Developing a causal framework of internet of things adoption barriers for agile manufacturing in post COVID-19

Abstract

In order to remain competitive in a post-COVID-19 scenario, manufacturing organizations should use the pandemic scenario as a lever to make IoT adoption more agile and faster by reinventing their business models. Although, IoT has vast potential in multiple sensors’ direct integration without human intervention, its adoption has observed significant barriers during COVID-19. In this context, the current research aims to identify, prioritize, categorize, and investigate inter-dependencies among the significant barriers to adopting IoT technologies in the manufacturing industry. The analysis used two-stage integrated AHP-TOPSIS (Analytical Hierarchy Process-Technique for Order Preference by Similarity to Ideal Solution) and ISM (Interpretative Structural Modeling) techniques. For AHP and ISM techniques, a group of 8 experts from academia and industry with at least 10 years of experience was surveyed to capture their perceptions. Whereas, for TOPSIS, data were collected from 433 operational-level users. The research discloses that the significant driver barriers that hamper the adoption of IoT technologies are: poor internet infrastructure, high adoption & operating costs and lack of government regulations. This study will assist practitioners, decision-makers and organizations in understanding the significant barriers to seamless IoT implementation.

Keywords

internet of things barriers manufacturing sector analytical hierarchical process technaique for order of preference by similarity to ideal solution interpretive structural equation modeling

Introduction

The world has witnessed unforeseen disruptions in industrialized operations due to coronavirus outbursts. The COVID-19 pandemic has hit worldwide trade at a matchless rapidity due to lockdowns.¹ The situation has forced firms, particularly those depending on manual processes, to transform their operations into robust, agile, smarter ways² and achieve resilience to recover quickly from misfortune and change, allowing them to continue operations. It is important to note that the ability of an enterprise to change from failure to success is considered resilience.³ Whereas, robustness measures the insensitivity to disturbances of the system and focuses on the ability of a system to withstand variations whereas manufacturing focuses on the systematic creation of products from raw material.⁴ Furthermore, Seric and Winkler,⁵ Javaid, Haleem⁶ stressed that digitalization could be significant in managing hazards and restoring operations in the post-COVID 19 era. In fact, COVID-19 has accelerated the inevitable digital transformation through the Internet of Things (IoT).⁷ The aim of IoT is to use predictive analytics and big data technologies to increase efficiency, profitability and productivity, and it is broadly spread in an entrenched grid of autonomous and intelligent devices.⁸ It is a digital virtual transformation from the physical business ecosystem into a smart and advanced entity. Conceptually, IoT aims to connect any “Thing” through error-free networks without considering the time, place and motion.⁹ Things mean sensor devices that result in efficient collaboration among stakeholders and more agile manufacturing operations.¹⁰ Information is exchanged in a secure and robust way through IoT in both kinds of businesses (manufacturing and services).¹¹ Robustness measures the insensitivity of the system to disturbances, and focuses on the ability of a system to withstand variations, whereas manufacturing focuses on the systematic creation of products from raw materials.³

IoT adoption offers tangible commercial benefits such as low cost and risk, and enhanced operational processes, as well as additional benefits of virtualization, flexibility, adaption, transparency and visibility in operations.^12,13 In addition, the proactive resilience of different network systems, including holistic supply chain networks, is enhanced through IoT in different ways: real-time monitoring of goods in transit, prediction of potential disruptions, speedy identification of problems aiding quicker responses to troubles and provision of insight into each node’s performance.¹⁴ For achieving sustainable success in today’s manufacturing system, the development of resilience is imperative because, in the face of various disruptions, resilient manufacturing systems can withstand and navigate uncertainties, confirming continued operations while maintaining their core functionality. Also, to accommodate product demand changes and supply chain disruptions, the processes of resilient manufacturing systems can be easily adapted.³ The trend of this technology is vast, covering all business sectors and major industries, and no industry is left out from the influence of the overall emergence of IoT technology.¹⁵

Organizations must consider IT and IoT as fundamental enablers for agile and next-generation advanced manufacturing.^16,17 Thus, interpreting barriers and facilitators to agile manufacturing is essential for manufacturing organizations to enhance organizational agility.¹⁸ However, the literature lacks studies on IoT adoption in agile manufacturing; several obstacles exist while implementing IoT in agile manufacturing. Therefore, eradication of such barriers is essential for the operative employment of IoT in the manufacturing industry, especially in the COVID-19 scenario.

In this context, the current paper is one of the earliest studies that attempt to prioritize the IoT adoption barriers and catch the communal affiliation among the several IoT application barriers from the perspective of agile manufacturing in the post-COVID-19 period. The prioritization of the prevailing obstacles to IoT execution was carried out using AHP-TOPSIS, while an investigation of the causative association amid barriers was performed using the Interpretive Structural Modeling (ISM) technique. The remaining paper has been sequenced as follows: section 2 is a literature review on IoT and agile manufacturing, and offers a deep insight into the barriers to IoT adoption; section 3 discusses the research methodology; section 4 deliberates the meticulous execution of the AHP-TOPSIS and ISM techniques, and presents the findings; section 5 discusses the results and the implications, while conclusions and study limitations are presented in section 6.

Literature review

Internet of things

The internet has advanced into a real-world solicitation in a broader platform of IoT using routine and daily objects. At various stages, there are virtual connections among the physical items like products, equipment and machines, and such devices are controlled and monitored through cyber systems.¹⁹ Based on protocols and standards, there are capabilities to self-configure itself in an active network infrastructure of IoT,²⁰ and it recognizes every entity as “Things”. There must be a seamless integration of this cyber-physical interface of IoT system architecture, as all the things are connected in the system. The set of connections, security, and business processes and models are the various factors of IoT architectures. There is a need to consider the multiple aspects of interoperability, scalability, and extensibility in IoT architecture designing for integrating the existing heterogeneous network devices and smooth transition.²¹ IoT harmonizes how people connect to the physical world using public services such as cloud computing.¹⁵

Due to the integration of digital technologies into physical systems, IoT and cyber-physical systems look like interchangeable concepts, but their focuses and characteristics are distinct. IoT is a broader concept that minimizes human intervention and focuses on system-to-system interactions. Whereas, the cyber-physical system emphasizes human-to-system interactions.²² However, the application of both concepts is to detect security threats, network intrusion, attack paths and malware. In recent years, both have witnessed a number of success stories and rapid development.²³

Internet of things technologies

The base technology for IoT is an RFID system (widely used in industries) to broadcast the info to an end-user. Internet protocols & applications, intelligent sensors, embedded devices and barcoding are the other methodologies adopted by IoT.²⁴ This technology has far-reaching trends, but its diffusion and perceived adoption are nascent.¹⁵ However, certain other techniques have priority in the organizations, as shown in Table 1.

Table 1.

IoT technologies.

IoT technology	Description
IoT security	There are usually risks of data thefts and cyber-attacks because, across the world, trillions of devices are interacting in IoT systems. Therefore, standardization and trustworthiness are instantly needed.²⁵
IoT analytics	There is a need to analyze the data using algorithms and modern analytical tools because IoT captures voluminous data.²⁶
IoT device management	Organizations must have the requisite capability to align millions of connected devices.²⁷
IoT Processors	There will be a demand for processors that can deliver encryption solutions and robust security.²⁴
Distributed Stream processing	For data analysis in real-time and to support distributed streams, parallel computing architecture platforms are needed.²⁴
Platforms	There should be efficient device management, data acquirement, storage analysis, and sharing under the platforms of IoT infrastructure system components.²⁸

Internet of things for agile manufacturing

In the global economy, more data are generated by manufacturing than in any other sector. However, due to a lack of support systems and technology, most of the data are used to its curtailed potential or lost. However, IoT can contribute substantially to task automation, quality assurance, effective monitoring, and proactive maintenance of the machine's systems when combined with data analytics. Manufacturing organizations like Toyota are already shifting from traditional manufacturing to more digitized ones to meet market demand.¹⁵

