PaperlayerView exact revision on arXiv

Source-linked AI summary

The Internet of Things for Smart Manufacturing: A Review

Hui Yang, Soundar Kumara, Satish Bukkapatnam, Fugee Tsung

arXiv:2607.16172v1eess.SYstat.AP

TL;DR

Smart manufacturing requires IoT-based methods that can manage sensing data and integrate physical manufacturing enterprises with cyberspace. This paper reviews enabling technologies and applications, presents an IoMT–cloud framework for virtual machine networks, and surveys cybersecurity, policies, challenges, and opportunities.

  • Problem

    Large-scale IoT sensing creates big-data management, information-processing, and manufacturing-control needs, while machine-signature data remain underused for diagnostics, prognostics, and optimization.

  • Method

    The paper reviews IoT technologies and manufacturing applications, studies IoMT and cloud computing for virtual machine networks, and examines cybersecurity, policies, challenges, and opportunities.

  • Results

    The review identifies IoT and cloud-based integration, analytics, and platforms as foundations for data-driven manufacturing, with IoT applications projected to generate $1.2 to $3.7 trillion annually by 2025.

  • Takeaways & Limitations

    IoMT provides a basis for cyber-physical integration and improved manufacturing decision-making, while cybersecurity and policy remain important for smart-factory development.

  • Takeaways & Limitations

    IoT remains under development and faces technical cyber-physical integration issues involving communication, big data, control, and cloud bandwidth.

Abstract

from arXiv · show

The modern manufacturing industry is investing in new technologies such as the Internet of Things (IoT), big data analytics, cloud computing and cybersecurity to cope with system complexity, increase information visibility, improve production performance, and gain competitive advantages in the global market. These advances are rapidly enabling a new generation of smart manufacturing, i.e., a cyber-physical system tightly integrating manufacturing enterprises in the physical world with virtual enterprises in cyberspace. To a great extent, realizing the full potential of cyber-physical systems depends on the development of new methodologies on the Internet of Manufacturing Things (IoMT) for data-enabled engineering innovations. This paper presents a review of the IoT technologies and systems that are the drivers and foundations of data-driven innovations in smart manufacturing. We discuss the evolution of internet from computer networks to human networks to the latest era of smart and connected networks of manufacturing things (e.g., materials, sensors, equipment, people, products, and supply chain). In addition, we present a new framework that leverages IoMT and cloud computing to develop a virtual machine network. We further extend our review to IoMT cybersecurity issues that are of paramount importance to businesses and operations, as well as IoT and smart manufacturing policies that are laid out by governments around the world for the future of smart factory. Finally, we present the challenges and opportunities arising from IoMT. We hope this work will help catalyze more in-depth investigations and multi-disciplinary research efforts to advance IoMT technologies.

I. Introduction

Smart manufacturing is moving from shop-floor automation toward cyber-physical, data-driven enterprises that connect manufacturing things and cyberspace. The paper reviews this evolution, proposes an IoMT–cloud framework, and surveys related policies, cybersecurity issues, challenges, and opportunities.

  • IoT connects manufacturing materials, sensors, actuators, controllers, robots, operators, machines, products, and handling equipment into an integrated enterprise network.
  • Large-scale IoT sensing produces cloud-stored big data, creating a need for new methods in data management, information processing, and manufacturing process control.
  • Smart manufacturing depends on data-driven innovations that support greater autonomy and optimization beyond shop-floor automation.
  • The paper presents an IoMT and cloud-computing framework for developing virtual machine networks.
  • The review covers IoT technologies, cybersecurity issues, government manufacturing policies, and IoMT challenges and opportunities.

B. IoT Sensing

IoT sensing and communication technologies provide the infrastructure for connecting heterogeneous manufacturing things. The section covers sensing components, protocols, manufacturing standards, and architectural frameworks that support data exchange and integration.

