Time-Sensitive Networking (TSN) for Industrial Automation: Current Advances and Future Directions

I. INTRODUCTION

• Incompatible vendor protocols make industrial automation systems hierarchical, custom-built, and difficult to integrate, despite the need for interoperable real-time control.The introduction identifies deterministic communication as essential to mission-critical control processes.Source: §I · p. 1
• TSN targets shared Ethernet networks carrying critical and non-critical traffic by providing timing guarantees and temporal isolation for critical flows.The TSN Task Group develops general-purpose protocols for fields including factory and process automation.Source: §I · p. 1
• The paper reviews TSN standardization and research for industrial automation, then examines standards’ support for current and future use cases and the challenges of deployment.Automotive, aerospace, and healthcare applications are outside the survey’s focus.Source: §I · pp. 1–2

A. Industrial Automation

• Industrial automation connects sensors, actuators, robots, control systems, and IT systems to manage processes and machinery across industries.The paper situates this background amid CPS and IoT-driven changes in architecture and system development.Source: A. Industrial Automation · p. 2, §II · p. 2

1) Recent Trends in Industrial Automation:

• Industrial systems need determinism, reliability, availability, interoperability, and support for diverse traffic, with latency and jitter central to deterministic communication.Critical and best-effort traffic have different QoS demands; examples include isochronous control loops with deadlines under 2 ms and cyclic traffic with 2–20 ms latency requirements.Source: 1) Recent Trends in Industrial… · pp. 2–3

2) Deterministic Real-Time Communication:

• Packet latency is end-to-end delay from transmission initiation to complete reception, and industrial applications often require frames to meet a predefined bound.Some applications instead use probabilistic bounds, where occasional violations have negligible effect on perceived multimedia quality.Source: 2) Deterministic Real-Time Communication · p. 3
• Jitter is variation in packet latency; low jitter supports predictable communication and timing precision, particularly for motion control.The passage gives examples of low-jitter requirements for machine tools, automotive radar, and professional audio.Source: 2) Deterministic Real-Time Communication · p. 3
• Latency and jitter are the primary QoS measures for industrial networks; bounding both makes communication deterministic within a specified, predictable time frame.Determinism is relevant to applications including condition monitoring, process automation, and smart manufacturing.Source: 2) Deterministic Real-Time Communication · p. 3

3) The Future of Industrial Ethernet:

• Ethernet fieldbus systems provide real-time services through differing approaches, from TCP/IP best-effort bridging to MAC-layer scheduling and clock synchronization.The described classes span cycle times from 100 ms in Class A, within 10 ms for the Fast Ethernet Class B example, and below 1 ms for Class C.Source: 3) The Future of Industrial Ethernet · pp. 3–4
• No industry-established Ethernet fieldbus meets all summarized industrial requirements, and TSN’s task group works to improve Ethernet reliability and real-time capabilities.The stated TSN improvements include lower latency, determinism, fault tolerance without additional equipment, safety and security support, and vendor interoperability.Source: 3) The Future of Industrial Ethernet · p. 4

B. IEEE 802.1 Overview

• TSN is an IEEE 802.1 standardization effort to integrate real-time capabilities into Ethernet and could eventually replace current industrial communication systems.The overview places this effort within the IEEE 802.1Q bridging standard and its amendments.Source: B. IEEE 802.1 Overview · p. 4
• IEEE 802.1Q specifies LAN and WLAN bridging and the protocol layers above the MAC and LLC layers.A bridge is a network entity with functions specified by the corresponding standards.Source: B. IEEE 802.1 Overview · p. 4
• IEEE 802.1Q traffic classes use port-local priorities from 0 to 7, with 0 lowest and 7 highest; up to eight queues may be implemented per port.Source: B. IEEE 802.1 Overview · p. 4, Table I · p. 4
• Bridges switch, shape, and police traffic, while industrial use also requires worst-case end-to-end latency, efficiency, bandwidth isolation, and zero congestion loss.The passage identifies store-and-forward and cut-through as the two switching operations.Source: B. IEEE 802.1 Overview · p. 4
• Cut-through forwarding can start before a complete frame arrives, but it may forward invalid frames that affect shaping and later valid frames.IEEE 802.1Q does not specify whether cut-through is legitimate in a bridge.Source: B. IEEE 802.1 Overview · pp. 4–5
• IEEE focuses TSN standardization on physical- and link-layer techniques, while the IETF DetNet Working Group focuses on network and higher-layer techniques.Source: B. IEEE 802.1 Overview · p. 5

