Industrial Networks and Communication Protocols

Introduction

Industrial automation depends on reliable exchange of information among sensors, actuators, programmable logic controllers (PLCs), drives, supervisory computers, safety systems, and enterprise applications. A PLC may execute a control program correctly yet fail to control a process if a network is incorrectly terminated, addresses conflict, messages arrive too late, or devices interpret data differently. Industrial communication is therefore a systems-engineering discipline involving physical media, data-link rules, application semantics, timing, cybersecurity, and maintenance.

This chapter examines serial links, Modbus, PROFIBUS, PROFINET, EtherNet/IP, CAN-based networks, OPC UA, and time-sensitive Ethernet. It also develops methods for selecting, designing, commissioning, and diagnosing industrial networks.

Learning Objectives

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

• distinguish physical, data-link, transport, and application-layer responsibilities;
• select suitable topologies and media for industrial environments;
• configure and diagnose serial, Modbus, PROFIBUS, PROFINET, EtherNet/IP, and CAN-based systems;
• explain determinism, latency, jitter, bandwidth, and interoperability;
• compare OPC UA with field-level protocols;
• evaluate industrial Ethernet and time-sensitive networking (TSN) for automation;
• apply a structured network troubleshooting method.

1. Industrial Communication as a Layered System

A communication system converts process information into signals that can be transmitted, interpreted, and acted upon. Its principal layers are:

1. Physical layer: cable, connector, fiber, radio, voltage levels, signaling rate, and electromagnetic compatibility.
2. Data-link layer: framing, addressing, error detection, media access, and sometimes real-time scheduling.
3. Network and transport layers: logical routing, segmentation, flow control, and reliable delivery where provided.
4. Application layer: meaning of data, services, device profiles, diagnostics, and control commands.

Industrial protocols do not all implement these layers in the same way. Modbus RTU commonly operates over RS-485 and defines an application message format, while IP-based industrial Ethernet uses Ethernet and IP beneath a specialized real-time or application protocol. OPC UA is primarily an information-model and service architecture rather than a replacement for every fieldbus.

Important performance terms include:

• Bandwidth: maximum data rate, normally measured in bit/s.
• Latency: elapsed time between transmission and reception of a message.
• Jitter: variation in latency or message timing.
• Throughput: useful information delivered per unit time.
• Determinism: ability to bound timing behavior sufficiently for the application.
• Availability: proportion of time that the communication service is operational.
• Interoperability: ability of devices from different suppliers to exchange meaningful information.

A network with high bandwidth is not automatically deterministic. Ordinary Ethernet can carry a large amount of data, but congestion, queuing, and retransmission can produce variable delay. Conversely, a lower-speed fieldbus can provide predictable cyclic timing for a modest number of devices.

2. Topology, Media, and Physical Design

A bus connects multiple stations to a shared trunk. It uses cable efficiently but requires correct termination and is vulnerable to a trunk fault. A star connects stations to a central switch or hub; a single branch failure normally affects one station, although failure of the central device can affect many stations. A ring provides an alternate path when implemented with redundancy protocols, but configuration and recovery behavior must be understood. A line or daisy chain is economical and common in machine-level Ethernet devices, but an interruption can disconnect downstream devices.

Copper twisted pair is inexpensive and supports many industrial Ethernet installations. Shielding, bonding, cable segregation, and connector quality are essential near variable-frequency drives, contactors, welding equipment, and high-current conductors. Fiber-optic cable provides electrical isolation, long distance, and immunity to electromagnetic interference, but requires appropriate transceivers and careful handling.

For a balanced electrical bus, termination resistors reduce reflections. In a simplified transmission-line model, the terminating resistance should approximate the characteristic impedance of the cable. Termination is installed at the physical ends of the bus, not at every node. A common RS-485 installation uses approximately 120 Ω termination, but the exact value must follow the physical-layer specification and cable characteristics.

Figure: Common industrial network topologies and their principal physical characteristics.