Numerous enterprises intend to implement lean manufacturing systems in their manufacturing processes to meet the customization demands, increase product variety, reduce lead customer times, and eliminate waste. Furthermore, to encounter highly volatile market demand, companies aim to keep the system more agile.²⁹ Machine and mechanical setups could be made agile up to some extent by enhancing the combinatory sets of electrical, mechanical and software parameters.³⁰ Agility means “capacity to move easily and quickly.” Agile manufacturing is characterized by satisfying individual customer preferences where fluctuations may happen in the roles of competitor firms, suppliers and customers. Agile manufacturing is a significant attempt to respond to the threat of newly industrialized nations.³¹ Traditional manufacturing is the conventional production method and depends on established processes and technology, whereas agile manufacturing augments efficiency and adaptability to provide customer-driven products and services by embracing advanced automation and prompt prototyping.³² There is a need for continuous development of new technological approaches for an agile manufacturing base, and IT is argued to be the essential condition to realize real agile manufacturing.³³ Information is a more important constraining factor in complex global organizations than labor. Without an efficient infrastructure of information access, sharing and distribution among all stakeholders, concurrent development of products cannot be achieved. The supportive role of IT is a novel feature of an agile environment.³⁴

Currently, all successful enterprises invariably rely on IT, but real agile manufacturing is a strategic process of thriving in the competitive environment by excellently responding to changing markets.³³ One of the key enlisted strategies and fundamental enablers for agile manufacturing is “automation and information technology”. There is a lot of change in the working culture of organizations with advancements in IT, and organizations are spending a large budget on constructing IT infrastructure to attain higher levels of firm agility.³⁵ In order to enhance the consumers’ experience and make it more convenient, personalized and engaging, digital technologies such as IoT play an important role³⁶ as the advancement of IT supports all key shifts of the manufacturing paradigm. IoT gives speedy access to information about physical objects, leading to state-of-the-art services with high productivity and high efficacy. IoT changed the methods of handling and operating production, delivery, maintenance, transportation and product recycling. Hence, IoT is competent in offering robust solutions at all levels.³⁷ IoT competencies are imperative for organizations to sense and grab business opportunities and challenges.³⁵ In short, by adopting IoT infrastructure, manufacturing organizations can benefit significantly in all aspects.³⁷ Moreover, the application of IoT in manufacturing enterprises advances the competence of production processes, followed by fulfilment of customer specifications.³⁸

Barriers prone to the Internet of Things adoption

IoT is supposed to have an enormous influence in the future on the global economic platform.³⁹ It is specified that the optimization and transformation of the international economy will depend merely on the implementation level of IoT devices.¹⁵ The current study surveyed accessible literature to ascertain possible barriers to adopting IoT in the following.

Lack of standardization

Over the entire IoT network, to ensure data safety and efficient and smooth integration between various vendors, IoT communication protocol established global standards. IoT is proficient in handling end-user platforms, cloud networks and sensitive devices because of the existence of global standards.⁴⁰ Standardization is useful for information interchanging, composting, controlling, and surveying.⁴¹ Various standards like communication, identification and security play a serious role in implementing IoT successfully. A common adoption barrier has been identified as a lack of standardization for most of the new technologies.⁴² For IoT practitioners, every day makes a difficult situation because of the fragmentation of standards with new ones.⁴³ Furthermore, standardization will be required to avoid chaos.⁴⁴

Inadequate internet connectivity

Poor internet connectivity challenges developing countries.⁴⁵ In IoT adoption and diffusion, one of the key challenges is electricity problems and poor internet connectivity. The Internet must be available to all partners.^8,43

Lack of mobility/reliability

Mostly, IoT objects have a certain degree of mobility and are not fixed. This is of paramount importance to efficiently manage and support the mobility of this large amount of objects.⁴⁶ Reliability challenges are enhanced due to the massive number of smart technologies and widespread participation. The failure risk is superior and the interconnections among them are less trustworthy with mobility.^47,48

Security and privacy

There are substantial fluctuations regarding cybersecurity in the companies’ perception in a post-COVID-19 world.⁴⁹ IoT security challenges must be closely considered and nullified to enhance public trust and experience a significant adoption rate.¹⁵ There may be calls for efficient security mechanisms because billions of devices are connected through IoT. In employing IoT networks, a major concern is the personal privacy issue (data ownership) because it can be easy to trace and hack connected objects and devices.⁴³ Security issues are central in IoT, and a series of properties, such as privacy, non-repudiation, availability, integrity, confidentiality and authentication, must be guaranteed to ensure the security of data and the entire IoT system.⁴⁶ IoT systems are vulnerable to external security risks due to different issues like authorization, software protection, internet connectivity, and data encryption.²⁴ Client privacy, access control, data authentication and resilience to attacks are also challenges in security and privacy protection.²¹ When the user’s signals are interrupted, his or her privacy may be compromised, and this issue should be addressed to provide user confidence in terms of privacy.⁵⁰ In order to ensure control over the flow and exchange of sensitive information and safeguarding of personal data across the manufacturing chains, holistic multi-layer approaches to security are required.⁵¹

High adoption and operating costs

There may be irreversibility of investments and technological risk related to financial loss because high-end technical infrastructure support is required for IoT solution implementation.^24,52 Moreover, the maintenance cost of these devices will rise in the future.⁵³ Manufacturers need benchmarks about IoT’s ability to lead to improvements across cost, time, and quality axes to provide a compelling proposition for IoT adoption.⁵¹ In addition, many sensing and triggering devices are employed in IoT applications, which raises their cost.⁴³

Lack of government regulations

An important consideration within the IoT industry is effective governance, and the effective governance plan is harder to achieve with the involvement of heterogeneous technologies.⁵⁴ Legal information systems need to be developed to support the expansion of IoT and massive development as it ensures security standards. Judicial laws should be guided by the limitations of sensitive frequency bands.⁴¹ Organizations and governments must work synchronously to support and promote technological initiatives and resolutions.⁵⁵ There is a continual downward trend because of weak governance in IoT networks and devices’ security, and the absence of effective regulation and standards.⁵⁴

Long payback period

Massive series of sensing and stimulating devices are required in manufacturing organizations. There may be an increase in the payback period, depending on the finances and IoT implementation size.⁴³ Due to the long payback period and unawareness of capturing the potential of IoT, small and medium enterprises may be discouraged from upgrading their systems.¹³ Furthermore, being financially unstable, many organizations do not believe in long-term payback and investment⁵⁶ because investment in IoT always carries a substantial risk of monetary loss.²⁴

Lack of long-term vision and top management commitment

An organization will lose its competitive advantage and fail to leverage the opportunities without a good long-term vision.⁵⁷ Employee commitment and involvement are crucial for implementing any new process or upgradation program in an enterprise. Therefore, it is essential to frame an efficient policy for human resources from the management team as it includes molding the mentality of employees to trust the organization, and for the betterment of the organization, changing the beliefs, outlook, and routines of employees.^24,29 Due to a lack of familiarity, perceived complexity, cultural resistance to change, security apprehensions, and high cost, top leaders may face challenges in understanding IoT.⁵⁸

Compatibility issues and seamless integration

Compatibility and integration problems prevail while embracing IoT due to heterogeneity in technology adoption.⁵⁰ Integration of existing legacy systems with IoT technologies acts as an IoT adoption barrier. Conformity with the current industrial automation is necessary for smooth integration between systems.⁵⁹ Haddud, DeSouza²⁴ also confirmed that integrating IoT with used strategic, analytics, and operational technologies is the greatest barrier to successfully adopting IoT.

Limited skilled workforce

High-end functional and technical skills are required in IoT systems to make certain system’s adaptable, and there should be a user-friendly interface, installation, and management of the IoT network.^8,44 So, acquiring the desirable supportive staff with the right skills and familiarity for designing and deploying IoT solutions is another key IoT adoption barrier.²⁴ The system is adversely influenced by the lack of technical skills in the workforce.⁵⁹

Lack of identification and validation

Mainly, the literature on IoT covers technological and functional challenges, while multi-application research is still in the emerging stage.⁵⁹ It discourages small enterprises from adopting innovative & troublesome technology and implementing new business models^24,44 because the top challenge to IoT adoption was perceived as a lack of clear comprehension of the IoT benefits.²⁴

Complex Architecture

Currently, there is a non-existence of universally acceptable IoT architecture.⁵⁰ Compliance with the IoT philosophy and the requirements and constraints of the manufacturing environment to manage the devices and services is the role of architecture.³⁰ The upgraded technology can be adopted only with the proper architecture.⁴⁸ In the case of wireless systems, there is a critical challenge of initial plan, design and IoT architecture selection.²⁴ There is a need for standards-based reference architecture implementation to integrate digital enterprise systems in the manufacturing chain and for industrial automation.⁵¹ Designing an efficient, flexible, and secure architecture that is able to cope with complex IoT scenarios is a major challenge for sensor networking and storage. In this concern, the service-oriented architecture (SOA) approach has often been followed in the last couple of years.⁵⁴