C. Time-Sensitive Networking (TSN)

• TSN standardizes features to address the performance gap in AVB and the incompatible proprietary protocols used in industrial automation.Its stated goals include temporarily isolating non-critical traffic to guarantee critical-traffic timing behavior.Source: C. Time-Sensitive Networking (TSN) · p. 5
• TSN supports interoperability, scalability, and flexibility through open, vendor-independent standards that can integrate with technologies such as OPC UA.Ethernet-based networking also allows nodes to be added and discovered through standard Ethernet protocols.Source: C. Time-Sensitive Networking (TSN) · p. 5
• A typical TSN switch uses gated priority queues, a global scheduler, and transmission selection to make communication delays predictable and deterministic.A queue transmits only when its gate is open.Source: C. Time-Sensitive Networking (TSN) · p. 5, Figure 2 · p. 5
• The toolbox is flexible, but protocols may not operate independently and some configuration approaches are mutually exclusive.The overview excludes some draft standards still in progress from its Section V discussion.Source: C. Time-Sensitive Networking (TSN) · p. 5, Figure 3 · p. 6
• Its standards include synchronization through IEEE 802.1AS, traffic shaping and scheduling through IEEE 802.1Qbv, IEEE 802.1Qbu, and IEEE 802.3br, reliability through IEEE 802.1CB, and resource management through IEEE 802.1Qcc.IEEE 802.1AS-Rev adds support for fault tolerance and multiple active synchronization masters.Source: C. Time-Sensitive Networking (TSN) · pp. 5–6

III. TSN STANDARDIZATION

• The TSN standards span time synchronization, guaranteed end-to-end latency, reliability, and resource management, with some standards contributing to multiple aspects.IEEE 802.1AS is an example for synchronization, while IEEE 802.1Qbv provides scheduled traffic.Source: §III · p. 6

A. Time Synchronization

• gPTP synchronizes network clocks through messages between a GrandMaster and Clock Slaves, using peer-path delay to calculate link latency and switch residence time.IEEE 802.1AS-Rev supports redundant synchronization trees and multiple gPTP time domains.Source: A. Time Synchronization · pp. 6–7

1) IEEE 802.1Qav Forwarding and Queuing of Time-Sensitive Streams:

• Credit-based shaping regulates transmission rates and bandwidth allocation to provide bounded latency for time-sensitive traffic.A queue's credit accumulates while it waits to transmit and is consumed during transmission.Source: 1) IEEE 802.1Qav Forwarding and… · p. 7
• CBS delay scales may remain too high for industrial control, motivating IEEE 802.1Qbv, IEEE 802.1Qch, and IEEE 802.1Qcr.CBS with SRP targets less than 250 µs per bridge and up to 2 ms for class A and 50 ms for class B in a simple network setup.Source: 1) IEEE 802.1Qav Forwarding and… · p. 7

2) IEEE 802.1Qbv

• TAS schedules critical traffic in protected time windows, using synchronized gate-control lists that repeat on a cycle.Guard bands close all gates after best-effort traffic to prevent large frames from interfering with periodic traffic.Source: 2) IEEE 802.1Qbv · p. 8

3) IEEE 802.3br and 802.1Qbu Interspersing Express Traffic and Frame Preemption:

• IEEE 802.1Qbu and IEEE 802.3br allow urgent frames to preempt non-critical frames already in transmission, improving determinism and real-time performance.With scheduled traffic, frame preemption can reduce the guard band from an MTU of 1522 bytes to 127 bytes, though preemption induces overhead.Source: 3) IEEE 802.3br and 802.1Qbu… · p. 8

4) IEEE 802.1Qch Cyclic Queuing and Forwarding (CQF):