3. Serial Communication and Modbus

Serial communication transmits bits sequentially. The most common industrial interfaces are RS-232 and RS-485. RS-232 is generally point-to-point and has limited distance and noise tolerance. RS-485 uses differential signaling and supports multiple stations on a bus when compatible transceivers, cable, biasing, and termination are correctly applied.

A serial configuration must agree at both ends. Typical parameters include baud rate, data bits, parity, and stop bits—for example, 19,200 bit/s, 8 data bits, even parity, and one stop bit. A mismatch can produce framing errors or apparently random values. The serial line rate is not equal to the useful application rate because start, parity, stop, address, function, data, and error-checking bits consume capacity.

Modbus is a request-response protocol. In Modbus RTU, a master/client initiates communication and a slave/server responds. The RTU frame includes an address, function code, data, and CRC. A silent interval separates frames; therefore, correct timing and serial configuration are important. Modbus TCP carries Modbus application messages in TCP/IP and normally uses port 502. It removes the RS-485 electrical limitations but does not automatically provide hard real-time behavior.

Common function codes include reading coils, reading discrete inputs, reading holding registers, reading input registers, and writing single or multiple values. A frequent engineering problem is register interpretation. A 32-bit floating-point value may occupy two 16-bit registers, and vendors may differ in word order or byte order. The engineer must consult the device map rather than assume an ordering.

For a request-response network, an approximate scan time can be estimated as:

Here, T_cycle is the complete polling-cycle time, T_request and T_response are transmission times, T_turnaround is the changeover delay, T_processing is device processing time, and T_other includes protocol gaps and retries. The approximation applies when traffic is sequential and the terms are measured or conservatively bounded. It should not be treated as an exact guarantee under faults or variable operating-system scheduling.

For a serial frame, transmission time is approximately:

where N_bits is the total number of transmitted bits and R_baud is the symbol rate in baud. For ordinary binary serial signaling, baud and bit/s are often numerically equal, but they are conceptually distinct.

4. PROFIBUS and PROFINET

PROFIBUS is a fieldbus family widely used for distributed I/O, drives, and process instrumentation. PROFIBUS DP commonly uses RS-485. Devices share a bus, and termination must be active at the two physical ends. Cyclic master-class communication exchanges process data, while acyclic services support parameterization and diagnostics. PROFIBUS PA uses a process-automation physical layer suitable for field instrumentation and can support power and communication over the same medium in appropriate installations.

PROFINET is an Industrial Ethernet system. It uses Ethernet infrastructure and provides cyclic real-time communication, device diagnostics, engineering identification, and integration with standard TCP/IP services. PROFINET RT prioritizes cyclic traffic through an optimized protocol implementation. PROFINET IRT uses more tightly scheduled communication for applications requiring highly synchronized motion performance. The exact timing capability depends on device class, network design, controller load, topology, and configuration.

A PROFINET device is commonly identified during commissioning by a device name, while IP addressing supports engineering and diagnostics. Replacing a device therefore requires attention to both the configured device identity and its network parameters. A managed switch can provide port statistics, link status, VLAN functions, redundancy, and diagnostics, but a switch cannot correct an incorrectly configured device name or incompatible GSD/GSDML data.

5. EtherNet/IP and CIP

EtherNet/IP uses standard Ethernet and IP together with the Common Industrial Protocol (CIP). CIP defines objects, services, device profiles, and data models. Explicit messaging is used for configuration, diagnostics, and less time-critical transactions. Implicit or I/O messaging transfers cyclic process data between an originator and a target, commonly using UDP for efficiency.

EtherNet/IP installations must account for multicast behavior, switch configuration, address management, and traffic loading. Managed switches may use Internet Group Management Protocol (IGMP) snooping to limit multicast flooding. Device-level ring technologies can improve resilience, but the actual recovery time and compatibility must be verified for the installed devices.

CIP Safety extends the architecture for safety-related communication, but ordinary network availability is not equivalent to functional safety. Safety functions require a safety-rated design, validated devices, risk assessment, and compliance with applicable standards such as IEC 61508, IEC 62061, or ISO 13849, depending on the application.

