the code of other developed platforms is still being tested in different scenarios. In the initial exploration of the available platforms, we consider all the developed platforms irrespective of their maturity level. However, in the analytic process for selection, we only considered the fully matured platforms as alternatives.

Type of Blockchain: Although any type of blockchain-based platform can be used to manage the product lifecycle, as data privacy, transaction speed, and scalability are the prime concern in PLM, the focus in the exploration and selection processes is only on the permissioned blockchain.

Programming Languages: So far, many programming languages, i.e., C++, Python, Java Script, Solidity, etc., have been used in the development of blockchains. In addition to their native languages, these blockchain platforms also have the capability to support other languages. In this study, we considered all the blockchain platforms irrespective of their native and supported languages.

Consensus Protocols: The available platforms use different consensus protocols or algorithms, i.e., proof of work (PoW), proof of stake (PoS), practical byzantine fault tolerance (PBFT), and so on. This consensus protocol enables all the parties of the blockchain network to come to a common agreement on the present data state of the ledger. The business can define its own or can choose from the well-known available protocols. In one way or the other, these consensus protocols distinguish the platforms from one another. In the initial exploration, we considered platforms irrespective of their consensus protocols. However, the final analytics process considers the platforms with the most suitable protocols.

Smart Contract: Smart contracts are computer programs or transaction protocols stored on a blockchain that intend to execute automatically once a predefined condition is fulfilled. These smart contracts are responsible for the process of validation and enforcing the action on the blockchain. However, not all blockchain platforms need to support the concept of smart contracts. This study does not consider this factor and, hence, takes all the platforms irrespective of their case and whether they support the concept of smart contracts or not.

Scalability: Today’s product lifecycles are usually complex and decentralized in nature and hence, many stakeholders are involved in data generation and sharing. Therefore, blockchain platforms should be scalable enough to grow and manage an increase in the number of participants and transaction data. In the initial survey, we do not consider this factor; however, only highly scalable platforms are considered in the final selection through AHP.

A number of blockchain-based platforms introduced by different organizations have been explored as illustrated in

Table A1. List of available blockchain-based platforms.

ConsenSys Quorum	36. Bankchain	71. Straitis	106. Lexicon by Ternio
2. Hyperledger Sawtooth	37. Blockchain Evidence Locker	72. Tezos	107. LifeHash
3. 3. Kaleido	38. Bubichain	73. Universa	108. Lukka
4. Ethereum	39. Chain Core	74. Xinfin	109. MintMe.com Coin
5. Hyperledger Iroha	40. Digital Asset Platform	75. IOTA	110. NEM
6. IBM Blockchain Platform	41. Domus Tower Blockchain	76. Oracle Blockchain Platform	111. New Street Technologies
7. Corda	42. Factom Harmony	77. Achain	112. NuCypher
8. Hyperledger Fabric	43. GemOS	78. Eaelf Enterprise (Blockchain)	113. Ocyan
9. BlockCypher	44. Hydrachain	79. Aequalis	114. OneLedger
10. NEO	45. Hyperledger Indy	80. AERGO	115. Openchain
11. EOS	46. Monax	81. akaChain	116. Open Trade Blockchain (OTB)
12. Centrifuge	47. Nexledger	82. Axoni	117. Overledger Network Enterprise
13. Solana	48. Omni	83. Bethel	118. Paradigm Identity Blockchain
14. Factom	49. Onchain	84. Binance X	119. Platform 6
15. GetBlock	50. OpenCSD	85. BitFury	120. PONTON X/P Messenger
16. Blockedge	51. ParallelChain	86. BlockApps	121. ProximaX Sirius Chain
17. Multichain	52. pNetwork	87. Blockchain Gateway	122. Quantum Resistant Ledger
18. QANplatform	53. Polkadot	88. Blockstream	123. Quartz, The Smart Ledgers
19. Ripple	54. RSK	89. Bscexchange Finance	124. SASEUL Origin
20. Stellar Platform	55. SettleMint	90. Coinbase Institutional	125. SASEUL Origin Advanced Edition
21. Stratis Platform	56. Signchain Signature	91. CoreStarter	126. Sextant for DAML
22. Elements	57. StreamCore	92. DEIP	127. Snapper Future Tech
23. XDC Network	58. Symbiont Assembly	93. Development Env for TrueChain Dapp	128. Stark Drones
24. Xooa	59. VeChain ToolChain	94. DigiByte	129. Stezy Blockchain Platform
25. Algorand	60. Waves	95. Digital Twin Intelligent Automation Platform	130. SubQuery
26. BigChainDB	61. Zilliqa	96. GBC.AI	131. SubQuery
27. Blockdaemon	62. JSON-RPC Node API	97. Gospel Technology	132. Synthetix.io
28. Credits	63. Luniverse	98. Guardtime KSI	133. Tatum
29. Velas	64. Nxt Platform	99. Harmony Launcher	134. Tendermint
30. Blockstack	65. Monero	100. HCL CoTrust Blockchain Platform v1.4.8	135. Ticket Severity Forecasting
31. Chainalysis KYT	66. Chain Core	101. HCL CoTrust Blockchain Platform v2.1	136. UB Framework
32. Swirlds	67. Coco	102. Hedera Hashgraph	137. Vapor Pro
33. Tangle	68. Lisk	103. HyperCash	138. VeChain
34. AERGO Enterprise	69. Qtum	104. I/O Coin	139. VikRee
35. AxCore	70. Smilo	105. IoTeX Pantheon	140. Zeeve platform