• CQF synchronizes enqueue and dequeue operations in alternating cycles; a frame received in cycle x is sent in cycle x + 1.Its end-to-end latency depends on route length and cycle size T.Source: 4) IEEE 802.1Qch Cyclic Queuing and… · pp. 8–9, Figure 5 · p. 8
• CQF is designed for limited-scale networks with time synchronization, and its cycle length must be designed carefully to guarantee bounded, deterministic latency.When combined with frame preemption, CQF may use shorter cycles, provided all fragments arrive within the scheduled cycle.Source: 4) IEEE 802.1Qch Cyclic Queuing and… · pp. 8–9

5) IEEE 802.1Qcr Asynchronous Traffic Shaping (ATS):

• ATS reshapes streams at each hop asynchronously, prioritizing urgent traffic while supporting deterministic, relatively low delay without prescribed traffic patterns.Its token-bucket emulation controls average rate while allowing limited bursts; it suits hybrid periodic and sporadic traffic.Source: 5) IEEE 802.1Qcr Asynchronous Traffic… · p. 9
• The shaper summary compares synchronization needs, main technology, topology dependence, and triggering mechanisms across shapers.Topology dependence concerns whether longer paths generally increase end-to-end delay.Source: 5) IEEE 802.1Qcr Asynchronous Traffic… · p. 9, Table II · p. 10

C. Reliability

• TSN reliability mechanisms exploit channel redundancy through replicated frames, multiple paths, and per-stream filtering and policing.IEEE 802.1CB handles frame replicas, IEEE 802.1Qca manages multiple paths, and IEEE 802.1Qci defines filtering and policing.Source: C. Reliability · p. 9

1) IEEE 802.1CB Frame Replication and Elimination for Reliability (FRER):

• IEEE 802.1CB identifies and replicates frames, sends duplicates across separate paths, then removes duplicates using sequence numbers at the destination or relays.Replication can reduce packet loss impact from congestion or failures, but uses additional network resources; replication volume can be limited by traffic class and path link quality.Source: 1) IEEE 802.1CB Frame Replication and… · pp. 9–10

2) IEEE 802.1Qca Path Control and Reservation (PCR):

• IEEE 802.1Qca uses TLV and IS-IS to manage multiple routes, supporting explicit path and bandwidth control, redundancy, and restoration after topology changes.Its protection mechanisms include redundant trees and loop-free alternates for unicast flows.Source: 2) IEEE 802.1Qca Path Control and… · p. 10

3) IEEE 802.1Qci Per-Stream Filtering and Policing (PSFT):

• IEEE 802.1Qci filters, polices, and selects service classes per stream, applying mitigation when frames violate configured rules.The standard protects QoS among streams sharing an egress queue and can drop unauthorized or malicious transmissions.Source: 3) IEEE 802.1Qci Per-Stream Filtering… · p. 10

D. Resource Management

• TSN resource management allocates and uses network resources to meet industrial requirements, including bandwidth management, traffic prioritization, and QoS maintenance.The section covers IEEE 802.1Qcp, IEEE 802.1Qcc, and IEEE 802.1CS.Source: D. Resource Management · p. 10

1) IEEE 802.1Qcp YANG Data Model:

• IEEE 802.1Qcp defines a YANG data model for configuring and reporting the status of IEEE 802.1 bridge devices.YANG organizes configuration and state data, operations, and notifications in a hierarchical tree; its industry-wide implementation provides a universal interface for integrating resource management across devices.Source: 1) IEEE 802.1Qcp YANG Data Model · p. 10

2) IEEE 802.1Qcc SRP Enhancements and Performance Improvements:

• IEEE 802.1Qcc extends SRP to support more complex shaping and defines a UNI through which user stream requirements guide network configuration.It defines fully centralized, centralized-network/distributed-user, and fully distributed configuration models.Source: 2) IEEE 802.1Qcc SRP Enhancements and… · pp. 10–11, Figure 7 · p. 11

3) IEEE 802.1CS Link-Local Reservation Protocol (LRP):