6. CAN and CAN-Based Networks

Controller Area Network (CAN) is a multi-master broadcast bus. Messages are identified by an identifier that also determines arbitration priority: a numerically lower identifier has higher priority in standard CAN arbitration. A transmitting node monitors the bus, and dominant and recessive bit behavior permits nondestructive arbitration. This means a higher-priority message can continue while lower-priority transmitters defer.

CAN includes error detection mechanisms such as bit monitoring, bit stuffing checks, frame checks, acknowledgment, and cyclic redundancy checking. Fault confinement uses error counters and can cause a node to enter error-passive or bus-off states. CANopen adds an application-layer object dictionary, process data objects (PDOs), service data objects (SDOs), network management, and standardized device profiles. DeviceNet is another CAN-based industrial network with its own application conventions and physical-layer implementation.

CAN requires proper termination, commonly 120 Ω at both physical ends of the trunk. Stub length, total cable length, and bit rate are interdependent. High bit rates require shorter networks and better signal integrity. A major design issue is worst-case message latency: high-priority traffic can delay lower-priority messages, especially under heavy bus utilization. Therefore, identifiers should be assigned according to control urgency, not merely device identity.

7. OPC UA and Information Interoperability

OPC Unified Architecture (OPC UA) provides secure, platform-independent communication and a rich information model. It can expose variables, methods, events, metadata, relationships, and historical information. Unlike a simple register map, an OPC UA model can express that a value is a temperature, identify its engineering unit, indicate its quality and timestamp, and associate it with equipment.

OPC UA supports client-server communication and publish-subscribe mechanisms. Security features include application authentication, encryption, message signing, user authorization, and certificate management. Security is effective only when certificates, identities, trust lists, roles, updates, and backups are managed throughout the asset lifecycle.

OPC UA is particularly valuable between PLCs, SCADA systems, manufacturing execution systems, historians, and analytics platforms. It may coexist with Modbus, PROFINET, or EtherNet/IP rather than replace them. A gateway can translate a field device’s data into an OPC UA information model, but translation cannot recover information that the source device never measured or define semantics that were not agreed.

8. Determinism and Time-Sensitive Ethernet

Control applications differ in timing requirements. A supervisory temperature trend may tolerate hundreds of milliseconds, while a coordinated motion system may require tightly bounded cycle time and synchronization. The relevant design variables include cycle period, maximum latency, jitter, clock accuracy, traffic load, and fault-recovery time.

Time-Sensitive Networking is a group of IEEE Ethernet standards intended to provide bounded latency, low jitter, time synchronization, and improved reliability over Ethernet. Important mechanisms include:

• IEEE 802.1AS for time synchronization;
• IEEE 802.1Qbv for scheduled traffic transmission;
• IEEE 802.1Qbu and 802.3br for frame preemption;
• IEEE 802.1Qci for per-stream filtering and policing;
• IEEE 802.1CB for frame replication and elimination in suitable redundant designs.

TSN is not a single protocol or automatic guarantee. Deterministic behavior requires compatible end stations, switches, timing plans, traffic specifications, configuration tools, and validation. TSN can support convergence of information technology and operational technology networks, but safety, security, and lifecycle management remain engineering responsibilities.

9. Network Capacity and Timing Analysis

A first approximation of link utilization is:

where U is utilization as a dimensionless fraction, N_i is the number of bytes transmitted by traffic stream i per period, R is link capacity in bit/s, and T_i is the period in seconds. The summation applies to all streams sharing the link. Protocol headers, inter-frame gaps, acknowledgments, retransmissions, and bursts must be added for a realistic estimate. Utilization below 100% is necessary but not sufficient for deterministic performance.

Worked example. A 100 Mbit/s link carries four cyclic streams. Each sends 200 bytes every 10 ms. Ignoring headers and gaps:

The calculated payload utilization is 0.64%. The result does not prove that latency is acceptable because a burst, a low-priority queue, switch processing, or competing traffic may still cause delay. A design margin is therefore required, and actual traffic should be measured.

10. A Structured Commissioning and Diagnostic Method

Troubleshooting should proceed from the physical layer upward rather than beginning with application assumptions.