  • IoT Sensing: IoT things have unique identities and can sense, collect, or exchange environmental and operational data across interconnected networks.
  • IoT Sensing: RFID, wireless sensor networks, and mobile computing contribute complementary capabilities for identifying objects, sensing system dynamics, and connecting devices.
  • IoT Data Protocols and Architectures: IoT systems integrate heterogeneous things with different communication, processing, storage, and power-supply characteristics through diverse data-link protocols.
  • IoT Data Protocols and Architectures: IPv6 provides approximately 2^128, or 3.4×10^38, addresses to support unique addressing as the number of connected IoT things grows.
  • IoT Data Protocols and Architectures: 6LowPAN enables IPv6 packets to traverse heterogeneous platforms including WiFi, ZigBee, and 802.15.4.
  • IoT Data Protocols and Architectures: MTConnect standardizes manufacturing data vocabulary, semantics, and communication between IoT-enabled machines and user applications using Internet technologies.
  • IoT Data Protocols and Architectures: MTConnect uses adapters, agents, and applications to translate raw data into compliant information for processing, storage, analytics, and visualization.
  • IoT Data Protocols and Architectures: Industry 4.0 architectures include RAMI 4.0 and OPC Unified Architecture, while competition makes a single reference architecture difficult to establish.

D. IoT Platforms

IoT platforms provide software infrastructure for integrating physical things with cyber-world applications. The section highlights cloud computing, virtual and augmented reality, and big-data analytics as key platform capabilities and technologies.

  • D. IoT Platforms: IoT platforms enable physical things and cyber-world applications to communicate and integrate through cloud, embedded, data-management, machine-learning, and analytics mechanisms.
  • D. IoT Platforms: Cloud computing supplies Internet-based storage, data management, KPI computation, visualization, and analytics through IaaS, PaaS, and SaaS services.
  • D. IoT Platforms: VR and AR integration supports asset utilization, workforce training, root-cause diagnosis, digital design, and predictive maintenance.
  • D. IoT Platforms: IoT sensing generates big data characterized by high volume, velocity, veracity, and variety, including diverse and uncertain manufacturing signals.
  • D. IoT Platforms: Big-data analytics provide methods for processing large-scale IoT data and supporting manufacturing process control.

III. Sensor Networks, Manufacturing Services, and Applications

Manufacturing execution systems connect shop-floor control systems with management applications to support transparent data sharing, analytics, and digital performance management. IoT and cloud platforms extend this integration across ERP, MES, and PCS through real-time bidirectional data flows.

  • Manufacturing Execution Systems: MES establishes transparent data sharing and information exchange between machines, controllers, and manufacturing management departments.Gateway computers transmit real-time control-system data to database servers for management-level monitoring and analytics.
  • Manufacturing Execution Systems: IoT-enabled control systems and MTConnect are moving MES platforms to the cloud, simplifying data communication, storage, analytics, and reporting.Cloud-based MES addresses proprietary definitions that make real-time data streams difficult to decode.
  • ERP–MES–PCS Integration: ERP, MES, and PCS exchange data bidirectionally, with ERP sending planning information downward and PCS returning operational feedback upward.The flow follows ISA 95 activity models while emphasizing data movement across enterprise, execution, and process-control levels.
  • ERP–MES–PCS Integration: Real-time feedback on asset utilization, quality, labor, and process performance improves the accuracy and reliability of cost analysis, work-in-process predictions, and inventory control.This feedback enables ERP-side purchasing and bill-of-materials adjustments.

B. Sensor-based Manufacturing Informatics and Control

Sensor-based manufacturing informatics transforms large, heterogeneous data streams into features, patterns, and decisions for monitoring, diagnosis, optimization, and diverse IoT manufacturing applications. The review spans cloud manufacturing, cyber-physical systems, energy, operations, safety, and logistics while noting that industrial evidence remains limited.

  • Sensor-based Informatics: Advanced sensing populates ERP, MES, and PCS with data that remain underused for real-time monitoring, fault diagnosis, and performance optimization.New methods are needed to extract useful features and patterns and exploit the resulting knowledge.
  • Sensor-based Informatics: Data representation, pattern recognition, feature extraction, and simulation support the transformation of sensor data into information for manufacturing control and decision making.Methods discussed include Fourier and wavelet representations, PCA, clustering, Bayesian networks, and simulation modeling.
  • IoT Manufacturing Applications: IoT manufacturing applications include cloud manufacturing, cyber-physical manufacturing, energy-efficiency management, operations management, safety and ergonomics, and supply-chain logistics.The reviewed applications connect sensing and computing infrastructures with manufacturing services, control, and resource coordination.
  • IoT Manufacturing Applications: IoT-based systems support energy monitoring and optimization, predictive maintenance, dynamic production scheduling, accountability, and collaborative warehouse operations.Examples include selective-laser-sintering energy control, coal-mine equipment monitoring, and RFID-enabled decentralized warehouse management.
  • Research Needs: Most industrial IoT case studies are presented for marketing purposes, leaving urgent research needs in system optimization, data modeling, and cybersecurity.The review therefore treats current industrial evidence as an incomplete basis for assessing IoT manufacturing systems.