• IEEE 802.1CS replicates registration databases link-locally and supports databases on the order of 1 Mbyte, compared with MRP’s 1500-byte capacity.Local operation enables resource management without centralized management.Source: 3) IEEE 802.1CS Link-Local Reservation… · p. 11

4) IEEE 802.1Qdd Resource Allocation Protocol (RAP):

• IEEE 802.1Qdd RAP uses LRP for dynamic unicast and multicast reservation in fully distributed networks and supports latency calculation and reporting.RAP standardization was still ongoing at Draft 0.9 when the paper was written.Source: 4) IEEE 802.1Qdd Resource Allocation… · p. 11

• TSN is described as outperforming PROFINET IRT in cycle time below 50 microseconds, jitter less than ±100 nanoseconds, and scalability beyond 10,000 nodes.The paper also cites openness, vendor-independent interoperability, and IT/OT integration as TSN advantages.Source: 4) IEEE 802.1Qdd Resource Allocation… · pp. 11–12

A. Why Do We Need TSN in Industrial Automation?

• TSN’s interoperability, convergence, and determinism are presented as benefits for industrial automation, where incompatible vendor protocols can cause lock-in and costly integration choices.The paper describes TSN’s open network standard as allowing multi-vendor devices to interoperate and be configured through a single standard interface.Source: A. Why Do We Need TSN in Industrial… · p. 12

B. TSN-based Converged Industrial Networks

• TSN can carry critical and non-critical traffic on one network while providing deterministic performance guarantees for critical traffic.The paper describes TSN as a Layer 2 building block for bounded latency and jitter, requiring higher-layer protocols for end-to-end QoS.Source: A. Why Do We Need TSN in Industrial… · p. 12, B. TSN-based Converged Industrial… · p. 12

1) Fieldbus over TSN:

• Fieldbus traffic can use TSN either through a new TSN network or through active gateways that convert network traffic to TSN-compatible Ethernet frames.The paper notes that current fieldbus systems do not meet the convergence needs of emerging applications such as close IT/OT integration.Source: 1) Fieldbus over TSN · p. 13
• PROFINET over TSN supplements PROFINET at the Ethernet layer without changing its existing services or profiles, while EtherCAT over TSN requires no changes to EtherCAT slaves.The described EtherCAT approach uses segments as TSN structuring elements; the section also discusses EtherNet/IP and CC-Link IE TSN integrations.Source: 1) Fieldbus over TSN · p. 13

2) OPC UA over TSN:

• OPC UA over TSN’s PubSub model supports deterministic real-time transmission while retaining OPC UA’s flexibility and openness; it complements rather than replaces fieldbus-over-TSN systems.The paper distinguishes OPC UA’s factory- and enterprise-level communication role from fieldbus communication between field devices and controllers.Source: 2) OPC UA over TSN · p. 13
• Reported OPC UA-over-TSN efforts include a two-tier manufacturing architecture validated in a laboratory system and an implementation using open62541 for PubSub transport.A field-device case study reports implementation using commercial off-the-shelf software and hardware components.Source: 2) OPC UA over TSN · pp. 13–14

C. Traffic Scheduling

• TAS uses network-wide synchronization and time-triggered scheduling for deterministic guarantees, with switch gates configured by a network-wide schedule.The schedule combines gate-control-list settings and traffic-to-queue mappings, configured by the CNC.Source: C. Traffic Scheduling · p. 14
• A comparison of 17 TAS-based scheduling methods found no method dominated across all scenarios, and differing experimental settings complicate fair comparisons.The review concludes that TSN scheduling remains an open problem and that findings may be valid only under specific settings.Source: C. Traffic Scheduling · p. 15

D. TSN Testbeds

• Physical testbeds offer high-fidelity evaluation with real hardware, while theoretical analysis abstracts real-world behavior and simulations lack hardware components.Testbeds can support controlled experiments against stringent industrial requirements.Source: D. TSN Testbeds · p. 15
• Testbeds can assess interoperability, network configuration challenges, and cybersecurity vulnerabilities, while their development faces cost, sharing, and fidelity challenges.Replicating real-world industrial conditions is difficult and may require substantial hardware, software, and skilled-personnel resources.Source: D. TSN Testbeds · p. 15
• Industrial TSN testbeds fall into general, OPC UA TSN, and wireless TSN categories, respectively examining core TSN functions, multi-vendor information flow, and wireless media.Wireless testbeds explore Wi-Fi and 5G.Source: D. TSN Testbeds · p. 15