1. Define the symptom: identify which devices, services, and time intervals are affected.
2. Check safety and process state: place the machine in a safe condition before manipulating communication or actuators.
3. Inspect physical installation: verify power, connectors, shield continuity, polarity, cable type, bend radius, termination, and link LEDs.
4. Check addressing and identity: look for duplicate IP addresses, incorrect station addresses, incorrect device names, and incompatible subnet settings.
5. Check configuration: compare baud rate, parity, GSD/GSDML or electronic data sheets, firmware, I/O sizes, and timing parameters.
6. Observe diagnostics: examine PLC fault buffers, device status words, switch counters, CRC errors, dropped packets, and bus-off events.
7. Capture traffic: use an appropriate protocol analyzer or mirror port. Confirm request frequency, response validity, retries, malformed frames, and timing.
8. Test systematically: change one variable at a time and document the result.
9. Restore and verify: confirm normal operation, alarms, safety behavior, and configuration backup.

Typical symptoms provide clues. CRC errors suggest noise, cabling, termination, grounding, or excessive length. A device that appears online but has invalid process values may have a register offset, data-type, scaling, or byte-order problem. Intermittent Ethernet disconnections may result from duplex mismatch, poor connectors, power interruptions, duplicate addressing, multicast overload, or an unmanaged switch unsuitable for the application.

11. Illustrative Case: Packaging Cell Network Upgrade

Background

A fictional packaging cell contains a PLC, six distributed I/O stations, two servo drives, a barcode reader, and a supervisory computer. The original installation uses a long RS-485 multidrop network. Operators report slow fault recovery and occasional incorrect barcode values.

Problem Analysis

Inspection finds that one termination resistor is installed at an intermediate station rather than at the cable end. The barcode reader returns two registers for a 32-bit value, but the PLC program assumes the opposite word order. The polling program also reads every device sequentially, so a nonresponsive station creates a long timeout.

Approach and Solution

The bus is rewired so that termination is located at both physical ends, and cable routing is separated from motor power wiring. The barcode data is tested with known values and the correct word order is documented. A timeout and retry policy is introduced, with a bounded number of retries and an explicit fault state rather than indefinite blocking. For the next machine revision, distributed I/O and drives are migrated to a managed industrial Ethernet network with device diagnostics and a ring-capable topology. The supervisory system receives standardized equipment data through an OPC UA server, while real-time control remains at the PLC and drive network.

Outcome

Communication errors disappear during the acceptance test, barcode values are correct, and a failed station is identified by a diagnostic message rather than by unexplained process delay. The Ethernet design improves maintainability, but the project retains physical separation between control traffic and enterprise traffic and applies role-based access to the supervisory interface.

Learning Points

The case demonstrates that protocol selection alone does not solve communication problems. Physical installation, data representation, timeout behavior, diagnostics, and system boundaries must be engineered together. It also illustrates a sound division of responsibility: deterministic control remains local, while OPC UA provides structured information for higher-level systems.

12. Interoperability, Cybersecurity, and Governance

Interoperability has technical and semantic dimensions. Two devices may exchange bytes successfully yet disagree about units, scaling, signedness, timestamps, quality codes, or safe states. Interface specifications should therefore define data type, range, unit, update rate, validity, ownership, failure behavior, and version.

Industrial networks should follow defense-in-depth principles: asset inventory, network segmentation, least privilege, controlled remote access, secure configuration backups, patch assessment, application allowlisting where practical, logging, time synchronization, and tested recovery procedures. Disabling unused ports and services reduces attack surface. Protocols without authentication or encryption should be isolated or protected by compensating controls.

Cybersecurity changes must be managed with operational constraints. An uncontrolled patch or switch replacement can stop production or invalidate a safety assessment. Governance should assign responsibility for certificates, passwords, firmware, backups, network diagrams, change approval, and incident response. Relevant standards and regulations vary by jurisdiction and industry; IEC 62443 provides an important industrial cybersecurity framework, while functional safety standards address hazards and risk reduction rather than general cybersecurity alone.

