IoT Connectivity and Industrial Communication

Introduction

An Internet of Things (IoT) system is only as useful as its communication architecture. Sensors must report measurements, actuators must receive commands, controllers must exchange status information, and enterprise systems must obtain trustworthy data. In mechatronics, communication is especially demanding because mechanical motion, electrical control, computation, and networking interact in the same system. A communication link may therefore be judged not only by data rate, but also by timing, electrical robustness, safety, energy use, maintainability, and cost.

Connectivity technologies can be divided into wired and wireless systems. Wired technologies generally provide predictable timing, high immunity to radio interference, and continuous power, but installation may be expensive or physically restrictive. Wireless technologies simplify deployment and support mobile equipment, but they introduce propagation uncertainty, spectrum sharing, battery constraints, and cybersecurity concerns.

The central engineering task is not to identify the “best” technology. It is to match communication characteristics to application requirements. A motion-control loop, a condition-monitoring sensor, a warehouse tag, and a remote utility meter require different solutions.

Learning Objectives

After studying this chapter, the reader should be able to:

1. Explain the roles of physical, data-link, network, transport, and application layers in industrial IoT.
2. Compare Ethernet, CAN, Modbus, industrial Ethernet, Wi-Fi, Bluetooth LE, Zigbee/Thread, RFID/NFC, LPWAN, cellular IoT, and 5G.
3. Evaluate technologies using range, latency, reliability, bandwidth, power consumption, cost, and determinism.
4. Select a suitable connectivity architecture for a mechatronic application.
5. Identify limitations, interoperability issues, and cybersecurity responsibilities.

1. Communication Requirements in Mechatronics

A communication requirement should be expressed quantitatively where possible. Important attributes include:

• Range: the physical distance over which communication is dependable.
• Bandwidth: the maximum information-carrying capacity, commonly measured in bit/s.
• Latency: the time between transmitting information and receiving or applying it.
• Jitter: variation in latency from one message to another.
• Reliability: the probability that information is delivered correctly and within the required time.
• Determinism: the degree to which transmission timing is bounded and predictable.
• Energy consumption: particularly important for battery-powered nodes.
• Topology: the arrangement of nodes, such as bus, star, ring, mesh, or point-to-point.
• Scalability: the ability to add devices without unacceptable performance degradation.
• Interoperability: the ability of equipment from different suppliers to communicate correctly.

For a periodic control loop, the communication deadline must be compatible with the control period. If a loop has sampling period T_s, then sensing, computation, communication, and actuation must all be completed within the available cycle time. A useful engineering condition is:

Here, all quantities are times in seconds. The inequality is a design constraint rather than a universal control law. It assumes a single cycle and does not by itself guarantee stability; the control algorithm, plant dynamics, sampling quality, and jitter must also be considered.

For a message of L bits sent at data rate R bit/s, the ideal serialization time is:

This excludes propagation delay, arbitration, protocol headers, waiting time, retransmissions, processing, and queuing. Consequently, it is a lower bound, not a complete latency model.

A practical link budget must also consider protocol overhead. If a payload contains P bits and the complete transmitted frame contains F bits, protocol efficiency is:

where η is dimensionless. A high nominal data rate does not necessarily provide high application throughput if frames are small, acknowledgements are frequent, or retransmissions occur.

Determinism and traffic classes

A network is deterministic when its timing can be bounded sufficiently for the application. Conventional Ethernet uses contention and queuing mechanisms that may produce variable delay. Industrial Ethernet systems improve predictability using managed switches, traffic scheduling, synchronization, prioritization, redundancy, or specialized protocols. Some systems support hard real-time operation, in which missing a deadline is unacceptable; others require only soft real-time behavior, in which occasional delay reduces performance but does not cause immediate failure.

Safety-critical communication adds a separate requirement. A communication protocol may be reliable yet not safety-certified. Functional safety requires systematic risk reduction, diagnostics, fault reaction, and appropriate standards-based development.

2. Layered Communication Architecture

The layered model separates responsibilities:

• Physical layer: electrical or radio signals, connectors, modulation, cabling, and transmission speed.
• Data-link layer: framing, addressing, error detection, medium access, and sometimes retransmission.
• Network layer: routing between different networks.
• Transport layer: segmentation, flow control, and end-to-end delivery where required.
• Application layer: meaning of data, commands, device profiles, and object models.