INNOVYT

The section presents company examples of cloud-based infrastructure and life-science supply-chain support. These examples illustrate commercial uses of cloud and IoT technologies rather than a comparative evaluation.

  • Commercial Cloud Examples: LightInTheBox uses Amazon AWS to build a highly available customer website and reduce operating expenses.The cloud infrastructure supports transactions across locations and allows computing resources to be adjusted as needed.
  • Commercial Cloud Examples: TraceLink developed the Life Science Cloud platform to support compliance across a global life-science network and pharmaceutical supply chain.AWS supports the requirements of hundreds of pharmaceutical companies and their partners.

IV. Case Study - IoT and Cloud Computing to Build Cyber-physical

The case study frames IoMT as a network of connected manufacturing objects whose data require specialized management and processing. IoMT and cloud computing support virtual machine networks, cyber-space analytics, and feedback to physical manufacturing systems.

  • IoMT Framework: IoMT integrates sensors, computing units, physical objects, and services so manufacturing things can communicate and exchange data.This network forms a backbone for smart manufacturing and provides access to large volumes of manufacturing data.
  • IoMT Framework: Existing methodologies are insufficient for internet-like IoMT structures and the high-volume data gathered throughout manufacturing enterprises.The paper identifies a need for new data-driven methods and tools tailored to IoMT.
  • Data Management: IoMT data management must address high-volume, high-velocity manufacturing data through access, structure, compression, synthesis, traceability, and retrieval techniques.Manufacturing data differ substantially from data in domains such as computer science, environmental science, and healthcare.
  • Information Processing: IoMT information processing uses data representation, feature extraction, and visualization to reveal hidden information and characterize manufacturing-system patterns.Representation can use frequency, wavelet, or state-space domains, while feature extraction seeks parsimonious state-sensitive features.
  • Decision Making: Cyber-physical manufacturing reflects the physical enterprise in cyberspace through data-driven processing, modeling, and simulation, then feeds analytical actions or control schemes back to physical systems.The supported decisions include machine monitoring, fault diagnosis, predictive maintenance, inventory optimization, supply-chain management, and safety.
  • Case Study: The case study uses machine signatures and cloud computing to model large-scale IoMT machine networks and support condition monitoring.It focuses on dissimilarities between high-dimensional signatures across many machines to construct a virtual machine network.

IV.A. Physical Machine Networks – Process Monitoring and Control

The paper develops IoMT physical machine networks that compare machine profiles, organize machines and parts into communities, and support monitoring, classification, planning, and optimization. It combines customized P2P and population M2M models with parallel graph analytics for large-scale manufacturing data.

  • Network construction: The approach uses pairwise profile dissimilarities rather than a predefined normal signature to cluster large numbers of machine profiles into homogeneous groups.This supports richer data representation and visual analysis of machine conditions.
  • Network construction: Virtual machine-to-machine networks exchange real-time attributes so machines with similar signatures, profiles, or events form communities.The study focuses on enabling communication among networked machines through shared attributes.
  • Manufacturing applications: IoMT sensing can generate tens of thousands of power profiles, enabling proactive process adjustment and machine maintenance while creating major network-construction challenges.The stated goals include improving product quality and reducing re-work rates.
  • Network models: Customized P2P networks monitor repeated production profiles, while population M2M networks represent machines using dominant patterns or aggregated properties.The choice of M2M node attributes depends on the application domain.
  • Scalable analytics: Parallel graph algorithms are proposed because large IoMT populations and data volumes make serial network modeling and real-time analytics computationally expensive.The paper presents pattern matching, network modeling, predictive analytics, and parallel computing as initial technical steps.

IV.C.1 Pattern Matching

Pattern matching measures profile dissimilarity by accounting for changes in signal morphology and timing. The reviewed approach uses dynamic time warping to align profiles before constructing pairwise similarity and dissimilarity information.