of Appendix A. Most of the presented platforms are generic, i.e., they are not limited to just financial applications and, hence, can be used as an alternative to the traditional PLM systems to provide a framework for organizing and securing the data generated at different phases of the product lifecycle. However, based on certain metrics, different search engines, i.e., G2.com [39], Capterra.com [40], Gartner.com [41], Hacker noon [42], and Value coders [43], as well as research blogs published by LeewayHertz [44], Tech Target [45], HFS [46], etc., have considered some of them as top blockchain-based platforms as presented in

Table A2. Top blockchain-based platforms.

G2.com	HackerNoon	Capterra	LeewayHertz	HFS	TechTarget	Gartner	ValueCoders
Ethereum	Ethereum	Etherium	XDC Network	Etherium	Etherium	IBM Blockchain	Ethereum
Kaleido	Hyperledger Sawtooth	Solana	Stellar	Hyperledger Fabric	IBM Blockchain	Chainalysis KYT	Hyperledger Fabric
Azure Blockchain	Hyperledger Fabric	Azure Blockchain	Tezos	R3 Corda	Hyperledger Fabric	Ripple	OpenChain
IBM Blockchain	Hyperledger Iroha	BlockCypher	Hyperledger Fabric	Ripple	Hyperledger Saw	Ethereum	MultiChain
Corda	OpenChain	Blockedge	Hyperledger Sawtooth	Quorum	R3 Corda	Hyperledger Fabric	EOS
Hyperledger	Stellar	Centrifuge	Hedera Hashgraph		Tezos	Stellar	Ripple
	NEO	Quorum	Ripple		EOSIO	Microsoft Azure	Stellar
	EOS	Corda	Quorum		Stellar	Oracle Blockchain	R3 Corda
	Hedera Hashgraph	Hyperledger Fab	Hyperledger Iroha		ConsenSys Quorum	Bitcoin	Quorum
	R3 Corda	IBM Blockchain	Corda			Blockstream	Monero
	Quorum	Kaleido	EOS			Hyperledger Iroha	IBM Blockchain
	MultiChain	Luniverse	OpenChain			Swirlds	Hyperledger
	Ripple	Oracle Blockchain	Ethereum			Quorum	Neo Blockchain
	Credits	QANplatform	Dragonchain			Tangle	Hedera Hashgraph
	Elements	RippleNet	NEO			AERGO Enterprise	Hyperledger Iroha

of Appendix A. These top platforms have versatile capabilities and can be applied in any production industry to improve the management of the entire product lifecycle.

3. Selection of Blockchain-Based Platform for PLM

An Analytic Hierarchy Process (AHP) has been used for the selection of the most suitable blockchain-based platform for PLM. AHP is one of the most advanced methods among multi-criteria decision making (MCDM) available for decision making in a complex situation. Compared to other MCDM techniques, AHP is one of the most simple and easy-to-use techniques. This tool is highly scalable, and its hierarchical structure can easily be adjusted to fit. Furthermore, AHP is a data-intensive technique, therefore different-sized problems can be addressed through it. Furthermore, when decision makers need to make a decision in uncertain circumstances, then other MDCM techniques, i.e., basic fuzzy or Pythagorean fuzzy AHP, are the best option. Even in case of incomplete data or information, Grey AHP is better to be used. However, in this work, the data collected from various resources are deterministic and complete, therefore we used a simple AHP technique in our selection process. To ensure a more compact form and easy understanding, different notation as presented in

Table A3. Notation for alternatives and criteria.

Alternatives	Notation	Criteria	Notation
R3 Corda	RC	Throughput (Transaction per sec)	TH
Hyperledger Fabric	HF	Block Confirmation Time (Sec)	BT
Stellar	ST	User’s Feedback	UF
Ripple	RI	Platform Maturity	PM
Quorum	QU	Community Activity	CA
Hedera Hashgraph	HH	Popularity in Market	PT
OpenChain	OC	Published Research	PB
MultiChain	MC
NEO Platform	NE
NEO Platform	EO

of Appendix A has been used throughout this section. The AHP can be applied in three main steps, i.e., development of the hierarchical tree, pairwise comparison, and ranking of the alternatives for final selection. These steps are applied and discussed in detail in this section.

3.1. Development of Hierarchical Tree

In the AHP technique, the complex problem is converted into a hierarchical tree for decision making [47,48]. The hierarchical tree, as illustrated in Figure 1, can be developed in three steps, i.e., defining the goals or objectives, setting the decision criteria, and choosing among the available alternatives.

Define the Objective: The goal or objective of applying AHP in this work is to select the best blockchain-based platform among the available option for product lifecycle management.

Set the Decision Criteria: Blockchain-based platforms have several factors based on which they can be compared while using them for different applications. This work set the criteria based on the data available for the concerned blockchain-based available platform as given in

Table 2. Decision criteria for selection of platform.