High energy consumption

The critical area of concern is efficient power consumption for IoT implementation. Energy efficiency is a crucial need for an operational Internet of Things device. Power is a significant concern for IoT’s commercial applications.⁴⁶ System power consumption can be enhanced in a way that various alternative power storage devices lower the power usage of the electronic module. There will be a rising call for energy with the increase in the number of data centers, networks and IoT devices. Consequently, energy costs are expected to increase continuously.^48,60

Transparency and trust

At every production stage, transparency and trust are important for agile manufacturing. IoT adoption creates the problem of trust as the cloud architecture is centralized,⁶¹ and trust issues are typically associated with emergent technologies.⁶² That is why most stakeholders did not trust the provided IoT system.⁶³ For IoT acceptance, adoption and continued use, trust remains the number one barrier and is a central policy concern that needs to be handled.⁶²

Displacement of human resources and employees’ resistance to new technologies

IoT adoption at social and organizational levels will create relevant challenges concerning reorganizing skilled manpower.⁴⁴ The displacement of human resources will be crucial when organizations implement disruptive technologies.²⁴ Thus, instead of displacing or replacing them, the involvement of existing human resources is important.⁶⁴ Therefore, employees always resist new technologies and practices. It is challenging to change the mentality of workers from their old way of working and basic principles and to recognize employees who are not relaxed with the use of new technologies.²⁹

Scalability issues

In order to address the significant issue of scalability, organizations must keep themselves updated as more physical devices are assessed to get joined in the forthcoming time with the increase in the size of the IoT network.⁸

Research methodology

Research design

The present study uses integrated AHP-TOPSIS and ISM techniques to better understand the identified 16 IoT adoption barriers. For evaluating the relative importance of each barrier and assigning weights, AHP is useful. While ranking the identified alternatives or barriers, the weights obtained from AHP are used as inputs in TOPSIS. Then, the ISM method was used to describe the relationship among various barriers. Finally, to classify these barriers, MICMAC analysis was carried out.

Sample and procedure

In the current study, the manufacturing sector of Pakistan and top-ranked educational institutions were the sources of data collection. The authors painstakingly nominated modern manufacturing firms from the country's largest provinces where the IoT concept has been implemented. To prompt expert opinions while using AHP and ISM, there were invitations to experienced managers and academicians with more than ten years of experience. Researchers can choose experts based on their knowledge, so the purposive sampling technique was applied here. Five from industry and three from academia (a total of eight experts) participated among the 18 contacted experts. In the case of ISM, three rounds of (Delphi) discussions were executed to come to a consensus about the association amongst barriers.

There are different opinions about the sample size, and a minimum of eight experts is recommended because of the absence of a significant correlation between the panel members.⁶⁵ Hyun, Cho⁶⁶ selected seven experts for their research to efficiently evaluate various alternatives, as the Delphi method requires extensive knowledge and experience. Bolaños, Fontela⁶⁷ and Govindan, Azevedo⁶⁸ stated that in case of ISM participation, there should not be too many participators. Therefore, Etemadi, Van Gelder⁶⁹ and Kumar, Bhamu⁷⁰ engaged five and eight experts, respectively, to answer the questions. Similarly, in the study of Deshpande and Nagendra⁷¹ only five experts participated. While using Delphi, most studies engaged the number of eight panelists.⁷² Also, in the current study, the views of eight-panel members with expertise were considered effective for analysis.

Using convenient sampling, data were collected for TOPSIS from operational-level users, including supervisors and managers of various functional areas of the manufacturing organizations. Questionnaires were distributed to 710 users to rate on a five-point Likert scale varying from “strongly disagree” (1) to “strongly agree” (5). 61% was the effective response rate, and 433 questionnaires were completely filled out. For both studies, the respondents’ characteristics have been summarized in Table 2.

Table 2.

Demographics of the respondents.

Technique	Demographic	Frequency	Percent
AHP and ISM	Experts (Having 10+ Years of Experience)
	Academia	3	37.5
	Industry	5	62.5
TOPSIS	Operational Level Users (Managers and Supervisors)
	Age (Years)
	25-30	78	18
	31-35	143	33
	36-40	121	28
	Above 40	91	21
	Education (Years)
	14 Years	351	81
	16 Years	56	13
	18 & above	26	6

Techniques used in the study

Analytical hierarchy process

Following are the steps of AHP:^73–75

Step 1

After goal setting, a decision hierarchy is designed. The 1^st level is about prioritizing the IoT implementation barriers. There are attributes-based groups of all barriers at 2^nd level. The 3^rd level is about all barriers.

Step 2

Now, to create matrices of pairwise comparisons of all the barriers and groups, experts are solicited using a 1-9 preference scale described by Saaty.⁷⁶

Step 3

There is the start of the mathematical process for normalizing each matrix. A normalized matrix is obtained when each entry in the column of a matrix is divided by a respective column sum. The eigenvector of a matrix is obtained by averaging all values in a normalized matrix row. Then, the principal eigenvalue known as λ_max is obtained by adding the product of the eigenvector and the sum of the columns of the reciprocal matrix.

Step 4

A consistency test is conducted to verify the assessment of experts:

C R (Consistency Ratio) = \frac{C I}{R I}

where, RI = Random Index, and

C I (Consistency Index) = \frac{λ_{\max} - n}{n - 1}

(e.g., the value of RI for a 4 × 4 matrix is 0.89).⁷⁷ A review is required if $C R > 0.1$ which shows that elements are not properly compared.

TOPSIS

Based on 16 IoT adoption barriers, a questionnaire was developed for TOPSIS to determine the final ranking. The principal notion in the TOPSIS approach is that the most preferred alternative should be farthest away from the negative ideal solution (NIS) and closest to the positive ideal solution (PIS).⁷⁸ Quick identification of the best alternative and from decision-makers, only limited subjective input is needed are the benefits of TOPSIS.⁷⁹ Following are the steps involved in TOPSIS:

Step 1

Initially, normalize that decision matrix ( $x_{i j}$ ). Normalized value $(r_{i j})$ calculated as:

r_{i j} = \frac{x_{i j}}{\sqrt{\sum x_{i j}^{2}}}

(1)

where, i = 1,…..,m and j = 1,……,n

Step 2

In this step, the weighted normalized decision matrix is constructed by using equation (2):

ν_{i j} = r_{i j} \times w_{j}

(2)

Where $w_{j}$ is the weight acquired from AHP in the current study.

Step 3

Here, the positive ideal solution V+ and the negative ideal solution V- are determined. In each column of the weighted normalized decision matrix, maximum and minimum values are denoted by $A_{+}$ and $A_{-}$ , respectively are:

A_{+} = {(\max (v_{i j} | i = 1,2, \dots, m) | j \in J_{-}), (\min (v_{i j} | i = 1,2, \dots ., m) | j \in J_{+})}

(3)

A_{-} = {(\min (v_{i j} | i = 1,2, \dots, m) | j \in J_{+}), (\max (v_{i j} | i = 1,2, \dots ., m) | j \in J_{-})}

(4)

Step 4

Now, a separation measure is planned for each alternative. From a positive ideal alternative, separation is computed as:

S_{+}^{i} = \sqrt{\sum_{j = 1}^{m} {({v_{i j} - v_{j}^{+}}_{.})}^{2}}, i = 1,2,3 \dots . ., m

(5)

$S_{+}^{i}$ stands for distance i from the positive ideal

From a negative ideal alternative, separation is computed as:

S_{-}^{i} = \sqrt{\sum_{j = 1}^{m} {({v_{i j} - v_{j}^{-}}_{.})}^{2}}, i = 1,2,3 \dots . ., m

(6)

$S_{-}^{i}$ stands for distance i from the negative ideal

Step 5

Now, for each barrier, relative closeness $(C_{i})$ to the ideal solution is computed as:

C_{i} = \frac{S_{-}^{i}}{S_{-}^{i} + S_{+}^{i}}

(7)

Step 6

On the basis of relative closeness values, barriers are ranked in decreasing order of preference.