  • Profile variation: Manufacturing profile waveforms vary across operation stages and between nominally identical parts because of machine, tool, material, and uncertainty factors.Power consumption is used as an illustration of these variations.
  • Similarity measures: Correlation captures only linear dependence, whereas mutual information characterizes linear and nonlinear relationships but requires stationarity.The paper situates mutual information among methods for estimating profile interrelationships.
  • Dynamic time warping: Dynamic time warping aligns one-dimensional and multidimensional signatures before measuring morphological dissimilarity.Without alignment, direct profile differences may contaminate useful information and produce meaningless results.
  • Dynamic time warping: The dynamic-programming search produces an optimal warping path and normalized dissimilarity, and repeated pairwise matching yields a warping matrix.The resulting matrix records profile-level similarity and dissimilarity for P2P or M2M analysis.

IV.C.2 Network Modeling

Network modeling converts profile dissimilarities into a machine-network representation whose node distances preserve pairwise relationships. The resulting structure supports clustering, predictive analytics, and scalable optimization, while leaving broader topology-based exploitation as an open challenge.

  • Network representation: The paper identifies network theory as a way to exploit dissimilarity matrices that are difficult to use directly as predictive-model inputs.Machines become nodes and profile similarities or dissimilarities become links.
  • Network representation: Network nodes are positioned in high-dimensional space so node-to-node distances preserve the dissimilarity between corresponding profiles.The model is illustrated with six machine profiles.
  • Network representation: Representing profiles as network nodes reduces data dimensionality and identifies compact features of machine condition.Euclidean or other application-dependent distance measures may be used.
  • Scalable optimization: Cloud computing and parallel algorithms are proposed to distribute large-scale optimization because serial computation can require prohibitive time.Stochastic gradient methods are described as suitable for processing limited subsets of large IoMT datasets.
  • Predictive analytics: Once constructed, virtual machine networks encode changing machine conditions as network dynamics and support predictive analytics for anticipatory manufacturing.Optimized node attributes can serve as features for condition monitoring and predictive modeling.
  • Open challenges: Network topology and community structure can group machines with similar conditions, but additional work is needed to uncover actionable patterns in attributes, links, communities, and topology.The paper frames facility design, job reassignment, and preventive maintenance as questions for future analysis.

V. IoT and Cybersecurity in Manufacturing

The paper treats cybersecurity as a central requirement for IoMT because connected manufacturing devices, cloud databases, and information networks expand exposure to malicious attacks. It reviews the NIST manufacturing framework and security approaches including cryptography, intrusion identification, and blockchain.

  • Cybersecurity motivation: IoT interconnection makes manufacturing systems vulnerable to cyber-attacks, making cybersecurity a primary concern for smart-manufacturing adoption.Manufacturing equipment is part of critical infrastructure and may attract malicious attackers.
  • Cybersecurity framework: The NIST cybersecurity framework for manufacturing comprises Identify, Protect, Detect, Respond, and Recover components.These components address assets, safeguards, incident detection, impact containment, and capability restoration.
  • Related approaches: Reviewed manufacturing-security work includes vulnerability identification across enterprise and supply-chain data flows and mapping cyber-physical vulnerabilities to their effects.These examples extend cybersecurity analysis beyond individual devices.
  • Security techniques: The paper identifies cryptographic solutions, intrusion identification, and blockchain technology as techniques for protecting IoMT security and privacy.Examples include access control, key management, intrusion detection, decentralized device ownership, and peer-to-peer interaction.

VI. IoT Manufacturing Policies and Strategies

Governments and industrial organizations have established policies and strategies to promote IoT adoption, digitalize manufacturing, and strengthen economic competitiveness. These initiatives address technological, workforce, environmental, and market challenges across regions.

  • Policies and strategies are described as key drivers of IoT development and practical implementation for the transition toward smart manufacturing.
  • China: China’s Made in China 2025 strategy promotes data-driven innovation and smart technologies to support sustainable growth and upgrade national manufacturing capabilities.
  • European Union: The EU’s manufacturing agenda emphasizes IoT, big data, AI, additive manufacturing, robotics, and blockchain to create clean, high-performing, and socially sustainable factories.
  • European Union: EU initiatives target a smooth transition to a smart economy, next-generation products and services, stronger manufacturing innovation, and higher EU GDP.
  • United Kingdom: The UK’s Future of Manufacturing plan addresses personalized products, skilled labor shortages, sustainable production, and digitalization through IoT, analytics, intelligent systems, robotics, and related technologies.
  • Industrial organizations: The Industrial Internet Consortium had more than 258 academic and industrial members, over 20 testbeds, and an estimated $1.2 to $3.7 trillion annual manufacturing impact by 2025.