	Platform	RC	HF	ST	RI	QU	HH	OC	MC	NE	EO	Source
Criteria
Interoperability		Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Platform Website
Data Transparency		Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Platform Website
Scalability		High	High	High	High	High	High	High	High	High	High	Platform Website
Level of Decentralization		PT *	PT *	PT *	PT *	PT *	PT *	PT *	PT *	PT *	PT *	[45]
Data Privacy		Strong	Strong	Strong	Strong	Strong	Strong	Strong	Strong	Strong	Strong	Platform Website
Ubiquitous Access		Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Platform Website
Collaborative Data Provision		Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Nature of Blockchain
Data Security		Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Platform Website
Throughput (TPS)		1678	3500	83.26	1500	2300	10,000	100	2000	10,000	4000	Platform Website
Block Confirmation Time (Sec)		1	1	4	3.5	1	3 to 5	1	1	15	1	Platform Website
User Feedback	Customization, Ease of deployment, administration	4.3	4.3	4.5	4.2	4	4	4	4.5	4.2	4.4	[23,41,49]
Platform Maturity	Founded	2015	2015	2014	2012	2016	2017	2015	2014	2017	2017	Platform Website
	Employees	250–500	78 (25–100)	100–250	500–1000	25–100	100–250	1–25	1–25	45	106	[50]
Community Activity	Commits	9725	14,113	7970	12,687	14,303	1105	1268	1454	1385	20,554	[51]
	Contributors	190	316	83	96	485	16	24	4	61	227	[51]
	Industry Adoption	Very High	Extremely High	High	High	Very High	High	High	High	High	High	Websites, Blogs
Popularity in Market	Twitter Followers	51,600	78,100	753,200	2,500,000	7472	243,300	696	130,600	431,900	261,100	[52]
	LinkedIn Followers	37,444	39,405	34,255	275,963	84,512	27,093	756	244	1284	91	[53]
Published Research	Science Direct	190	738	234	661	199	21	22	141	182	18	[54]
	Web of Science	28	800	18	150	39	2	3	45	32	31	[55]

* Partially Decentralized and can be incremental, which allows for multiple stakeholders to participate.

. However, these top platforms perform almost equally well in terms of certain settled decision criteria, i.e., interoperability, data transparency, scalability, ubiquitous access, level of decentralization, collaborative data provision, and data security. Furthermore, as blockchain is a new technology, a clear comparison of the available platforms based on performance-based metrics and key performance indicators is not completely mature and is still in the development stage. Therefore, this work considered the factors shown in Figure 1 as the final decision criteria. As AHP is based on the findings from cognitive science, it is usually suggested that five to nine criteria are the ideal range [38]. Therefore, certain relevant criteria are integrated together as illustrated in Table 2.

Choose the alternatives: As presented in Table A1 of Appendix A and discussed in Section 2, many blockchain-based platforms are available in the market. These available platforms consist of dedicated platforms (with their own protocols) as well as certain other service-providing platforms. The service-providing platforms just provide services as third parties and hence cannot be counted as dedicated platforms. Therefore, these service providers’ platforms have not been considered while choosing our final alternatives. Furthermore, different factors, i.e., transaction speed, data privacy, scalability, etc., are of prime concern in PLM. Therefore, we only considered the permissioned (private) blockchain platforms in our final selection. Finally, based on the defined criteria, the most suitable permissioned platforms are finalized as given in Table 2 and illustrated as available alternatives in decision-making as presented in Figure 1.

3.2. Pairwise Comparison

As the analytic hierarchy process involves matrix algebra for measurements based on an expert’s judgment, pairwise comparison of criteria, as well as corresponding alternatives, is one of the most crucial parts of this process. Both qualitative and quantitative criteria can be considered and compared to derive weights and priorities. To give a relative weightage to each criterion and alternative, we define the scale as presented in

Table 3. Scale for pairwise comparison of criterion/alternatives.

Relative Weightage	Importance	Explanation
1	Low	Criteria/Alternative has low importance compared to other criteria/alternative
3	Moderate	Criteria/Alternative has moderate importance compared to other criteria/alternative
5	High	Criteria/Alternative has high importance compared to other criteria/alternative
7	Very High	Criteria/Alternative has very high importance compared to other criteria/alternative
9	Extremely High	Criteria/Alternative has extremely high importance compared to other criteria/alternative

.

Based on the available data in Table 2 and a thorough discussion with experts from different research groups and industries, we assigned importance to each criterion as well as to all alternatives in terms of individual criteria as illustrated in

Table 4. Relative importance of criterion and alternatives.

	Platform	RC	HF	ST	RI	QU	HH	OC	MC	NE	EO
Criteria
TH	Extremely High	Moderate	Very High	Low	Moderate	High	Extremely High	Low	High	Extremely High	Very High
BT	Extremely High	High	High	Moderate	Moderate	High	Moderate	High	High	Low	High
UF	Very High	Very High	Very High	Extremely High	High	Moderate	Moderate	Moderate	Extremely High	High	Very High
PM	Moderate	Very High	High	High	Very High	High	High	Low	Low	Moderate	Moderate
CA	Very High	Very High	Extremely High	High	High	Extremely High	Low	Low	Low	High	Very High
PT	Low	High	Very High	Very High	Extremely High	Moderate	High	Moderate	High	High	Moderate
PR	High	Moderate	Extremely High	High	Very High	High	Low	Low	Moderate	High	Moderate

. A complete pairwise comparison of criteria and alternatives for the available data is given in detail in

Table A4. Pairwise comparison of criteria.