Interpretive Structural Modeling

Interpretive Structural Modeling (ISM) is considered a well-established methodology than Fuzzy Cognition Maps, DEMATEL, Graph Theory, Structural Equation Modeling (SEM) and Analytical Network Process (ANP) for ascertaining the relationships between distinct variables.^72,80 ISM approach was proposed by Warfield.⁸¹ The main logic behind this approach is to build a structural model by using the practical experience and knowledge of experts. Finding the interrelationship among elements is the focus of ISM, which is helpful for individuals. This approach develops a well-defined hierarchal model from unclear and poorly articulated data. ISM is helpful for both individuals and groups to provide a clear picture of the interrelationships.^82,83

Following are the steps of the ISM methodology:^84,85

1. Documents the IoT adoption barriers through the survey.

2. Establish the contextual relationships among identified elements to show the leading role of each element.

3. To show a pairwise relationship between elements, develop the structural self-interaction matrix (SSIM).

4. From SSIM, form the Initial Reachability Matrix (IRM) by converting the information in each entry of SSIM into 1s and 0s. Now, develop the Final Reachability Matrix (FRM) after detecting transitivity.

5. Partitioning the FRM into different levels.

6. Draw a directed graph based on FRM.

7. Build the ISM model after replacing the nodal elements with statements.

8. Analyze the ISM model to check for any inconsistency.

MICMAC

Duperrin and Godet⁸⁶ developed MICMAC (Cross-Impact Matrix Multiplication Applied to Classification) analysis, “an indirect classification technique”, which is helpful for critically investigating the scope of each barrier. This technique addresses the driving and dependence power of all obstacles individually. There are four clusters in MICMAC, namely, Autonomous, Dependent, Linkage and Independent, in which different IoT adoption barriers have been placed. Many ISM-based studies use the MICMAC approach to categorize elements because the ISM lacks this capability.⁸⁷

A real-case application of the proposed model (data analysis and results)

Selection of the most prominent IoT adoption barriers among numerous alternatives that dominate each other in different characteristics is difficult, so the proposed model is applied. The numeric application with the results has been described here based on the steps in the previous section.

Computation of weights

A three-level hierarchy structure of different IoT adoption barriers with respect to AHP analysis was developed. The first level aimed to identify the most prominent IoT adoption barriers, unlike the conventional AHP structure. Four groups (made by experts, as shown in Table 3) of all barriers were placed at the criteria level. At the third level, there were all IoT adoption barriers/alternatives. Pairwise comparisons of barriers and groups were carried out using the fundamental preference scale of absolute numbers established by Saaty.⁷⁶ The assigned and normalized relative weights of all IoT adoption barriers were obtained through experts' judgment and by following the AHP steps described in the literature, as displayed in column 11 of Table 4.

Table 3.

Groups of barriers.

Group	Barrier no.	IoT adoption barriers
Group 1 Adaptability barriers	B1	Lack of standardization
	B2	Inadequate internet connectivity
	B9	Compatibility issues and seamless integration
	B12	Complex architecture
Group 2 Pecuniary barriers	B5	High adoption and operating costs
	B7	Long payback period
	B11	Lack of identification and validation
	B13	High energy consumption
Group 3 People barriers	B6	Lack of government regulations
	B8	Lack of long-term vision and top management commitment
	B10	Limited skilled workforce
	B15	Displacement of human resources and employees’ resistance to new technologies
Group 4 Reliability barriers	B3	Lack of mobility/Reliability
	B4	Security and privacy
	B14	Transparency and trust
	B16	Scalability issues

Table 4.

The importance weights of all barriers.

Practices	Expert 1	Expert 2	Expert 3	Expert 4	Expert 5	Expert 6	Expert 7	Expert 8	Relative Preference Weights (Geometric Mean)	Global preference weights
Adaptability barriers									.5056
B1	.1169	.1530	.1186	.1467	.2400	.3369	.1379	.1207	.1595	.806
B2	.4868	.4335	.4392	.4214	.4873	.3577	.3951	.4902	.4365	.2207
B9	.3041	.3281	.3069	.3588	.1794	.0849	.3565	.3037	.2554	.1291
B12	.0922	.0854	.1353	.0731	.0933	.2205	.1106	.0854	.1053	.0532
C.I	.0474	.0644	.0325	.0575	.0694	.0326	.0656	.0323	.0478
C.R	.0527	.0715	.0361	.0639	.0771	.0362	.0729	.0359	.0531
Pecuniary barriers									.2642
B5	.5631	.4886	.4659	.4715	.3826	.5722	.5190	.4170	.4810	.1271
B7	.0904	.0944	.0947	.2550	.1272	.0843	.0807	.2224	.1188	.0314
B11	.1382	.1471	.1812	.1653	.1423	.1601	.1401	.2502	.1625	.0429
B13	.2082	.2700	.2582	.1083	.3479	.1834	.2601	.1105	.2030	.0536
C.I	.0682	.0154	.0416	.0153	.0071	.0251	.0033	.0153	.0167
C.R	.0758	.0171	.0463	.0170	.0078	.0279	.0037	.0170	.0186
People barriers									.1434
B6	.5798	.4800	.5443	.5564	.6161	.4883	.5190	.5317	.5378	.0771
B8	.1637	.1799	.1497	.2285	.1427	.2328	.0807	.1856	.1631	.0234
B10	.1637	.2484	.2399	.1366	.1697	.0893	.1401	.1856	.1641	.0235
B15	.0929	.0917	.0662	.0786	.0715	.1896	.2601	.0972	.1054	.0151
C.I	.0093	.0360	.0273	.0103	.0133	.0083	.0033	.0014	.0091
C.R	.0103	.0400	.0303	.0115	.0148	.0092	.0037	.0015	.0101
Reliability barriers									.0868
B3	.0969	.1058	.1111	.1808	.1967	.1468	.2125	.2239	.1519	.0132
B4	.5312	.5140	.4706	.5011	.5959	.5059	.4954	.5247	.5162	.0448
B14	.0925	.0947	.1003	.2121	.1320	.0852	.1001	.0810	.1068	.0093
B16	.2795	.2855	.3179	.1060	.0755	.2621	.1920	.1704	.1906	.0165
C.I	.0015	.0069	.0284	.0202	.0393	.0669	.0052	.0153	.0137
C.R	.0016	.0076	.0316	.0225	.0437	.0743	.0057	.0170	.0152

Determination of final ranking

710 users (managers and supervisors) were given questionnaires to complete at the operational level of various manufacturing businesses, and they were asked to score each item on a five-point Likert scale. Answers to all the questions in the instrument were provided by sixty-one per cent (61%) of the respondents. For this study, 433 users completely fulfilled the questionnaire after discarding incomplete questionnaires. A decision matrix has been prepared based on information provided by users, shown in Table 5.

Table 5.

Initial decision matrix.

Alternatives/IoT Adoption Barriers		X _i 5	X _i 4	X _i 3	X _i 2	X _i 1
B1	Lack of standardization	170	137	96	23	7
B2	Inadequate internet connectivity	193	110	90	32	8
B3	Lack of mobility/Reliability	7	82	164	81	99
B4	Security and privacy issues	134	180	84	20	15
B5	High adoption and operating costs	202	115	83	29	14
B6	Lack of government regulations	190	110	93	32	8
B7	Long payback period	184	165	57	22	5
B8	Lack of long-term vision and top management commitment	185	140	79	25	4
B9	Compatibility issues and seamless integration	129	140	120	30	14
B10	Limited skilled workforce	11	85	165	81	91
B11	Lack of identification and validation	9	90	174	71	89
B12	Complex architecture	165	134	96	20	18
B13	High energy consumption	39	115	203	40	36
B14	Transparency and trust	34	121	202	48	32
B15	Displacement of human resources and employees’ resistance to new technologies	162	152	83	28	8
B16	Scalability issues	140	176	77	31	9

Then, the decision matrix is normalized using equation (1) shown in Table 6.

Table 6.

Normalized decision matrix.