13. Contemporary Directions and Professional Practice

Industrial networks are increasingly converging with IT systems, cloud services, edge computing, digital twins, and machine-learning applications. These technologies can improve predictive maintenance and anomaly detection, but they introduce data-quality, latency, privacy, and cybersecurity concerns. Artificial intelligence should not be allowed to bypass validated interlocks or safety logic. It is most appropriately used for advisory functions, classification, maintenance prediction, and analysis of historical network or process data unless a separately validated safety case exists.

A competent automation engineer should be able to read wiring and topology drawings, use a protocol analyzer, interpret device manuals, calculate approximate traffic load, configure managed switches, document data semantics, and communicate with electricians, controls engineers, IT personnel, and process operators. Employability increasingly depends on the ability to integrate legacy serial equipment with modern Ethernet and information systems without compromising determinism or safety.

Best-Practice Checklist

• Match protocol and topology to timing, distance, environment, and availability requirements.
• Use specified cable, connectors, shielding, grounding, and termination.
• Document addresses, device names, firmware, data types, byte order, units, and scaling.
• Separate cyclic control traffic from noncritical or enterprise traffic where necessary.
• Establish bounded timeouts, retries, and defined fault states.
• Use managed-switch diagnostics and capture traffic during commissioning.
• Validate redundancy and recovery rather than assuming that a ring is fault tolerant.
• Apply cybersecurity controls without compromising safety or deterministic control.
• Back up configurations and maintain current topology and asset records.
• Test changes in a controlled environment before production deployment.

Chapter Summary

Industrial communication combines physical signaling, framing, addressing, application semantics, timing, diagnostics, and cybersecurity. RS-485 and Modbus remain useful for economical serial communication, but correct termination, timing, and register interpretation are essential. PROFIBUS provides established fieldbus capabilities, while PROFINET and EtherNet/IP extend industrial communication over Ethernet with real-time and diagnostic services. CAN provides efficient, priority-based arbitration and strong error handling for embedded control networks. OPC UA contributes secure, structured information exchange across supervisory and enterprise boundaries. TSN extends Ethernet toward bounded latency and synchronized traffic, but requires compatible equipment and deliberate engineering.

Network design must consider topology, electromagnetic environment, utilization, latency, jitter, determinism, redundancy, and maintainability. Diagnosis is most effective when it proceeds from physical installation through addressing and configuration to captured protocol behavior. Interoperability requires agreement not only on bytes but also on data meaning. Finally, safety and cybersecurity are lifecycle responsibilities that cannot be obtained merely by selecting a modern protocol.

Key Terms

Physical layer: The part of communication concerned with signals, media, connectors, and electrical or optical characteristics.

RS-485: A differential serial physical interface commonly used for multidrop industrial buses.

Modbus RTU: A binary Modbus variant commonly transmitted over serial links.

Modbus TCP: Modbus application messaging transported using TCP/IP.

PROFIBUS: A family of industrial fieldbus technologies for automation and process applications.

PROFINET: An Industrial Ethernet system supporting cyclic real-time communication and diagnostics.

EtherNet/IP: An Industrial Ethernet protocol using CIP over Ethernet and IP networks.

CAN: A multi-master broadcast bus with priority arbitration and extensive error detection.

CANopen: An application-layer protocol and device-profile system based on CAN.

OPC UA: A secure, platform-independent architecture for services and information modeling.

Latency: Time taken for information to travel from source to destination.

Jitter: Variation in message timing or latency.

Determinism: Predictability and boundedness of communication timing.

Termination: Electrical impedance applied at bus ends to reduce signal reflections.

Cyclic communication: Repeated exchange of process data at defined intervals.

Explicit messaging: On-demand communication commonly used for configuration and diagnostics.

TSN: Time-Sensitive Networking mechanisms for scheduled, synchronized, and bounded Ethernet traffic.

Interoperability: Ability of independently produced systems to exchange and use information correctly.

IGMP snooping: Switch behavior that limits multicast forwarding to interested ports.

Bus-off: A CAN node state in which excessive errors cause the node to stop participating in the bus.