	TH	BT	UF	PM	CA	PT	PB
TH	9/9	9/9	9/7	9/3	9/7	9/1	9/5
BT	9/9	9/9	9/7	9/3	9/7	9/1	9/5
UF	7/9	7/9	7/7	7/3	7/7	7/1	7/5
PM	3/9	3/9	3/7	3/3	3/7	3/1	3/5
CA	7/9	7/9	7/7	7/3	7/7	7/1	7/5
PT	1/9	1/9	1/7	1/3	1/7	1/1	1/5
PB	5/9	5/9	5/7	5/3	5/7	5/1	5/5

,

Table A5. Pairwise comparison in terms of throughput.

	RC	HF	ST	RI	QU	HH	OC	MC	NE	EO
RC	3/3	3/7	3/1	3/3	3/5	3/9	3/1	3/5	3/9	3/7
HF	7/3	7/7	7/1	7/3	7/5	7/9	7/1	7/5	7/9	7/7
ST	1/3	1/7	1/1	1/3	1/5	1/9	1/1	1/5	1/9	1/7
RI	3/3	3/7	3/1	3/3	3/5	3/9	3/1	3/5	3/9	3/7
QU	5/3	5/7	5/1	5/3	5/5	5/9	5/1	5/5	5/9	5/7
HH	9/3	9/7	9/1	9/3	9/5	9/9	9/1	9/5	9/9	9/7
OC	1/3	1/7	1/1	1/3	1/5	1/9	1/1	1/5	1/9	1/7
MC	5/3	5/7	5/1	5/3	5/5	5/9	5/1	5/5	5/9	5/7
NE	9/3	9/7	9/1	9/3	9/5	9/9	9/1	9/5	9/9	9/7
EO	7/3	7/7	7/1	7/3	7/5	7/9	7/1	7/5	7/9	7/7

,

Table A6. Pairwise comparison in terms of block confirmation time.

	RC	HF	ST	RI	QU	HH	OC	MC	NE	EO
RC	5/5	5/5	5/3	5/3	5/5	5/3	5/5	5/5	5/1	5/5
HF	5/5	5/5	5/3	5/3	5/5	5/3	5/5	5/5	5/1	5/5
ST	3/5	3/5	3/3	3/3	3/5	3/3	3/5	3/5	3/1	3/5
RI	3/5	3/5	3/3	3/3	3/5	3/3	3/5	3/5	3/1	3/5
QU	5/5	5/5	5/3	5/3	5/5	5/3	5/5	5/5	5/1	5/5
HH	3/5	3/5	3/3	3/3	3/5	3/3	3/5	3/5	3/1	3/5
OC	5/5	5/5	5/3	5/3	5/5	5/3	5/5	5/5	5/1	5/5
MC	5/5	5/5	5/3	5/3	5/5	5/3	5/5	5/5	5/1	5/5
NE	1/5	1/5	1/3	1/3	1/5	1/3	1/5	1/5	1/1	1/5
EO	5/5	5/5	5/3	5/3	5/5	5/3	5/5	5/5	5/1	5/5

,

Table A7. Pairwise comparison in terms of user feedback.

	RC	HF	ST	RI	QU	HH	OC	MC	NE	EO
RC	7/7	7/7	7/9	7/5	7/3	7/3	7/3	7/9	7/5	7/7
HF	7/7	7/7	7/9	7/5	7/3	7/3	7/3	7/9	7/5	7/7
ST	9/7	9/7	9/9	9/5	9/3	9/3	9/3	9/9	9/5	9/7
RI	5/7	5/7	5/9	5/5	5/3	5/3	5/3	5/9	5/5	5/7
QU	3/7	3/7	3/9	3/5	3/3	3/3	3/3	3/9	3/5	3/7
HH	3/7	3/7	3/9	3/5	3/3	3/3	3/3	3/9	3/5	3/7
OC	3/7	3/7	3/9	3/5	3/3	3/3	3/3	3/9	3/5	3/7
MC	9/7	9/7	9/9	9/5	9/3	9/3	9/3	9/9	9/5	9/7
NE	5/7	5/7	5/9	5/5	5/3	5/3	5/3	5/9	5/5	5/7
EO	7/7	7/7	7/9	7/5	7/3	7/3	7/3	7/9	7/5	7/7

,

Table A8. Pairwise comparison in terms of maturity.

	RC	HF	ST	RI	QU	HH	OC	MC	NE	EO
RC	7/7	7/5	7/5	7/7	7/5	7/5	7/1	7/1	7/3	7/3
HF	5/7	5/5	5/5	5/7	5/5	5/5	5/1	5/1	5/3	5/3
ST	5/7	5/5	5/5	5/7	5/5	5/5	5/1	5/1	5/3	5/3
RI	7/7	7/5	7/5	7/7	7/5	7/5	7/1	7/1	7/3	7/3
QU	5/7	5/5	5/5	5/7	5/5	5/5	5/1	5/1	5/3	5/3
HH	5/7	5/5	5/5	5/7	5/5	5/5	5/1	5/1	5/3	5/3
OC	1/7	1/5	1/5	1/7	1/5	1/5	1/1	1/1	1/3	1/3
MC	1/7	1/5	1/5	1/7	1/5	1/5	1/1	1/1	1/3	1/3
NE	3/7	3/5	3/5	3/7	3/5	3/5	3/1	3/1	3/3	3/3
EO	3/7	3/5	3/5	3/7	3/5	3/5	3/1	3/1	3/3	3/3

,

Table A9. Pairwise comparison in terms of community activity.