Practices		$X_{i} 5 / \sqrt Σ X_{i j}^{2}$	$X_{i} 4 / \sqrt Σ X_{i j}^{2}$	$X_{i} 3 / \sqrt Σ X_{i j}^{2}$	$X_{i} 2 / \sqrt Σ X_{i j}^{2}$	$X_{i} 1 / \sqrt Σ X_{i j}^{2}$
B1	Lack of standardization	0.7092	0.5715	0.4005	0.0959	0.0292
B2	Inadequate internet connectivity	0.7977	0.4547	0.3720	0.1323	0.0331
B3	Lack of mobility/Reliability	0.0313	0.3666	0.7332	0.3621	0.4426
B4	Security and privacy issues	0.5562	0.7472	0.3487	0.0830	0.0623
B5	High adoption and operating costs	0.8115	0.4620	0.3335	0.1165	0.0562
B6	Lack of government regulations	0.7894	0.4570	0.3864	0.1329	0.0332
B7	Long payback period	0.7226	0.6480	0.2239	0.0864	0.0196
B8	Lack of long-term vision and top management commitment	0.7508	0.5682	0.3206	0.1015	0.0162
B9	Compatibility issues and seamless integration	0.5671	0.6155	0.5276	0.1319	0.0616
B10	Limited skilled workforce	0.0495	0.3824	0.7423	0.3644	0.4094
B11	Lack of identification and validation	0.0397	0.3969	0.7673	0.3131	0.3925
B12	Complex architecture	0.7028	0.5708	0.4089	0.0852	0.0767
B13	High energy consumption	0.1608	0.4740	0.8368	0.1649	0.1484
B14	Transparency and trust	0.1389	0.4943	0.8252	0.1961	0.1307
B15	Displacement of human resources and employees’ resistance to new technologies	0.6780	0.6362	0.3474	0.1172	0.0335
B16	Scalability issues	0.5836	0.7337	0.3210	0.1292	0.0375

After obtaining the normalized decision matrix, it is weighted by multiplying it with the weighting factor (obtained from AHP) using equation (2) exhibited in Table 7.

Table 7.

Weighted normalized decision matrix.

Practices		V5j	V4j	V3j	V2j	V1j
B1	Lack of standardization	0.0572	0.0461	0.0323	0.0077	0.0024
B2	Inadequate internet connectivity	0.1761	0.1003	0.0821	0.0292	0.0073
B3	Lack of mobility/Reliability	0.0004	0.0048	0.0097	0.0048	0.0058
B4	Security and privacy issues	0.0249	0.0335	0.0156	0.0037	0.0028
B5	High adoption and operating costs	0.1031	0.0587	0.0424	0.0148	0.0071
B6	Lack of government regulations	0.0609	0.0353	0.0298	0.0103	0.0026
B7	Long payback period	0.0227	0.0203	0.0070	0.0027	0.0006
B8	Lack of long-term vision and top management commitment	0.0176	0.0133	0.0075	0.0024	0.0004
B9	Compatibility issues and seamless integration	0.0732	0.0795	0.0681	0.0170	0.0079
B10	Limited skilled workforce	0.0012	0.0090	0.0175	0.0086	0.0096
B11	Lack of identification and validation	0.0017	0.0170	0.0329	0.0134	0.0168
B12	Complex architecture	0.0374	0.0304	0.0218	0.0045	0.0041
B13	High energy consumption	0.0086	0.0254	0.0449	0.0088	0.0080
B14	Transparency and trust	0.0013	0.0046	0.0077	0.0018	0.0012
B15	Displacement of human resources and employees’ resistance to new technologies	0.0102	0.0096	0.0053	0.0018	0.0005
B16	Scalability issues	0.0097	0.0121	0.0053	0.0021	0.0006

Positive and negative ideal solutions, as shown in Table 8 were determined using equations (3) and (4).

Table 8.

Ideal best and negative ideal solution matrix.

	Ideal (Best) A+		Negative-Ideal A-
$Vi 1^{+}$	0.1761	$Vi 1^{-}$	0.0004
$Vi 2^{+}$	0.1003	$Vi 2^{-}$	0.0046
$Vi 3^{+}$	0.0821	$Vi 3^{-}$	0.0053
$Vi 4^{+}$	0.0170	$Vi 4^{-}$	0.0018
$Vi 5^{+}$	0.0168	$Vi 5^{-}$	0.0004

Using equations (5) and (6), Euclidean separation distances have been computed. After that, by using equation (7), relative closeness (Ci) to the ideal solution of each barrier was calculated. Table 9 summarizes the results along with the proposed rank.

Table 9.

Euclidean separation distance.

	$S i^{+}$	Distance $i$ from Positive Ideal	$S i^{-}$	Distance $i$ from Negative Ideal	Relative Closeness ( $Ci$ ) Values	The Proposed Rank
B1	$S 1^{+}$	0.1409	$S 1^{-}$	0.0756	0.3492	4
B2	$S 2^{+}$	0.0155	$S 2^{-}$	0.2162	0.9333	1
B3	$S 3^{+}$	0.2133	$S 3^{-}$	0.0076	0.034	15
B4	$S 4^{+}$	0.1792	$S 4^{-}$	0.0394	0.1802	8
B5	$S 5^{+}$	0.0935	$S 5^{-}$	0.1228	0.5678	2
B6	$S 6^{+}$	0.1431	$S 6^{-}$	0.0727	0.3367	5
B7	$S 7^{+}$	0.1898	$S 7^{-}$	0.0274	0.1260	10
B8	$S 8^{+}$	0.1968	$S 8^{-}$	0.0194	0.0896	11
B9	$S 9^{+}$	0.1062	$S 9^{-}$	0.1231	0.5368	3
B10	$S 10^{+}$	0.2079	$S 10^{-}$	0.0174	0.0770	12
B11	$S 11^{+}$	0.1994	$S 11^{-}$	0.0365	0.1546	9
B12	$S 12^{+}$	0.1676	$S 12^{-}$	0.0483	0.2236	6
B13	$S 13^{+}$	0.1876	$S 13^{-}$	0.0467	0.1993	7
B14	$S 14^{+}$	0.2139	$S 14^{-}$	0.0027	0.0124	16
B15	$S 15^{+}$	0.2053	$S 15^{-}$	0.0110	0.0511	14
B16	$S 16^{+}$	0.2046	$S 16^{-}$	0.0119	0.0551	13

Barriers with the maximum Ci values were the major causal IoT adoption barriers. IoT adoption barriers can be ranked as B2-B5-B9-B1-B6-B12-B13-B4-B11-B7-B8-B10-B16-B15-B3-B14 in the decreasing order of preference.

Making of causal model showing the relationship among barriers

To develop the ISM model, the top 10 IoT adoption barriers were extracted from Table 9 to avoid complexity.⁸⁸ The top 10 IoT barriers were recoded from B (barrier) to BM (barrier for model) for respondents’ convenience, as shown in Table 10. There is a need for a well-established modeling approach to elucidate the connections among 10 barriers; thus, the ISM technique is appropriate.

Table 10.

Renaming of barriers’ sequence.

Rank	Value	Barrier no.	Alternatives/Barriers
1	0.9333	BM1	Inadequate internet connectivity
2	0.5678	BM2	High adoption and operating costs
3	0.5368	BM3	Compatibility issues and seamless integration
4	0.3492	BM4	Lack of standardization
5	0.3367	BM5	Lack of government regulations
6	0.2236	BM6	Complex architecture
7	0.1993	BM7	High energy consumption
8	0.1802	BM8	Security and privacy issues
9	0.1546	BM9	Lack of identification and validation
10	0.1260	BM10	Long payback period

Structural self-interaction matrix development

Between any two barriers, a value of V, A, X or O was used to categorize the relationship in the structured questionnaire:

V = barrier $i$ impacts barrier $j$

A = barrier $j$ impacts barrier $i$

X = interrelationship between $(i, j)$ barriers

O = between $(i, j)$ barriers, there is no relationship

The contextual relationship between each pair of elements was developed using experts’ opinions based on the Delphi technique. Respondents completed the questionnaires, and SSIM was developed based on these relationships, as shown in Table 11.

Table 11.

Structural self-interaction matrix.

Reachability matrix formation

Using equations (8), V, A, X and O values are replaced into 1s and 0s and IRM is developed. Then, the final reachability matrix (FRM) is developed by incorporating transitivity, as shown in Table 12. FRM was computed according to the Boolean rules in equation (9). In equation (9), the final reachability matrix is shown by

R

r

denotes less than or equal to a number of barriers,

I

is a unit matrix, and

A

represents the initial reachability matrix.

(i, j) = {\begin{cases} V, (i, j) = 1, (j, i) = 0; \\ A, (i, j) = 0, (j, i) = 1; \\ X, (i, j) = 1, (j, i) = 1; \\ O, (i, j) = 0, (j, i) = 0; \end{cases}

(8)

R = {(A + I)}^{r} = {(A + I)}^{r - 1} \neq {(A + 1)}^{r - 2} \neq \dots \neq (A + 1), r \leq 9

(9)

Table 12.