Review Questions

Short Answer Questions

1. Distinguish bandwidth, latency, and jitter.
2. Why are termination resistors installed at the ends of an RS-485 or CAN bus?
3. State two differences between Modbus RTU and Modbus TCP.
4. What is the purpose of an OPC UA information model?
5. Why does high bandwidth not guarantee deterministic communication?

Long Answer Questions

1. Explain the design and commissioning requirements of a Modbus RTU network.
2. Compare PROFIBUS, PROFINET, and EtherNet/IP in terms of medium, communication style, and typical applications.
3. Describe CAN arbitration, error handling, and the implications of message priority.
4. Explain how topology, switching, traffic loading, and redundancy affect industrial Ethernet performance.
5. Discuss cybersecurity and functional-safety responsibilities in a connected automation system.

Analytical / Application Questions

1. A serial network shows intermittent CRC errors after a variable-frequency drive is installed. Develop a step-by-step diagnostic plan and identify at least four likely causes.
2. A controller exchanges four 200-byte messages every 10 ms on a 100 Mbit/s link. Calculate payload utilization using the equation in this chapter and explain why the result alone cannot establish determinism.
3. Design a communication architecture for a packaging line containing legacy serial instruments, real-time drives, PLC control, SCADA, and an enterprise data platform. Justify protocol boundaries, topology, security, and diagnostic provisions.

Multiple Choice Questions with Answers

Question 1

Which topology normally requires termination at the two physical ends of a shared bus?

A. Bus

B. Star

C. Fully connected mesh

D. Point-to-point optical link

Correct Answer: A. Bus

Question 2

In Modbus RTU, which device normally initiates a request-response transaction?

A. The slave/server

B. The master/client

C. The termination resistor

D. The switch

Correct Answer: B. The master/client

Question 3

What is a major function of OPC UA?

A. Providing only an RS-485 electrical interface

B. Replacing all PLC scan cycles

C. Providing secure services and structured information modeling

D. Guaranteeing hard real-time motion control in every installation

Correct Answer: C. Providing secure services and structured information modeling

Question 4

In CAN arbitration, which message generally has higher priority?

A. The message with the numerically highest identifier

B. The message with the numerically lowest identifier

C. The message transmitted by the physically closest node

D. The message with the longest payload

Correct Answer: B. The message with the numerically lowest identifier

Question 5

Which condition is most directly associated with network determinism?

A. Unbounded and unpredictable delay

B. Predictable and sufficiently bounded timing

C. Maximum cable length regardless of bit rate

D. Use of wireless communication only

Correct Answer: B. Predictable and sufficiently bounded timing

Question 6

A 100 Mbit/s link carries four 200-byte payloads every 10 ms. Ignoring overhead, what is the payload utilization?

A. 0.064%

B. 0.64%

C. 6.4%

D. 64%

Correct Answer: B. 0.64%

Question 7

Which technology is specifically associated with scheduled Ethernet traffic and time synchronization?

A. TSN

B. RS-232

C. Modbus ASCII only

D. Analog 4–20 mA

Correct Answer: A. TSN

Question 8

What is a likely consequence of incorrect Modbus register word order?

A. The cable automatically changes baud rate

B. A 32-bit value may be interpreted incorrectly

C. The Ethernet switch becomes a terminator

D. CAN arbitration is disabled

Correct Answer: B. A 32-bit value may be interpreted incorrectly

Question 9

Why may IGMP snooping be used in an EtherNet/IP installation?

A. To provide electrical termination for RS-485

B. To limit unnecessary multicast flooding

C. To convert TCP into PROFIBUS DP

D. To assign CAN identifiers

Correct Answer: B. To limit unnecessary multicast flooding

Question 10

Which diagnostic sequence is generally most appropriate?

A. Replace the PLC program before checking wiring

B. Begin with physical checks, then addressing, configuration, and traffic analysis

C. Increase baud rate until errors disappear

D. Disable all alarms permanently

Correct Answer: B. Begin with physical checks, then addressing, configuration, and traffic analysis