	RC	HF	ST	RI	QU	HH	OC	MC	NE	EO
RC	7/7	7/9	7/5	7/5	7/9	7/1	7/1	7/1	7/5	7/7
HF	9/7	9/9	9/5	9/5	9/9	9/1	9/1	9/1	9/5	9/7
ST	5/7	5/9	5/5	5/5	5/9	5/1	5/1	5/1	5/5	5/7
RI	5/7	5/9	5/5	5/5	5/9	5/1	5/1	5/1	5/5	5/7
QU	9/7	9/9	9/5	9/5	9/9	9/1	9/1	9/1	9/5	9/7
HH	1/7	1/9	1/5	1/5	1/9	1/1	1/1	1/1	1/5	1/7
OC	1/7	1/9	1/5	1/5	1/9	1/1	1/1	1/1	1/5	1/7
MC	1/7	1/9	1/5	1/5	1/9	1/1	1/1	1/1	1/5	1/7
NE	5/7	5/9	5/5	5/5	5/9	5/1	5/1	5/1	5/5	5/7
EO	7/7	7/9	7/5	7/5	7/9	7/1	7/1	7/1	7/5	7/7

,

Table A10. Pairwise comparison in terms of popularity in market.

	RC	HF	ST	RI	QU	HH	OC	MC	NE	EO
RC	5/5	5/7	5/7	5/9	5/3	5/5	5/3	5/5	5/5	5/3
HF	7/5	7/7	7/7	7/9	7/3	7/5	7/3	7/5	7/5	7/3
ST	7/5	7/7	7/7	7/9	7/3	7/5	7/3	7/5	7/5	7/3
RI	9/5	9/7	9/7	9/9	9/3	9/5	9/3	9/5	9/5	9/3
QU	3/5	3/7	3/7	3/9	3/3	3/5	3/3	3/5	3/5	3/3
HH	5/5	5/7	5/7	5/9	5/3	5/5	5/3	5/5	5/5	5/3
OC	3/5	3/7	3/7	3/9	3/3	3/5	3/3	3/5	3/5	3/3
MC	5/5	5/7	5/7	5/9	5/3	5/5	5/3	5/5	5/5	5/3
NE	5/5	5/7	5/7	5/9	5/3	5/5	5/3	5/5	5/5	5/3
EO	3/5	3/7	3/7	3/9	3/3	3/5	3/3	3/5	3/5	3/3

and

Table A11. Pairwise comparison in terms of published research.

	RC	HF	ST	RI	QU	HH	OC	MC	NE	EO
RC	3/3	3/9	3/5	3/7	3/5	3/1	3/1	3/3	3/5	3/3
HF	9/3	9/9	9/5	9/7	9/5	9/1	9/1	9/3	9/5	9/3
ST	5/3	5/9	5/5	5/7	5/5	5/1	5/1	5/3	5/5	5/3
RI	7/3	7/9	7/5	7/7	7/5	7/1	7/1	7/3	7/5	7/3
QU	5/3	5/9	5/5	5/7	5/5	5/1	5/1	5/3	5/5	5/3
HH	1/3	1/9	1/5	1/7	1/5	1/1	1/1	1/3	1/5	1/3
OC	1/3	1/9	1/5	1/7	1/5	1/1	1/1	1/3	1/5	1/3
MC	3/3	3/9	3/5	3/7	3/5	3/1	3/1	3/3	3/5	3/3
NE	5/3	5/9	5/5	5/7	5/5	5/1	5/1	5/3	5/5	5/3
EO	3/3	3/9	3/5	3/7	3/5	3/1	3/1	3/3	3/5	3/3

of Appendix B.

3.3. Ranking the Alternatives

Different approaches can be adopted to obtain a ranking of priorities from a given pairwise matrix. However, a study demonstrated mathematically that the eigenvector-based solution is the best approach [56]. In this work, we solved the pairwise matrix for the eigenvector by squaring the matrix each time, summing, and then normalizing. Microsoft Excel software was used for calculation and was instructed to stop the iterations when the difference between sums in two consecutive calculations remained zero. These steps of finding the eigenvector are repeated for the pairwise comparison matrices of criteria as well as alternatives as follows:

Step 1: Calculate the eigenvector for the pairwise comparison matrix of criteria.