For example, Ethernet defines important physical and data-link behavior, while TCP/IP provides network and transport functions. Modbus defines an application protocol that can operate over serial links or TCP. CAN defines a robust message-oriented bus, while CANopen adds an application and device-profile layer.

Interoperability therefore has several meanings. Two devices may share the same cable standard but use incompatible application data. Conversely, a gateway can connect physically different networks while translating data models and timing behavior. Gateways are powerful, but they add latency, configuration effort, and possible failure points.

3. Wired Industrial Connectivity

3.1 Ethernet

Ethernet is a dominant wired technology because it is inexpensive, widely supported, and scalable. Modern Ethernet commonly uses twisted-pair or fiber-optic media and supports star topologies through switches. It provides high bandwidth and convenient integration with IP networks.

Standard switched Ethernet is not inherently deterministic. Frames may wait in switch queues, and traffic bursts can produce variable delay. It is therefore suitable for supervisory control, machine data collection, cameras, engineering workstations, and many industrial applications, but unmodified Ethernet should be assessed carefully for tight motion-control loops.

Fiber is valuable where long distance, electrical isolation, high electromagnetic immunity, or high bandwidth is required. Copper twisted pair is usually cheaper and can support Power over Ethernet in suitable applications, although power availability and thermal limits must be checked.

3.2 CAN

The Controller Area Network (CAN) is a multi-master, message-oriented bus widely used in vehicles, mobile machinery, robots, and embedded mechatronics. Nodes compete for access using nondestructive bitwise arbitration: the message with the highest priority wins without corrupting the winning frame. This gives high-priority messages short and bounded access delay under a properly engineered load.

CAN includes error detection mechanisms such as cyclic redundancy checking, bit monitoring, acknowledgement, and fault confinement. Faulty nodes can move to a bus-off state. CAN uses short frames and is robust, but its bandwidth and payload size are limited compared with Ethernet. Bus length depends on bit rate, cable characteristics, termination, and transceiver specifications. A typical CAN bus requires termination at both physical ends, not at every node.

CAN is strong for distributed sensors, actuators, motor controllers, and embedded control modules. It is less suitable for video, large files, or high-volume historical data. CAN FD extends payload capacity and may use a faster data phase, but compatibility with classic CAN equipment must be verified.

3.3 Modbus

Modbus is an application-layer protocol rather than a single physical medium. Modbus RTU commonly operates over RS-485, while Modbus TCP operates over Ethernet. It uses a simple client/server or request/response model and is popular because it is easy to implement and supported by many instruments, programmable logic controllers, and drives.

Modbus RTU is economical and adequate for low-to-moderate-speed register access. However, polling introduces delay, and the protocol has limited built-in semantics for advanced device discovery, time synchronization, or sophisticated diagnostics. Modbus TCP benefits from Ethernet bandwidth but inherits the timing variability of the underlying network and TCP processing. Modbus should not be assumed to provide hard real-time determinism merely because it uses industrial equipment.

3.4 Industrial Ethernet

Industrial Ethernet refers to Ethernet-based systems adapted for automation. Examples include EtherCAT, PROFINET, EtherNet/IP, Sercos, and Ethernet POWERLINK. They differ in frame processing, synchronization, topology, device profiles, and real-time mechanisms.

Common capabilities include:

• prioritized or scheduled traffic;
• hardware timestamping and clock synchronization;
• ring or dual-path redundancy;
• distributed I/O and motion profiles;
• diagnostics and device configuration;
• integration with ordinary IP traffic through appropriate segmentation.

EtherCAT, for example, processes frames as they pass through devices, supporting efficient cyclic exchange. Other systems use time-aware scheduling, producer-consumer models, or specialized synchronization. The exact performance depends on the protocol, controller, switch architecture, node count, cycle time, payload, and engineering configuration.

Industrial Ethernet can provide excellent performance, but it is not automatically interchangeable across vendors. A project must specify the protocol profile, conformance requirements, engineering tools, cabling, topology, and diagnostic procedures.