Final reachability matrix.

Barriers	BM1	BM2	BM3	BM4	BM5	BM6	BM7	BM8	BM9	BM10	Driving power
BM1	1	1	1	1	1	1	1	1	1	1	10
BM2	1	1	1	1	1	1	1	1	1	1	10
BM3	0	0	1	0	0	0	1	1	1	1	5
BM4	0	0	1	1	0	1	1	1	1	1	7
BM5	1	1	1	1	1	1	1	1	1	1	10
BM6	0	0	1	1	0	1	1	1	1	1	7
BM7	0	0	1	0	0	0	1	1	1	1	5
BM8	0	0	1	0	0	0	1	1	1	1	5
BM9	0	0	0	0	0	0	0	0	1	1	2
BM10	0	0	0	0	0	0	0	0	1	1	2
Dep. power	3	3	8	5	3	5	8	8	10	10

Hierarchical analysis

From FRM, reachability set R(Pi) and antecedent set A (Pi) are yielded for each barrier. A particular barrier and the other may help achieve the reachability set R(Pi), whereas a specific barrier and the other barriers may influence it to constitute the antecedent set A(Pi). For all barriers, intersection set C (Pi) represents the intersection of corresponding R (Pi) and A (Pi). The top-level position is given to that barrier, where R(Pi) = C(Pi), because it does not assist in achieving any other barrier above its own level. For further analysis, the top-level identified barrier is to be removed. This process is iterated to determine the levels of all the barriers, as shown in Table 13, and these levels support the structure of the digraph.

Table 13.

Hierarchical division of IoT adoption barriers.

Barriers	R(Pi)	A(Pi)	C(Pi)	Level
BM9	9,10	1,2,3,4,5,6,7,8,9,10	9,10	I
BM10	9,10	1,2,3,4,5,6,7,8,9,10	9,10	I
BM3	3,7,8	1,2,3,4,5,6,7,8	3,7,8	II
BM7	3,7,8	1,2,3,4,5,6,7,8	3,7,8	II
BM8	3,7,8	1,2,3,4,5,6,7,8	3,7,8	II
BM4	4,6	1,2,4,5,6	4,6	III
BM6	4,6	1,2,4,5,6	4,6	III
BM1	1,2,5	1,2,5	1,2,5	IV
BM2	1,2,5	1,2,5	1,2,5	IV
BM5	1,2,5	1,2,5	1,2,5	IV

Based on the above-defined interactions, a digraph was drawn, and by removing transitivity links, this digraph was converted to an interpretive structural model of barriers, as shown in Figure 1. Figure 1 shows that barriers have been placed in the same layer, which is at a similar level in the analysis mentioned above.

Figure 1.

Hierarchical structure of barriers (ISM graph).

Clustering of barriers (MICMAC analysis)

For investigating the scope of each barrier, MICMAC analysis is used. Analyzing each barrier's driving and dependence power is the key objective of MICMAC.⁸⁹ A diagram was built, as shown in Figure 2, and barriers have been categorized into four clusters.

1. Autonomous barriers are in the first cluster with weak driving and dependence powers. Because of weak links among them, such barriers are considered detached. In the current case, the autonomous cluster is empty, meaning the administration should pay severe devotion to all IoT adoption barriers.

2. The second cluster is about dependent barriers with strong dependence and weak driving power, and based on other barriers, their development mainly occurs. Such barriers are plotted in the right-bottom portion of the graph and exhibit the attributes of output variables. Five barriers, viz., Compatibility issues and seamless integration (BM3), High energy consumption (BM7), Security and privacy issues (BM8), Lack of identification and validation (BM9) and long payback period (BM10) are placed in this cluster.

3. Linkage barriers establish the third cluster with strong dependence as well as strong driving powers. They are placed in the right-top portion on both scales, and their values are above the middle. Such barriers not only drive the system but are also dependent on other barriers. In our study, Lack of standardization (BM4) and Complex Architecture (BM6) are in this cluster.

4. Independent barriers constitute the last cluster and are referred to as driver barriers. These barriers influence all the other barriers in the system considerably. In this research, Inadequate internet connectivity (BM1), High adoption and operating costs (BM2) and Lack of government regulations (BM5) are driving barriers.

Figure 2.

Driver power–dependence diagram.

Discussion

Figures 1 and 2 in the previous section summarize the studies' findings. Figure 1 shows the interactive relations between the barriers, whereas the distribution of these barriers with respect to their driving and dependence power is reflected in Figure 2. From the hierarchy structure shown in Figure 1, "driving barriers” can be seen at the lowest level. Including ‘Inadequate internet connectivity (BM1)’, ‘High adoption and operating costs (BM2)’ and ‘Lack of government regulations (BM5)’. The outcomes agree with the preceding studies that ‘Inadequate internet connectivity (BM1)’ is the main cause of IoT adoption and triggers the other barriers.⁸ The findings are also aligned with Sharma, Joshi⁴⁸ study, which states that ‘Inadequate internet connectivity (BM1)’ and ‘Lack of government regulations (BM5)’ represent the consistent behavior of being greatly influential. Furthermore, ‘High adoption and operating costs (BM2)’ is another important driving barrier. The previous study by Luthra, Garg⁹⁰ is similar to the current result.

‘Lack of standardization’ (BM4) and ‘Complex Architecture’ (BM6) are classified as linkage barriers in Figure 2. From the driving barriers, they receive the influence and, in turn, apply effects on dependent barriers. BM4 and BM6 will be triggered and affected by driving barriers (BM1, BM2 and BM5). The literature also suggests that BM1 (Inadequate internet connectivity) and BM5 (Lack of government regulations) influence BM4 (Lack of standardization) and BM6 (Complex Architecture).^8,48 The findings are also aligned with Luthra, Garg⁴³ study, which states that BM4 (Lack of standardization) and BM6 (Complex Architecture & IT infrastructure) capture third and fourth place, respectively, in IoT adoption challenges.

In turn, BM4 and BM6 will exert influence on the dependent barriers such as ‘Compatibility issues and seamless integration (BM3)’, ‘High energy consumption (BM7)’, ‘Security and privacy issues (BM8)’, ‘Lack of identification and validation (BM9)’ and ‘Long payback period (BM10)’. Such barriers display the attributes of output variables within the entire system. The findings of Kamble, Gunasekaran⁸ also indicate that BM10, BM9 and BM7 are highly energetic and dependent on the other barriers included in the system. These outcomes also align with Kamble, Gunasekaran,⁹¹ who referred to the ‘Lack of identification and validation & lack of a clear comprehension about the IoT benefits (BM9)’ as a dependent barrier. Consistent with earlier studies,⁴⁸ this study also shows that ‘Security and privacy issues (BM8)’ are perennial concerns when discussing IoT adoption and treating it as a barrier. This outcome also concurs with prior empirical studies,⁹² which show that security and privacy issues are dependent barriers. Tackling these barriers will spur and foster the IoT adoption process.

Theoretical and managerial implications

Manufacturing organizations are searching for different sustainability programs to beat the growing competitive pressure, and IoT adoption is an emerging technology initiative. Identifying the most prominent IoT adoption barriers for agile manufacturing in the post-COVID-19 scenarios was the main contribution of the present study because barriers overshadow most firms' opportunities for IoT-related investment. Ordinarily, managing all barriers at a similar time is not feasible for top management because of the absence of adequate assets. In this regard, the current study shows an entire framework of critical barriers they must first work on. It implies that policymakers should pay serious attention to three causal barriers of Inadequate internet connectivity (BM1)’, ‘High adoption and operating costs (BM2)’ and ‘Lack of government regulations (BM5) as they directly or indirectly control every other barrier. And to track down these obstacles, practitioners can extrapolate this model. By proposing implications for policymakers, this work helps manufacturing organizations reduce complexities in IoT adoption, and compliance with this framework will also boost consumer trust. At a massive scale, this study would enhance and foster IoT adoption both in social platforms and academic institutions.

Conclusion

This research aimed to classify the significant barriers associated with adopting IoT technologies in the manufacturing sector in the post-COVID-19 lockdown scenario through an expert survey of industry leaders. Using an extensive literature survey, 16 IoT adoption barriers were identified and validated by academicians and industry experts. To study IoT adoption barriers, 3 data analysis techniques, AHP, TOPSIS and ISM, have been uniquely employed. These techniques highlight the substantial driver barriers that might serve as a major stimulus in IoT adoption. These barriers were categorized into four classes based on their driving and dependence powers, and because of high driving power and less dependence, independent barriers are of crucial significance. Finally, a unique causal model was developed with the support of top-listed IoT adoption barriers to facilitate the root cause analysis.