VII. IoT Challenges and Opportunities in Manufacturing

The paper identifies challenges in sensing machine status, dynamically distributing tasks, and coordinating machines, while outlining opportunities in retrofitting, self-powered sensing, cloud analytics, and blockchain-enabled IoMT. These opportunities aim to improve connectivity, decision-making, and management across smart manufacturing systems.

  • Challenges: IoMT challenges include determining machine status, dynamically distributing tasks, and overcoming unreliable communication and limited external control of manufacturing assets.
  • Opportunity 1. Retrofit legacy machines for smart manufacturing: Legacy machines remain widely used but often lack real-time and in-process sensing and control, limiting information visibility and competitiveness for small manufacturers.
  • Opportunity 2. Self-powered machine status sensing: Self-powered sensors could eliminate wireline connections and batteries while improving portability, reducing maintenance costs, and supporting local signal processing.
  • Opportunity 3. Machine service and tasks scheduling and distributing: Sensed machine data can support dynamic task scheduling by accounting for machine availability, malfunctions, energy consumption, utilization, and changing system-level tasks.
  • Opportunity 5. Cloud computing and analytics: Cloud platforms can combine historical and in-situ machine data for analytics and algorithms that support manufacturing quality, power, safety, sustainability, distribution, and supply-chain management.
  • Opportunity 6. Blockchain enabled IoT: Blockchain is presented as a possible decentralized framework for secure IoMT data sharing because it uses peer-to-peer validation and cryptography, although further fundamental research is needed.

VIII. Conclusions

The conclusions frame IoT-enabled smart manufacturing as a data-driven cyber-physical integration challenge requiring standards, platforms, analytics, and secure control. The paper reviews this landscape and identifies unresolved technical issues and research opportunities for IoMT.

  • VIII. Conclusions: Existing sensors, IT systems, and OT systems are not yet closely integrated at the IoT level despite widespread sensor deployment.Industry is establishing standards and platforms to support their integration into new IoT frameworks.
  • VIII. Conclusions: IoT architectures such as RAMI 4.0 and OPC UA provide communication structures for Industry 4.0 integration.RAMI 4.0 defines three critical dimensions, while diverse architectures and platforms accelerate IoT-system development.
  • VIII. Conclusions: IoT-based cyber-physical manufacturing still faces communication, big-data, and control challenges.High-velocity sensor streams can exceed cloud transmission bandwidth and update-frequency limits, while manufacturing data vary in volume, variety, and veracity.
  • VIII. Conclusions: The paper reviews manufacturing IoT technologies and applications and preliminarily leverages IoMT with cloud computing to build virtual machine networks.The stated goal is to improve manufacturing decision-making through cyber-physical integration.
  • VIII. Conclusions: IoMT and smart manufacturing are presented as a promising research paradigm requiring deeper and more comprehensive multidisciplinary investigation.The paper specifically calls for novel IoMT technologies and analytical methodologies to improve manufacturing services and optimize systems.

Author Information

The authors are professors and research leaders in industrial and manufacturing engineering, systems engineering, and industrial engineering and decision analytics. Their research spans sensor-based modeling, manufacturing monitoring and prognostics, nonlinear dynamics, quality, and data analytics.

  • Author Information: Hui Yang is a Penn State associate professor whose research focuses on sensor-based modeling and analysis of complex systems.His applications include process monitoring, control, diagnostics, prognostics, quality improvement, and performance optimization.
  • Author Information: Soundar Kumara is a Penn State professor with an affiliate appointment in information sciences and technology.He is identified as a Fellow of IIE, CIRP, ASME, and AAAS.
  • Author Information: Satish T. S. Bukkapatnam is a Rockwell International Professor at Texas A&M University and directs its Institute for Manufacturing Systems.His research uses high-resolution nonlinear dynamic information, especially from wireless MEMS sensors, for monitoring and prognostics.
  • Author Information: Fugee Tsung is a HKUST professor, director of the Quality and Data Analytics Lab, and editor-in-chief of the Journal of Quality Technology.He is identified as a Fellow of IISE, ASQ, and ASA, and as an Academician of the International Academy for Quality.
Loading 2607.16172v1…