A. Time Synchronization

• Multi-hop gPTP synchronization can fail through propagated timestamp errors, asymmetric delays, clock drift, or compromised devices spreading erroneous time information,.These effects can accumulate across the synchronization spanning tree.Source: A. Time Synchronization · pp. 15–16
• IEEE 802.1AS relies on BMCA to switch grandmasters, while IEEE P802.1ASdm specifies two simultaneous time domains; resilience messaging can add bandwidth and control-plane overhead,,.Sync-message settings must balance accuracy against overhead.Source: A. Time Synchronization · p. 16
• Industrial switch hierarchies, differing synchronization quality, and device heterogeneity complicate centralized synchronization and require schemes to account for network topology.Because TSN operates at the MAC layer, small upper-layer time slips can affect it significantly.Source: A. Time Synchronization · p. 16

B. Latency Guarantee

• TAS schedule generation is NP-complete, and solutions range from specialized heuristics and metaheuristics to exact ILP or SMT solvers –.These approaches trade scheduling efficiency against precision and schedulability, particularly as model complexity and network scale change.Source: B. Latency Guarantee · p. 16
• More complex TAS models can improve schedulability but raise computational costs; heuristics scale efficiently yet may find no feasible schedule, while exact methods suit smaller networks.A feasible schedule may not be found within a reasonable time on resource-constrained systems.Source: B. Latency Guarantee · p. 16
• TAS requires co-design of end stations and switch schedules, and collisions between time-triggered and synchronization traffic can cause synchronization errors or missed deadlines,.Real-time TSN-compatible end-station design remains an open problem.Source: B. Latency Guarantee · p. 16

2) IEEE 802.1Qbu:

• Multi-level frame preemption can increase flexibility and reduce latency relative to one-level preemption, but increases configuration complexity and complicates worst-case analysis,.Combining preemption with multiple shapers further complicates configuration and can produce timing-incorrect configurations [162],.Source: 2) IEEE 802.1Qbu · pp. 16–17
• Each preemption adds 12 bytes of fixed overhead plus a 12-byte interframe gap, and each additional preemption level adds hardware overhead.The performance gains must be weighed against these costs.Source: 2) IEEE 802.1Qbu · p. 17

3) Other Shapers:

• CBS avoids best-effort starvation at the expense of higher-priority flow delay, while PS requires cycle alignment and ATS trades critical-flow determinism for better average stream latency,.CBS timing analysis is complex, and high-volume traffic may weaken its delay guarantees; ATS delay bounds remain conservative.Source: 3) Other Shapers · p. 17
• Shaper guarantees commonly assume homogeneous devices and global synchronization, whereas industrial networks include unscheduled or unsynchronized devices requiring tailored scheduling and timing analysis.Such mechanisms must preserve critical-flow timing despite differing device capabilities and synchronization states.Source: 3) Other Shapers · p. 17

C. Reliability

• TSN reliability standards leave implementation methods open; spatial redundancy handles permanent faults but reserves network resources, while retransmission-based temporal redundancy suits transient faults –.Both consume resources, and sharing them requires precise success-probability analysis; learning-based approaches are also discussed –.Source: C. Reliability · pp. 17–18

D. Resource Management

• IEEE 802.1Qcc provides network management and reconfiguration models, complemented by IEEE 802.1Qcp’s YANG configuration data template.No single management model suits every industrial scenario.Source: D. Resource Management · p. 18
• Centralized management offers global knowledge for precise configuration but creates a single point of failure, scalability limits, and potential controller bottlenecks; distributed management responds faster but converges more slowly.Distributed approaches may also cause transmission collisions and lower network performance than centralized methods.Source: D. Resource Management · p. 18
• The CNC and CUC functions and their UNI interface remain insufficiently defined, while ongoing IEEE P802.1Qdj work clarifies functions and communication capabilities.The choice among protocols such as NETCONF, CORECONF, and RESTCONF remains an open research issue.Source: D. Resource Management · p. 18
• Online reconfiguration must handle network dynamics without downtime while preserving critical-traffic QoS, but complex GCL and routing decisions make this difficult.Centralized methods have global information, yet SMT-based solutions may take too long for on-the-fly changes.Source: D. Resource Management · p. 18