For practitioners and policymakers, current research is invaluable in understanding the implementation of IoT and the barriers obstructing its application. Future researchers should replicate the study and collect data from other regions besides the manufacturing sector. As each country has different economic, political, social and environmental conditions, it is imperative to mention that different countries might generate different results. Also, there may be some other obstacles that have not been identified and included in this paper. In addition, probing ways to overcome the examined barriers may be the theme of further studies. Furthermore, through Structural Equation Modeling (SEM), the final model should be validated statistically.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Asif Mahmood

References

Narula

Kumar

Prakash

, et al. Modeling and analysis of challenges for industry 4.0 implementation in medical device industry to post COVID-19 scenario, 2021.

Ivanov

. Predicting the impacts of epidemic outbreaks on global supply chains: a simulation-based analysis on the coronavirus outbreak (COVID-19/SARS-CoV-2) case. Transport Res E Logist Transport Rev 2020; 136: 101922.

Zhang

Van Luttervelt

. Toward a resilient manufacturing system. CIRP annals 2011; 60: 469–472.

Zhang

Van Luttervelt

. Toward a resilient manufacturing system. CIRP annals. 2011; 60: 469–472.

Seric

Winkler

. COVID-19 could spur automation and reverse globalisation–to some extent. CEPR Policy Portal [en línea] https://voxeu.org/article/covid-19-could-spur-automation-and-reverseglobalisation-some-extent2020

Javaid

Haleem

Vaish

, et al. Robotics applications in COVID-19: a review. Journal of Industrial Integration and Management 2020; 5: 441–451.

Almeida

Santos

Monteiro

JAJIEMR

. The challenges and opportunities in the digitalization of companies in a post-COVID-19 World. 2020; 48: 97–103.

Kamble

Gunasekaran

Parekh

, et al. Modeling the internet of things adoption barriers in food retail supply chains. J Retailing Consum Serv 2019; 48: 154–168.

Baldini

Botterman

Neisse

, et al. Ethical design in the internet of things. Sci Eng Ethics 2018; 24: 905–925.

10.

Alaba

Othman

Hashem

IAT

, et al. Internet of things security. A survey 2017; 88: 10–28.

11.

Arunachalam

Kumar

Kawalek

. Understanding big data analytics capabilities in supply chain management: unravelling the issues, challenges and implications for practice. Transport Res E Logist Transport Rev 2018; 114: 416–436.

12.

Khan

Anwar

Azam

, et al. A model-driven approach for access control in internet of things (IoT) applications–an introduction to UMLOA. In: International Conference on Information and Software Technologies. Springer, 2018, pp. 198–209.

13.

Trappey

Govindarajan

, et al. A review of essential standards and patent landscapes for the Internet of Things: a key enabler for Industry 4.0. Adv Eng Inf 2017; 33: 208–229.

14.

Wang

Muddada

, et al. On Petri net implementation of proactive resilient holistic supply chain networks. Int J Adv Des Manuf Technol 2013; 69: 427–437.

15.

Rad

Ahmada

. Internet of things: trends, opportunities, and challenges. International Journal of Computer Science and Network Security 2017; 17: 89–95.

16.

Kumar

Singh

Jain

SKJIJQ

, et al. Agile manufacturing: a literature review and Pareto analysis. 2020; 37: 207-222.

17.

Trappey

Govindarajan

, et al. A review of essential standards and patent landscapes for the Internet of Things: a key enabler for Industry 4. Adv Eng Inf. 2017; 33: 208–229.

18.

Arshad Ali

Mahmood

Ikram

, et al. Configuring the drivers and carriers of process innovation in manufacturing organizations. 2020; 6: 154.

19.

Lee

. The internet of things (IoT): applications, investments, and challenges for enterprises. Bus Horiz 2015; 58: 431–440.

20.

Ferretti

Schiavone

. Internet of Things and business processes redesign in seaports. The case of Hamburg. Business Process Management Journal 2016. Online print.

21.

Da Xu

Zhao

. The internet of things: a survey. Inf Syst Front 2015; 17: 243–259.

22.

Khujamatov

Reypnazarov

Khasanov

, et al. Networking and computing in internet of things and cyber-physical systems. In 2020 IEEE 14th International Conference on Application of Information and Communication Technologies (AICT) 2020, 2020, pp. 1–6.

23.

Chui

Gupta

Liu

, et al. A survey of internet of things and cyber-physical systems: standards, algorithms, applications, security, challenges, and future directions. Information 2023; 14: 388.

24.

Haddud

DeSouza

Khare

, et al. Examining potential benefits and challenges associated with the Internet of Things integration in supply chains. J Manuf Technol Manag 2017. Online print.

25.

Kamble

Gunasekaran

Gawankar

. Sustainable Industry 4.0 framework: a systematic literature review identifying the current trends and future perspectives. Process Saf Environ Protect 2018; 117: 408–425.

26.

Mikusz

Clinch

Jones

, et al. Repurposing web analytics to support the IoT. Computer 2015; 48: 42–49.

27.

Perumal

Datta

Bonnet

. IoT device management framework for smart home scenarios. In: 2015 IEEE 4th global Conference on consumer electronics (GCCE), Osaka, 2015, pp.54-55. IEEE.

28.

Jia

Chen

Wang

, et al. ContexloT: towards providing contextual integrity to appified IoT platforms. NDSS 2017; 2: 2.

29.

Narkhede

Raut

Roy

, et al. Implementation barriers to lean-agile manufacturing systems for original equipment manufacturers: an integrated decision-making approach. Int J Adv Des Manuf Technol 2020; 108: 3193–3206.

30.

Houyou

Huth

H-P

Kloukinas

, et al. Agile manufacturing: general challenges and an IoT@ work perspective. In: Proceedings of 2012 IEEE 17th International Conference on Emerging Technologies & Factory Automation. Krakow, 2012, IEEE, pp. 1–7.

31.

Kidd

. Agile manufacturing: a strategy for the 21st century, 1995.

32.

Kumar

Singh

Jain

. Agile manufacturing: a literature review and Pareto analysis. Int J Qual Reliab Manag 2020; 37: 207–222.

33.

Jin‐Hai

Anderson

Harrison

. The evolution of agile manufacturing. Bus Process Manag J 2003; 9: 170–189.

34.

Guerra

Zhang

. Computer applications in agile manufacturing. In Agile manufacturing: the 21st century competitive strategy. Elsevier Science, 2001, pp. 317–337.

35.

Kumar

Singh

Jain

. A combined AHP and TOPSIS approach for prioritizing the attributes for successful implementation of agile manufacturing. Int J Prod Perform Manag 2020. Online print.

36.

Gregory

. The Internet of Things: revolutionizing the retail industry. Accenture Strategy 2015; 1: 1–8.

37.

Da Xu

Wang

. Internet of things for enterprise systems of modern manufacturing. IEEE Trans Ind Inf 2014; 10: 1537–1546.

38.

Ashima

Haleem

Bahl

, et al. Automation and manufacturing of smart materials in Additive Manufacturing technologies using Internet of Things towards the adoption of Industry 4.0. Mater Today Proc 2021. Online print.

39.

Loffler

Munstermann

Schumacher

, et al. Insurers need to plug into the internet of things or risk falling behind. European Insurance, 2016.

40.

Pang

Chen

Han

, et al. Value-centric design of the internet-of-things solution for food supply chain: value creation, sensor portfolio and information fusion. Inf Syst Front 2015; 17: 289–319.

41.

Bandyopadhyay

Sen

. Internet of things: applications and challenges in technology and standardization. Wireless Pers Commun 2011; 58: 49–69.

42.

Riggins

Wamba

. Research directions on the adoption, usage, and impact of the internet of things through the use of big data analytics. In 48th Hawaii international Conference on system Sciences 2015. Kauai, HI, 2015, IEEE, pp. 1531–1540.

43.

Luthra

Garg

Mangla

, et al. Analyzing challenges to Internet of Things (IoT) adoption and diffusion: an Indian context. Procedia Computer Science 2018; 125: 733–739.

44.