4. Wireless and Short-Range Connectivity

4.1 Wi-Fi

Wi-Fi provides relatively high bandwidth and is widely available in industrial computers, tablets, cameras, and mobile robots. It is suitable for operator interfaces, asset data, maintenance access, noncritical sensing, and mobile equipment. Wi-Fi operates in shared radio spectrum, so performance depends on channel planning, interference, obstacles, antenna placement, roaming, and access-point density.

Wi-Fi latency and packet loss can vary. Quality-of-service mechanisms can prioritize traffic, but they do not turn a shared wireless network into a guaranteed hard-real-time channel. Industrial deployment should use managed infrastructure, separate operational traffic from guest traffic, secure authentication, and a radio survey.

4.2 Bluetooth Low Energy

Bluetooth Low Energy (BLE) is designed for low-power short-range communication. It is useful for wearable devices, tool identification, commissioning, proximity sensing, beacons, and battery-powered sensors. BLE supports advertising and connection-oriented communication; its practical range varies greatly with transmit power, antenna design, data rate, obstacles, and regulatory limits.

BLE generally offers lower energy consumption than classic Bluetooth for bursty data. It is not normally chosen for deterministic closed-loop control. Security depends on pairing, authentication, key management, and appropriate device configuration; “short range” does not mean “secure by default.”

4.3 Zigbee and Thread

Zigbee is a low-power mesh technology commonly used for distributed sensing and building automation. Thread is an IPv6-based low-power mesh networking technology built on IEEE 802.15.4 radios. Mesh routing can extend coverage and improve resilience, but it adds routing overhead, configuration complexity, and variable path delay.

These technologies are appropriate for low-data-rate sensors, lighting, environmental monitoring, and building or plant instrumentation. They are generally not the first choice for high-speed motion control. Network planning must consider node density, battery replacement, interference, routing behavior, and the effect of sleeping devices on availability.

4.4 RFID and NFC

Radio-frequency identification (RFID) uses tags and readers to identify or track objects. Passive tags obtain energy from the reader field, allowing very low-cost labels but usually at short or moderate range and with limited data capacity. Active tags contain a battery and can provide longer range or sensing functions.

Near-field communication (NFC) is a short-range technology commonly used for deliberate tap-based interactions, commissioning, access control, and device pairing. RFID and NFC are identification technologies rather than general-purpose continuous sensor networks. Their value lies in associating a physical object with a digital record, such as a maintenance history or calibration certificate.

5. Wide-Area and Cellular IoT

5.1 LPWAN

Low-power wide-area networks (LPWANs) support small, infrequent messages over long distances with low energy consumption. Common approaches include LoRaWAN and licensed-spectrum narrowband systems. LPWAN is well suited to remote meters, environmental sensors, agriculture, and infrastructure monitoring.

The trade-off is low throughput, limited payload frequency, and comparatively high or variable latency. Battery life can be excellent when the device sleeps most of the time, but actual life depends on transmission power, message interval, acknowledgement policy, temperature, battery capacity, and network conditions. LPWAN is normally unsuitable for continuous motion control or high-rate waveform transmission.

5.2 Cellular IoT

Cellular IoT includes technologies such as LTE-M and NB-IoT, which use operator-managed licensed networks. They provide wide geographic coverage, SIM or eSIM-based identity, and carrier-grade infrastructure. LTE-M generally supports greater mobility and throughput than NB-IoT, whereas NB-IoT emphasizes low data rates and deep coverage.

Cellular solutions reduce the need to build private wide-area infrastructure, but recurring subscription costs, coverage dependence, power consumption, and operator policies must be considered. They are useful for fleet tracking, remote monitoring, distributed infrastructure, and equipment outside a plant’s premises.

5.3 5G

5G can support enhanced mobile broadband, massive machine-type communication, and ultra-reliable low-latency communication, although the exact service depends on spectrum, deployment mode, network configuration, and commercial offering. Private 5G networks can provide controlled coverage, device management, and traffic isolation within industrial sites.

5G may benefit mobile robots, augmented-reality maintenance, machine vision backhaul, and flexible factory layouts. However, advertised peak rates are not equivalent to guaranteed application performance. A design must verify latency distribution, handover behavior, coverage, synchronization, availability, and local spectrum regulation.