A. TSN Deployment

• TSN standards still require modification, testing, and validation before wide field deployment, and selecting mechanisms for each application is a key deployment concern.The review frames configuration synthesis as determining mechanisms that satisfy the application requirements.Source: A. TSN Deployment · p. 19, 1) Configuration Synthesis · p. 19

1) Configuration Synthesis:

• Network-wide configuration synthesis selects combined mechanisms to meet application requirements, and poor choices can complicate configurations or drive costly replacement and re-verification,.Designers must understand required functionality and trade off among TSN standards.Source: 1) Configuration Synthesis · p. 19

2) Coexistence of Shapers:

• Coexisting shapers such as TAS + CBS and TAS + CQF may interact and affect performance, making preservation of end-to-end timing analysis an open research question –.The review calls for study of the benefits and pitfalls of such combinations.Source: 2) Coexistence of Shapers · p. 19

3) Dynamic Reconfiguration:

• Dynamic TSN reconfiguration must accommodate topology and traffic changes at runtime; research should quantify reconfiguration overhead and explore incremental or pre-allocated approaches –.Offline schedules provide deterministic performance but do not enable flexible network reconfiguration.Source: 3) Dynamic Reconfiguration · p. 19

4) Security:

• TSN security remains open research: as systems become more publicly exposed, further work is needed on early threat detection and effective mitigation.The IEEE 802.1 Security TG is working with the TSN TG on secure capabilities.Source: 4) Security · p. 19

1) DetNet:

• TSN latency guarantees currently extend only up to seven hops, limiting scalability; TSN/DetNet integration targets larger routed networks but faces QoS, synchronization, and scheduling challenges –.Distributed or hierarchical scheduling is proposed as a possible direction.Source: B. Large-Scale Industrial Networks · p. 19, 1) DetNet · p. 19

2) Virtualization:

• Virtualization may improve flexibility and scalability, but VM-induced latency unpredictability can undermine determinism, while placement, migration, and edge containerization remain challenging –.TSN use in virtual environments is a relatively new research area.Source: 2) Virtualization · pp. 19–20

C. Wireless TSN

• Wireless TSN could reduce wiring costs and improve mobility, but variable capacity, channel stochasticity, and interference complicate deterministic timing and reliability guarantees,.Closed-loop control, mobile robots, and autonomous ground vehicles are identified as potential use cases.Source: C. Wireless TSN · p. 20

1) Time Synchronization:

• Wireless synchronization spans 802.11 schemes and 5G boundary/transparent clocks, with transparent clocks preferred for accuracy despite unresolved timestamping, handover, and delay-asymmetry challenges.Source: 1) Time Synchronization · p. 20

2) Traffic Scheduling:

• 802.11 scheduling research replaces contention with TDMA or 802.1Qbv, while 5G translators coordinate TSN schedules through QoS and buffering; mobility still demands scheduling enhancements.Source: 2) Traffic Scheduling · pp. 20–21

3) Reliability:

• Wireless reliability uses intra-frame, inter-frame, and multipath redundancy; passive redundancy avoids active retransmission's higher latency, while interference and power efficiency remain concerns.Source: 3) Reliability · p. 21

4) Wireless Security:

• Wireless links are more vulnerable to eavesdropping and tampering; security measures can add delay that affects TSN determinism.Source: 4) Wireless Security · p. 21
• Mobility, radio-resource costs, and aggregation overhead create design trade-offs; hybrid wired-wireless TSN must preserve end-to-end latency, synchronization, and critical-flow coexistence.Source: 4) Wireless Security · p. 21

VII. CONCLUSION

• Because fieldbus protocols lack cross-vendor interoperability, the review surveys published and draft TSN standards and points to standards optimization and technology integration.Source: §VII · pp. 21–22