Ryan

Watson

. Research challenges for the internet of things: what role can or play? Systems 2017; 5: 24.

45.

Hopalı

Vayvay

. Internet of Things (IoT) and its challenges for usability in developing countries. IJIESA 2018; 2: 6–9.

46.

Borgia

. The Internet of Things vision: key features, applications and open issues. Comput Commun 2014; 54: 1–31.

47.

Yan

Da Xu

. Developing vehicular data cloud services in the IoT environment. IEEE Trans Ind Inf 2014; 10: 1587–1595.

48.

Sharma

Joshi

Kannan

, et al. Internet of Things (IoT) adoption barriers of smart cities’ waste management: an Indian context. J Clean Prod 2020; 270: 122047.

49.

Almeida

Santos

Monteiro

. The challenges and opportunities in the digitalization of companies in a post-COVID-19 World. IEEE Eng Manag Rev 2020; 48: 97–103.

50.

Alaba

Othman

Hashem

IAT

, et al. Internet of Things security: a survey. J Netw Comput Appl 2017; 88: 10–28.

51.

Soldatos

Gusmeroli

Malo

, et al. Internet of things applications in future manufacturing. In Digitising industry-internet of things connecting the physical, digital and virtual worlds, 2016, pp. 153–183.

52.

ConsumerLab

. Wearable technology and the Internet of things: consumer views on wearables beyond health and wellness. Stockholm, Sweden: Ericsson, June, 2016.

53.

Chen

S-L

Chen

Y-Y

Hsu

. A new approach to integrate internet-of-things and software-as-a-service model for logistic systems: a case study. Sensors 2014; 14: 6144–6164.

54.

Saleem

Hammoudeh

Raza

, et al. IoT standardisation: challenges, perspectives and solution. In: Proceedings of the 2nd international conference on future networks and distributed systems, 2018, pp. 1–9.

55.

Rose

Eldridge

Chapin

. The internet of things: an Overview. The Internet Society (ISOC), 2015.

56.

Kumar

Raut

Nayal

, et al. To identify industry 4.0 and circular economy adoption barriers in the agriculture supply chain by using ISM-ANP. J Clean Prod 2021; 293: 126023.

57.

Rana

Shafique

Ali

, et al. Critical success factors in implementing enterprise resource planning (erp) system in PAKISTANI organizations. Governance and Management Review 2021; 4: 151.

58.

Alalawneh

Alkhatib

. The barriers to big data adoption in developing economies. Electron J Inf Syst Dev Ctries 2021; 87: e12151.

59.

Talavera

Tobón

Gómez

, et al. Review of IoT applications in agro-industrial and environmental fields. Comput Electron Agric 2017; 142: 283–297.

60.

Sethi

SR, S

. Internet of things: architectures, protocols, and applications. URL: (2017) https://wwwhindawicom/journals/jece/2017/9324035B93

61.

Shahbazi

Byun

Y-C

. Integration of blockchain, IoT and machine learning for multistage quality control and enhancing security in smart manufacturing. Sensors 2021; 21: 1467.

62.

Padyab

Habibipour

Rizk

, et al. Adoption barriers of IoT in large scale pilots. Information 2020; 11: 23.

63.

Pappas

Caputo

Pellegrini

, et al. The complexity of decision-making processes and IoT adoption in accommodation SMEs. J Bus Res 2021.

64.

Iansiti

Lakhani

. Digital ubiquity: how connections, sensors, and data are revolutionizing business. Harv Bus Rev 2014; 92: 19.

65.

Hallowell

Gambatese

. Qualitative research: application of the Delphi method to CEM research. J Construct Eng Manag 2010; 136: 99–107.

66.

Hyun

Cho

Koo

, et al. Effect of delivery methods on design performance in multifamily housing projects. J Construct Eng Manag 2008; 134: 468–482.

67.

Bolaños

Fontela

Nenclares

, et al. Using interpretive structural modelling in strategic decision-making groups. Manag Decis 2005; 43: 877–895.

68.

Govindan

Azevedo

Carvalho

, et al. Lean, green and resilient practices influence on supply chain performance: interpretive structural modeling approach. Int J Environ Sci Technol 2015; 12: 15–34.

69.

Etemadi

Van Gelder

Strozzi

. An ism modeling of barriers for blockchain/distributed ledger technology adoption in supply chains towards cybersecurity. Sustainability 2021; 13: 4672.

70.

Kumar

Bhamu

Sangwan

. Analysis of barriers to Industry 4.0 adoption in manufacturing organizations: an ISM approach. Procedia CIRP 2021; 98: 85–90.

71.

Deshpande

Nagendra

. An ISM-based framework of variables affecting transactions in the online patent marketplace. Int J Technol Market 2020; 14: 69–91.

72.

Mahmood

Arshad Ali

Nazam

, et al. Developing an interplay among the psychological barriers for the adoption of industry 4.0 phenomenon. PLoS One 2021; 16: e0255115.

73.

Mangla

Govindan

Luthra

. Prioritizing the barriers to achieve sustainable consumption and production trends in supply chains using fuzzy Analytical Hierarchy Process. J Clean Prod 2017; 151: 509–525.

74.

Kusumawardani

Agintiara

. Application of fuzzy AHP-TOPSIS method for decision making in human resource manager selection process. Procedia Computer Science 2015; 72: 638–646.

75.

Arshad

Mahmood

Salam

. Prioritising the practices that influence the operational performance of manufacturing organisations using hybrid ahp-topsis analysis. S Afr J Ind Eng 2020; 31: 65–77.

76.

Saaty

. Decision making with the analytic hierarchy process. Int J Serv Sci 2008; 1: 83–98.

77.

Golden

Wasil

Harker

. The analytic hierarchy process. Berlin, Heidelberg: Applications and Studies, 1989.

78.

Vinodh

Prasanna

Prakash

. Integrated Fuzzy AHP–TOPSIS for selecting the best plastic recycling method: a case study. Appl Math Model 2014; 38: 4662–4672.

79.

Dağdeviren

Yavuz

Kılınç

. Weapon selection using the AHP and TOPSIS methods under fuzzy environment. Expert Syst Appl 2009; 36: 8143–8151.

80.

Martins

Godina

Azevedo

, et al. Towards the development of a model for circularity: the circular car as a case study. Sustain Energy Technol Assessments 2021; 45: 101215.

81.

Warfield

. Binary matrices in system modeling. IEEE Transactions on Systems, Man, and Cybernetics 1973; 1: 441–449.

82.

Alawamleh

Popplewell

. Analysing virtual organisation risk sources: an analytical network process approach. Int J Netw Virtual Organ 2012; 10: 18–39.

83.

Arshad Ali

Mahmood

Ahmad

, et al. Boosting up operational performance of manufacturing organizations through interpretive structural modelling of enabling practices. Math Probl Eng 2020; 1: 2020.

84.

Kumar

Ahmed

Singh

, et al. Determination of hierarchical relationships among sustainable development goals using interpretive structural modeling. Environ Dev Sustain 2018; 20: 2119–2137.

85.

Singh

Kant

. Knowledge management barriers: an interpretive structural modeling approach. Int J Manag Sci Eng Manag 2008; 3: 141–150.

86.

Duperrin

J-C

Godet

. Méthode de hiérarchisation des éléments d'un système: essai de prospective du système de l'énergie nucléaire dans son contexte sociétal, 1973.

87.

Yadav

Shankar

Singh

. Hierarchy of critical success factors (CSF) for lean six sigma (LSS) in quality 4.0. International Journal of Global Business and Competitiveness 2021; 1: 1–14.

88.

Attri

Dev

Sharma

. Interpretive structural modelling (ISM) approach: an overview. Res J Manag Sci 2013; 2319: 1171.

89.

Mandal

Deshmukh

. Vendor selection using interpretive structural modelling (ISM). Int J Oper Prod Manag 1994; 14: 52–59.

90.

Luthra

Garg

Mangla

, et al. Analyzing challenges to internet of things (IoT) adoption and diffusion. An Indian context 2018; 125: 733–739.

91.

Kamble

Gunasekaran

Sharma

. Analysis of the driving and dependence power of barriers to adopt industry 4.0 in Indian manufacturing industry. Comput Ind 2018; 101: 107–119.

92.

Ajmera

Jain

. Modelling the barriers of Health 4.0–the fourth healthcare industrial revolution in India by TISM. Operations Management Research 2019; 12: 129–145.