6. Comparative Engineering Evaluation

The following table gives qualitative tendencies. Actual performance must be verified using the selected equipment and environment.

A useful selection procedure is:

1. Define the application’s payload, message frequency, deadline, availability, and safety class.
2. Decide whether the device is fixed, mobile, battery-powered, or energy-harvesting.
3. Establish environmental constraints: temperature, vibration, electromagnetic interference, hazardous location, and physical access.
4. Select candidate media and topologies.
5. Test worst-case latency, packet loss, recovery time, and coexistence—not merely average throughput.
6. Define cybersecurity controls, maintenance procedures, and obsolescence plans.

Figure: A requirements-driven process for selecting IoT connectivity.

7. Worked Example: Estimating Communication Load

Consider a machine with 24 sensors. Each sensor sends a 16-byte payload every 20 ms. The payload generation rate is:

The result assumes all sensors transmit exactly once per period and excludes headers, acknowledgements, inter-frame gaps, arbitration, retransmissions, and management traffic. If protocol overhead increases the total transmitted bits by 30%, the estimated offered load becomes:

A nominal 250 kbit/s link would appear adequate by average load alone, but it leaves little margin for bursts and retransmissions. A 1 Mbit/s link provides more capacity, yet the engineer must still verify worst-case access delay and whether all messages meet their deadlines. This example illustrates why average bandwidth is not a substitute for real-time analysis.

8. Applied Case: Retrofitting a Packaging Machine

Background

An illustrative packaging line contains servo drives, pneumatic valves, photoelectric sensors, a safety controller, and a supervisory dashboard. The original machine used point-to-point wiring and a small serial network. Management wants condition monitoring and simpler maintenance without compromising production timing.

Problem

The servo loop requires predictable cyclic updates. Vibration sensors generate larger data records but do not require sub-millisecond control. Maintenance technicians need wireless access, and the dashboard must send production data to a plant information system.

Analysis

A single wireless network would create unnecessary risk for the servo loop because radio interference and roaming can produce variable delay. Point-to-point wiring would preserve timing but increase cable count and make expansion difficult. The vibration data is unsuitable for a low-bandwidth fieldbus if raw waveforms are transmitted continuously.

Approach and solution

The retrofit uses industrial Ethernet with a real-time automation protocol for drives and distributed I/O. Safety communication is implemented using the safety mechanisms required by the selected safety system rather than relying on ordinary status messages. A separate VLAN and firewall zone carry supervisory and diagnostic traffic. Vibration nodes perform local feature extraction and send summary indicators over a suitable wired or wireless link. Wi-Fi provides authenticated maintenance access through a segregated network, while RFID identifies tooling and associates it with configuration records.

Traffic is prioritized and monitored. Switches, drives, and controllers are selected from compatible conformance classes. The commissioning test measures cycle-time jitter, packet loss, recovery after cable disconnection, wireless coverage, and the effect of diagnostic traffic.

Outcome

The architecture preserves deterministic control where it matters, reduces wiring for selected monitoring functions, and gives technicians better asset information. It also makes the distinction between control traffic, safety traffic, maintenance traffic, and enterprise data explicit.

Learning points

1. Different data classes should not automatically share identical connectivity.
2. A gateway or segmentation boundary can protect timing and security.
3. Wireless is valuable when mobility or retrofit cost justifies its variability.
4. Performance claims must be verified under worst-case operating conditions.

9. Cybersecurity, Governance, and Standards

Connecting a machine increases its attack surface. Threats include unauthorized commands, credential theft, spoofed sensor data, denial of service, ransomware propagation, and insecure firmware updates. Essential practices include asset inventory, unique identities, least privilege, network segmentation, secure remote access, authenticated updates, logging, time synchronization, backup, and incident response.

Security must be proportionate to risk. A sensor transmitting noncritical temperature data and a controller commanding a hazardous actuator should not receive identical treatment. Encryption protects confidentiality and can support integrity, but encryption alone does not guarantee availability, correct timing, or functional safety.