$$\mathit{Criteria} = \begin{array}{cc}& \begin{array}{c}\begin{array}{ccccccc} \mathrm{TH}\hspace{1em}& \mathrm{BT}\hspace{1em}& \mathrm{UF}\hspace{1em}& \mathrm{PM}\hspace{1em}& \mathrm{CA}\hspace{1em}& \mathrm{PT}\hspace{1em}& \mathrm{PB}\hspace{1em}\end{array}\end{array}\\ \begin{array}{c}\mathrm{TH}\\ \mathrm{BT}\\ \mathrm{UF}\\ \mathrm{PM}\\ \mathrm{CA}\\ \mathrm{PT}\\ \mathrm{PB}\end{array}& \left[\begin{array}{c}\begin{array}{ccccccc}1.0000& 1.0000& 1.2857& 3.0000& 1.2857& 9.0000& 1.8000\\ 1.0000& 1.0000& 1.2857& 3.0000& 1.2857& 9.0000& 1.8000\\ 0.7778& 0.7778& 1.0000& 2.3333& 1.0000& 7.0000& 1.4000\\ 0.3333& 0.3333& 0.4286& 1.0000& 0.4286& 3.0000& 0.6000\\ 0.7778& 0.7778& 1.0000& 2.3333& 1.0000& 7.0000& 1.4000\\ 0.1111& 0.1111& 0.1429& 0.3333& 0.1429& 1.0000& 0.2000\\ 0.5556& 0.5556& 0.7143& 1.6667& 0.7143& 5.0000& 1.0000\end{array}\end{array}\right]\end{array}$$

$$\mathit{Criteria} \ast \mathit{Criteria} = \begin{array}{cc}& \begin{array}{c}\begin{array}{ccccccc} \mathrm{TH} & \mathrm{BT} & \mathrm{UF} & \mathrm{PM} & \mathrm{CA} & \mathrm{PT} & \mathrm{PB} \end{array}\end{array}\\ \begin{array}{c}\mathrm{TH}\\ \mathrm{BT}\\ \mathrm{UF}\\ \mathrm{PM}\\ \mathrm{CA}\\ \mathrm{PT}\\ \mathrm{PB}\end{array}& \left[\begin{array}{c}\begin{array}{ccccccc}7.0000& 7.0000& 9.0000& 21.0000& 9.0000& 63.0000& 12.6000\\ 7.0000& 7.0000& 9.0000& 21.0000& 9.0000& 63.0000& 12.6000\\ 5.4444& 5.4444& 7.0000& 16.3333& 7.0000& 49.0000& 9.8000\\ 2.3333& 2.3333& 3.0000& 7.0000& 3.0000& 21.0000& 4.2000\\ 5.4444& 5.4444& 7.0000& 16.3333& 7.0000& 49.0000& 9.8000\\ 0.7778& 0.7778& 1.0000& 2.3333& 1.0000& 7.0000& 1.4000\\ 3.8889& 3.8889& 5.0000& 11.6667& 5.0000& 35.0000& 7.0000\end{array}\end{array}\right]\end{array}\begin{array}{c}\mathit{Eigenvector}\\ \left[\begin{array}{c}\begin{array}{c}0.2195\\ 0.2195\\ 0.1707\\ 0.0732\\ 0.1707\\ 0.0244\\ 0.1220\end{array}\end{array}\right]\end{array}$$

Step 2: Calculate the eigenvector for all the pairwise comparison matrices of alternatives in terms of each criterion.

$$\mathit{TH} \ast \mathit{TH} = \begin{array}{cc}& \begin{array}{c}\begin{array}{cccccccccc} \mathrm{RC} & \mathrm{HF} & \mathrm{ST} & \mathrm{RI} & \mathrm{QU} & \mathrm{HH} & \mathrm{OC} & \mathrm{MC} & \mathrm{NE} & \mathrm{EO} \end{array}\end{array}\\ \begin{array}{c}\mathrm{RC}\\ \mathrm{HF}\\ \mathrm{ST}\\ \mathrm{RI}\\ \mathrm{QU}\\ \mathrm{HH}\\ \mathrm{OC}\\ \mathrm{MC}\\ \mathrm{NE}\\ \mathrm{EO}\end{array}& \left[\begin{array}{c}\begin{array}{cccccccccc}10.0000& 4.2857& 30.0000& 10.0000& 6.0000& 3.3333& 30.0000& 6.0000& 3.3333& 4.2857\\ 23.3333& 10.0000& 70.0000& 23.3333& 14.0000& 7.7778& 70.0000& 14.0000& 7.7778& 10.0000\\ 3.3333& 1.4286& 10.0000& 3.3333& 2.0000& 1.1111& 10.0000& 2.0000& 1.1111& 1.4286\\ 10.0000& 4.2857& 30.0000& 10.0000& 6.0000& 3.3333& 30.0000& 6.0000& 3.3333& 4.2857\\ 16.6667& 7.1429& 50.0000& 16.6667& 10.0000& 5.5556& 50.0000& 10.0000& 5.5556& 7.1429\\ 30.0000& 12.8571& 90.0000& 30.0000& 18.0000& 10.0000& 90.0000& 18.0000& 10.0000& 12.8571\\ 3.3333& 1.4286& 10.0000& 3.3333& 2.0000& 1.1111& 10.0000& 2.0000& 1.1111& 1.4286\\ 16.6667& 7.1429& 50.0000& 16.6667& 10.0000& 5.5556& 50.0000& 10.0000& 5.5556& 7.1429\\ 30.0000& 12.8571& 90.0000& 30.0000& 18.0000& 10.0000& 90.0000& 18.0000& 10.0000& 12.8571\\ 23.3333& 10.0000& 70.0000& 23.3333& 14.0000& 7.7778& 70.0000& 14.0000& 7.7778& 10.0000\end{array}\end{array}\right]\end{array}\begin{array}{c}\mathit{Eigenvector}\\ \left[\begin{array}{c}\begin{array}{c}0.0600\\ 0.1400\\ 0.0200\\ 0.0600\\ 0.1000\\ 0.1800\\ 0.0200\\ 0.1000\\ 0.1800\\ 0.1400\end{array}\end{array}\right]\end{array}$$

The above matrix shows the eigenvector calculation for alternatives in terms of throughput. The eigenvectors’ calculations for alternatives in terms of other criteria can be seen in Appendix C.

Step 3: Calculate the final weightage by multiplying the eigenvectors of alternatives with the criteria-based eigenvector.

$$\mathit{Criteria} \ast \mathit{Criteria} = \begin{array}{cc}& \begin{array}{c}\begin{array}{ccccccc} \mathrm{TH} & \mathrm{BT} & \mathrm{UF} & \mathrm{PM} & \mathrm{CA} & \mathrm{PT} & \mathrm{PB} \end{array}\end{array}\\ \begin{array}{c}\mathrm{RC}\\ \mathrm{HF}\\ \mathrm{ST}\\ \mathrm{RI}\\ \mathrm{QU}\\ \mathrm{HH}\\ \mathrm{OC}\\ \mathrm{MC}\\ \mathrm{NE}\\ \mathrm{EO}\end{array}& \left[\begin{array}{c}\begin{array}{ccccccc}0.0600& 0.1250& 0.1207& 0.1667& 0.1400& 0.0962& 0.0714\\ 0.1400& 0.1250& 0.1207& 0.1190& 0.1800& 0.1346& 0.2143\\ 0.0200& 0.0750& 0.1552& 0.1190& 0.1000& 0.1346& 0.1190\\ 0.0600& 0.0750& 0.0862& 0.1667& 0.1000& 0.1731& 0.1667\\ 0.1000& 0.1250& 0.0517& 0.1190& 0.1800& 0.0577& 0.1190\\ 0.1800& 0.0750& 0.0517& 0.1190& 0.0200& 0.0962& 0.0238\\ 0.0200& 0.1250& 0.0517& 0.0238& 0.0200& 0.0577& 0.0238\\ 0.1000& 0.1250& 0.1552& 0.0238& 0.0200& 0.0962& 0.0714\\ 0.1800& 0.0250& 0.0862& 0.0714& 0.1000& 0.0962& 0.1190\\ 0.1400& 0.1250& 0.1207& 0.0714& 0.1400& 0.0577& 0.0714\end{array}\end{array}\right]\end{array}\begin{array}{c}\mathit{Criteria\; Eigenvector}\\ \left[\begin{array}{c}\begin{array}{c}0.2195\\ \\ 0.2195\\ \\ 0.1707\\ \\ 0.0732\\ \\ 0.1707\\ \\ 0.0244\\ \\ 0.1220\end{array}\end{array}\right]\end{array}=\begin{array}{c}\mathit{Final\; Weightage}\\ \left[\begin{array}{c}\begin{array}{c}0.1084\\ 0.1476\\ 0.0909\\ 0.0982\\ 0.1136\\ 0.0822\\ 0.0501\\ 0.0921\\ 0.0989\\ 0.1180\end{array}\end{array}\right]\end{array}$$

Step 4: Ranking the alternatives based on final weightage.

$$\mathit{Ranking} = \begin{array}{c}\mathit{Final\; Weightage}\\ \begin{array}{c}\text{RC}\\ \text{HF}\\ \text{ST}\\ \text{RI}\\ \text{QU}\\ \text{HH}\\ \text{OC}\\ \text{MC}\\ \text{NE}\\ \text{EO}\end{array}\left[\begin{array}{c}\begin{array}{c}0.1084\\ 0.1476\\ 0.0909\\ 0.0982\\ 0.1136\\ 0.0822\\ 0.0501\\ 0.0921\\ 0.0989\\ 0.1180\end{array}\end{array}\right]\end{array}=\begin{array}{c}\mathit{Ranking}\\ \left[\begin{array}{c}\begin{array}{c}4\\ 1\\ 8\\ 6\\ 3\\ 9\\ 10\\ 7\\ 5\\ 2\end{array}\end{array}\right]\end{array}$$

4. Results and Discussion

Although any permissioned blockchain platform can be used for data creation, storage, and retrieval throughout the lifecycle of a product, this study was carried out with the intention to select the platform that would not only manage the entire lifecycle of a product but also fulfill the requirements of industry 4.0. In this context, this study mainly consists of two parts, i.e., the identification of available blockchain platforms and the selection of the best platforms with the help of the AHP technique. The identified and top platforms are provided in Table A1 and Table A2, respectively, while the final results obtained via the AHP technique are summarized in

Table 5. Final results based on AHP.