Engineers should consult applicable national regulations, industrial cybersecurity guidance, radio-spectrum rules, machine-safety requirements, and organizational policies. Supplier support periods and end-of-life plans are governance issues as well as purchasing issues. Open standards can improve interoperability, but conformance testing and precise data models remain necessary.

10. Contemporary Developments and Best Practices

Current industrial architectures increasingly combine field-level protocols, IP networking, edge computing, cloud services, and digital twins. Edge processing reduces bandwidth and latency by filtering data near the machine. Artificial intelligence can classify vibration or detect anomalies at the edge, but its outputs should be treated as information for decision support unless the complete system has been validated for control or safety use.

Time-sensitive networking (TSN) extends standard Ethernet with mechanisms for synchronization, scheduling, and bounded traffic behavior. Its value is greatest where a converged network must carry both ordinary IP traffic and time-sensitive flows. Private cellular networks and 5G are developing options for flexible factories and mobile assets, but business cases should include deployment, spectrum, device, and operational costs.

Best practices are to:

• separate safety, control, supervisory, and enterprise traffic;
• document timing budgets and message priorities;
• use industrially rated cables, connectors, and enclosures;
• provide physical and logical redundancy where consequences justify it;
• test interference, temperature, vibration, and cable faults;
• monitor latency, jitter, packet loss, and device health;
• maintain configuration baselines and backups;
• design for replacement and future protocol migration;
• avoid selecting a technology solely because its peak data rate is high.

Challenges and Limitations

No communication technology removes the fundamental trade-offs among range, energy, capacity, cost, and timing. Wired systems require installation and may be vulnerable to connector damage. Wireless systems experience multipath propagation, interference, attenuation, and regulatory constraints. Mesh networks can lose performance as routes change. Cellular systems depend on operators and coverage. Industrial Ethernet can require proprietary tools and specialist skills. Legacy protocols may be easy to use but difficult to secure or extend.

The most important limitation of comparison tables is that they summarize typical behavior. Actual performance depends on antenna and cable design, traffic patterns, topology, firmware, environmental conditions, configuration, and maintenance. Engineering decisions should therefore combine standards knowledge with measurement and risk analysis.

Chapter Summary

IoT connectivity in mechatronics is a requirements-matching problem. Ethernet provides scalable high-bandwidth networking, CAN provides robust message-oriented embedded control, and Modbus provides simple register-based interoperability. Industrial Ethernet adds synchronization, prioritization, and real-time mechanisms for automation. Wi-Fi supports mobile and supervisory applications but has variable radio performance. BLE, Zigbee, and Thread support low-power local sensing; RFID and NFC provide identification and deliberate short-range transactions. LPWAN and cellular IoT connect dispersed low-data-rate assets, while 5G offers potential capacity, mobility, and private-network control.

Range, bandwidth, latency, jitter, reliability, power, cost, topology, security, and determinism must be evaluated together. Average throughput is insufficient for control design; serialization time, queuing, arbitration, retransmission, and worst-case behavior matter. A robust architecture separates traffic classes, validates performance under realistic faults and interference, and includes cybersecurity, lifecycle, and regulatory planning.

Key Terms

Bandwidth: Maximum rate at which information can be transmitted.

CAN: Robust message-oriented controller area network used in embedded control.

Determinism: Predictability and boundedness of communication timing.

Ethernet: Widely used wired local-area networking technology.

Industrial Ethernet: Ethernet-based networking adapted for automation and real-time operation.

Jitter: Variation in message latency or periodic timing.

Latency: Time between sending information and receiving or applying it.

LPWAN: Low-power wide-area networking for small, infrequent, long-range messages.

Mesh network: Network in which nodes can forward traffic for other nodes.

Modbus: Simple industrial application protocol commonly transported over serial links or TCP.

NFC: Very short-range communication for deliberate interactions such as tap-based identification.

Payload: Application data carried by a communication frame.

RFID: Radio-frequency identification using readers and electronic tags.

Thread: IPv6-based low-power mesh networking technology using IEEE 802.15.4 radios.

Throughput: Successfully delivered useful data rate.

Topology: Physical or logical arrangement of network nodes and links.

Wi-Fi: Wireless local-area networking technology based on IEEE 802.11.