Platform	Criteria							% Weightage	Ranking
	TH (% Age) 21.95	BT (% Age) 21.95	UF (% Age) 17.07	PM (% Age) 7.32	CA (% Age) 17.07	PT (% Age) 2.44	PB (% Age) 12.20
RC	6.00	12.50	12.07	16.67	14.00	9.62	7.14	10.84	4
HF	14.00	12.50	12.07	11.90	18.00	13.46	21.43	14.76	1
ST	2.00	7.50	15.52	11.90	10.00	13.46	11.90	9.09	8
RI	6.00	7.50	8.62	16.67	10.00	17.31	16.67	9.82	6
QU	10.00	12.50	5.17	11.90	18.00	5.77	11.90	11.36	3
HH	18.00	7.50	5.17	11.90	2.00	9.62	2.38	8.22	9
OC	2.00	12.50	5.17	2.38	2.00	5.77	2.38	5.01	10
MC	10.00	12.50	15.52	2.38	2.00	9.62	7.14	9.21	7
NE	18.00	2.50	8.62	7.14	10.00	9.62	11.90	9.89	5
EO	14.00	12.50	12.07	7.14	14.00	5.77	7.14	11.80	2

. The percentage data for criteria in the first row of Table 5 are obtained from the eigenvector of the criteria, while the corresponding column data are based on the eigenvector of alternatives with respect to each criterion. The last two columns of Table 5 represent the final weightage and ranking of alternatives, respectively.

The findings in Table 5 reveal that Hyperledger Fabric (relative weightage = 14.84%) is ranked first and is thus the best possible option followed by EOS.IO (relative weightage = 11.80%) and Quorum (relative weightage = 11.36%). Hence, these blockchain platforms should be the top priorities of the industries for the said purpose. On the contrary, OpenChain (relative weightage = 5.01%) is considered the worst option for industries to select among the alternatives. Furthermore, the overall low numeric values obtained for relative weightages are just because of the high number of alternatives considered in AHP. Therefore, consequently, it can be stated that the lower the number of alternatives considered in AHP, the higher the relative weightage each alternative would obtain or vice versa. These results are purely based on the defined importance of each criterion and the comparative weightage of alternatives in terms of criterion. Furthermore, for the purpose of high transaction speed, data privacy, and scalability, this work only considered the permissioned blockchains among the dedicated platforms.

In AHP, both qualitative and quantitative criteria could be considered and compared to derive weights and priorities. However, blockchain platforms have not yet been qualitatively compared on an exact scale in terms of different factors, i.e., interoperability (the ability of a platform to communicate and exchange the data with another platform), level of decentralization (transfer of control to a distributed network or nodes), etc. However, it is evident from the studies that the presented top blockchain-based platforms perform equally well when compared for those qualitative factors, i.e., interoperability, data transparency, scalability, ubiquitous access, level of decentralization, collaborative data provision, and data security. Hence, in addition to the quantitative criteria, this study considers only the built-in capabilities of the platforms as decision criteria, i.e., throughput (number of transactions processed per second), block confirmation time (the amount of time for the creation of a new block), etc.

5. Conclusions and Future Directions

This study identified and selected the most suitable blockchain-based platform that can be used for product lifecycle management. Initially, a total of 140 blockchain platforms were identified through various search engines. However, taking certain specific requirements into consideration, various platforms, i.e., those permissionless in nature or third-party service providers, were dropped from further analysis. It was further observed that among the available permissioned blockchain platforms, some are still not fully matured, and their consensus mechanism is still in the testing and validation stage. Therefore, top software reviews and selection websites were used to identify and consider the fully matured, highly scalable, and secured blockchain platforms. Out of those platforms, 10 were finalized as the best available alternatives for consideration in the final selection process.

A number of criteria were identified through different sources as given in Table 2. Among these defined criteria, seven decision criteria were finally settled on for the selection of the best blockchain platforms. Accordingly, based on the available data on different search engines and thorough discussion with experts from various research groups and industries, each criterion as well as alternatives were assigned relative importance as illustrated in Table 4. An analytical hierarchy process-based multi-criteria decision support system was then applied to rank and select the top 10 available alternatives.

To conclude our findings, we could say that Hyperledger Fabric has the highest relative score of 14.76% and is the best possible option that can be selected for the purpose of PLM. EOS.IO and Quorum are the second and third best options with relative scores of 11.80% and 11.36%, respectively. All the results obtained in this work are solely based on the defined importance of criteria and the comparative weightage of alternatives in terms of each criterion. However, the developed approach is generic and can be applied for any similar selection purpose with a range of criteria, alternatives, and their relative importance. To our knowledge, this work is the first and could be considered a pioneer in the selection of blockchain platforms for the purpose of product lifecycle management.

In continuation of the obtained results, immediate future work could be the development and implementation of a Hyperledger Fabric-based blockchain platform for product lifecycle management in a real production environment. This intended work could first identify and describe the basic components of the Hyperledger fabric blockchain for a complete product lifecycle management system, i.e., peers, channels, chain codes, ordering services, and so on. Then, various policies would be developed, i.e., what peers would have the authority to endorse the transaction, what peers just commit the transaction, and what the consensus mechanism would be. Although Hyperledger fabric has a built-in crash fault tolerance consensus, due to its modular architecture, one can just plug in and out their own consensus mechanism. Furthermore, this work could also emphasize how the data generated through different sources in a real production setup could be extracted, secured, and transferred to the Hyperledger fabric blockchain. This work could keenly focus on hash storing and transactions among the various stakeholders involved throughout the product lifecycle and a common cloud database for storing a huge amount of design, manufacturing, and other data.