5G: Fifth-generation cellular communication technology supporting enhanced capacity and industrial use cases.

Review Questions

Short Answer Questions

1. Define determinism and explain why it matters in a motion-control network.
2. Distinguish bandwidth from throughput.
3. Why is CAN suitable for embedded mechatronic control?
4. State two typical applications of LPWAN.
5. Why should safety traffic be treated separately from ordinary control traffic?

Long Answer Questions

1. Compare Ethernet, CAN, Modbus, and industrial Ethernet in terms of topology, timing, bandwidth, and application suitability.
2. Explain the principal trade-offs between Wi-Fi, BLE, Zigbee/Thread, and cellular IoT.
3. Describe a requirements-driven procedure for selecting connectivity in an industrial IoT system.
4. Explain why average data rate alone cannot prove that a network meets a control deadline.
5. Discuss cybersecurity and lifecycle considerations when connecting a legacy machine.

Analytical / Application Questions

1. A machine has 12 sensors, each producing 20 bytes every 10 ms. Estimate the payload data rate and explain what additional information is needed before selecting a network.
2. Design a connectivity architecture for a mobile robot that requires deterministic local motor control, wireless supervisory communication, and cloud-based fleet reporting. Justify each link.
3. A factory proposes replacing a wired servo network with Wi-Fi because Wi-Fi has a higher nominal data rate. Critically evaluate the proposal using latency, jitter, interference, safety, and testing requirements.

Multiple Choice Questions with Answers

Question 1

Which characteristic most directly describes determinism in an industrial network?

A. High peak bandwidth

B. Predictable and bounded communication timing

C. Low equipment purchase price

D. Large geographic coverage

Correct Answer: B. Predictable and bounded communication timing

Question 2

A 16-byte payload transmitted every 20 ms has an ideal payload rate of:

A. 6.4 kbit/s

B. 64 kbit/s

C. 6.4 Mbit/s

D. 640 bit/s

Correct Answer: B. 64 kbit/s

Question 3

Which technology is primarily a message-oriented embedded control bus?

A. CAN

B. NFC

C. LPWAN

D. Wi-Fi

Correct Answer: A. CAN

Question 4

What is a principal limitation of ordinary Modbus polling?

A. It cannot use numerical registers

B. It has no support for wired media

C. Response timing depends on request and response sequencing

D. It requires a mesh topology

Correct Answer: C. Response timing depends on request and response sequencing

Question 5

Which technology is most appropriate for passive object identification and tracking?

A. RFID

B. CAN FD

C. Industrial Ethernet

D. 5G URLLC

Correct Answer: A. RFID

Question 6

Which application is generally least suitable for LPWAN?

A. Remote water-meter reporting

B. Agricultural soil monitoring

C. High-speed closed-loop servo control

D. Periodic infrastructure status messages

Correct Answer: C. High-speed closed-loop servo control

Question 7

Why might fiber-optic Ethernet be selected in an industrial plant?

A. It eliminates all protocol overhead

B. It provides electrical isolation and strong immunity to electromagnetic interference

C. It always has lower latency than every copper network

D. It requires no network configuration

Correct Answer: B. It provides electrical isolation and strong immunity to electromagnetic interference

Question 8

What is a major reason to use edge processing for industrial sensor data?

A. To increase radio interference

B. To eliminate the need for sensor calibration

C. To reduce transmitted data and localize time-sensitive analysis

D. To guarantee functional safety automatically

Correct Answer: C. To reduce transmitted data and localize time-sensitive analysis

Question 9

Which statement about 5G is most accurate?

A. Its advertised peak rate automatically guarantees hard real-time behavior

B. It can support industrial mobility and capacity, but actual performance depends on deployment and configuration

C. It is exclusively a short-range wired technology

D. It cannot be used in private industrial networks

Correct Answer: B. It can support industrial mobility and capacity, but actual performance depends on deployment and configuration

Question 10

A network carries 153,600 bit/s of payload and protocol overhead increases transmitted bits by 30%. The approximate total offered rate is:

A. 107,520 bit/s

B. 153,600 bit/s

C. 199,680 bit/s

D. 460,800 bit/s

Correct Answer: C. 199,680 bit/s