PaperlayerView exact revision on arXiv

Source-linked AI summary

Real-Time Performance of Industrial IoT Communication Technologies: A Review

Ilja Behnke, Henrik Austad

arXiv:2311.08852v1cs.NIcs.DC

TL;DR

Industrial networks must support heterogeneous traffic and new IIoT scenarios with demanding real-time requirements. This article identifies scenarios and requirements, reviews networking technologies and recent work, and finds that wireless predictability remains a gap while wired flexibility and reconfiguration remain challenging.

  • Problem

    IIoT scenarios impose real-time networking requirements, but wireless solutions struggle to meet strict latency predictability or multimedia bandwidth needs at scale.

    Source: A. Bridging the Gap between IIoT… · p. 9

  • Method

    The article identifies IIoT scenarios, derives their networking requirements, reviews established technologies and recent peer-reviewed work, and assesses their applicability.

    Source: §I · p. 1, §IV · p. 7

  • Results

    Wireless TSN may help provide predictability, but radio remains inherently unpredictable; wired TSN flexibility and reconfigurability remain largely unsolved challenges.

    Source: A. Bridging the Gap between IIoT… · p. 9

  • Takeaways & Limitations

    Wireless technologies with a measure of predictability are a key gap-filler for IIoT scenarios, while combining TSN controllers and SDN is a promising avenue across WLAN and 5G.

    Source: A. Bridging the Gap between IIoT… · p. 9

Abstract

from arXiv · show

With the growing need for automation and the ongoing merge of OT and IT, industrial networks have to transport a high amount of heterogeneous data with mixed criticality such as control traffic, sensor data, and configuration messages. Current advances in IT technologies furthermore enable a new set of automation scenarios under the roof of Industry 4.0 and IIoT where industrial networks now have to meet new requirements in flexibility and reliability. The necessary real-time guarantees will place significant demands on the networks. In this paper, we identify IIoT use cases and infer real-time requirements along several axes before bridging the gap between real-time network technologies and the identified scenarios. We review real-time networking technologies and present peer-reviewed works from the past 5 years for industrial environments. We investigate how these can be applied to controllers, systems, and embedded devices. Finally, we discuss open challenges for real-time communication technologies to enable the identified scenarios. The review shows academic interest in the field of real-time communication technologies but also highlights a lack of a fixed set of standards important for trust in safety and reliability, especially where wireless technologies are concerned.

I. INTRODUCTION

Industry 4.0 and CPS scenarios connect operational and business networks while extending strict real-time demands to networked systems. The article surveys IIoT scenarios, derives their networking requirements, reviews technologies and recent work, and discusses open challenges.

I. INTRODUCTION

  • Industry 4.0 scenarios include autonomous drones, remote control, real-time monitoring, and predictive maintenance, alongside expectations for production agility and flexibility.Survey respondents also expected reduced time to market, while current sensor use focused on monitoring, tracking, and predictive maintenance.Source: §I · p. 1
  • Industrial networks face demands for greater flexibility and scalability as pervasive sensors and OT–IT interconnectedness support Industry 4.0 scenarios.Existing specialized protocols and standards address real-time needs, while research also considers scalability, business/factory-network integration, and mobile computing.Source: §I · p. 1
  • CPS control and safety tasks extend strict latency and predictability requirements from embedded systems to network routing and network participants.Conventional IP routing is best-effort, with limited predictability of latency and packet arrival; wireless jitter adds difficulty for mobile scenarios.Source: §I · p. 1
  • The article identifies IIoT scenarios, derives their real-time networking requirements, reviews established technologies and recent research, and discusses open challenges.It also presents existing standards and technologies and recent work on real-time IIoT networking solutions.Source: §I · p. 1

II. REGARDED IIOT SCENARIOS

The paper surveys IIoT scenario classes and uses them to derive rough network-requirement profiles for assessing real-time solutions. The classes span mobile control, monitoring and data analysis, worker safety, offloading, and wearable systems.

II. REGARDED IIOT SCENARIOS

  • The survey classifies scenarios from 31 peer-reviewed papers and articles, then derives network and real-time requirements to assess enabling technologies.The reviewed use cases were published in the preceding 10 years and appeared as either a paper's focus or an example implementation.Source: §II · p. 2

A. Scenario Classes

  • Autonomous vehicles depend on networks for collision prevention and efficiency, with dynamic navigation and real-time motor control posing challenges at scale and near workers.Their networking needs include high predictability and latency; bandwidth depends on whether sensor data such as video or LIDAR is transmitted.Source: A. Scenario Classes · p. 2
  • Remote control shifts controllers away from real-time devices, bridging formerly air-gapped industrial systems to external networks and conventional IP.These use cases require high real-time performance, sometimes wireless communication, but transmit relatively low-bandwidth control commands.Source: A. Scenario Classes · p. 2
  • Data analysis and monitoring transmit broad sensor streams that may share a medium with control traffic, while typically requiring high bandwidth rather than hard real-time guarantees.Use cases include wireless sensor aggregation, predictive maintenance, business decisions, and active energy management.Source: A. Scenario Classes · p. 2
  • Worker-safety systems range from sensor-triggered alarms to automatic hazard responses, with the latter requiring the highest predictability and lowest latency.Automatically reactive systems should operate independently, making distributed solutions unsuitable for those setups.Source: A. Scenario Classes · p. 2
  • Task offloading targets mostly soft real-time work, balancing latency and reliability across shared network and compute resources.Hard real-time tasks with high predictability requirements are excluded, and network delay can prevent offloading tasks with very tight latency limits.Source: A. Scenario Classes · p. 2
  • Wearables provide human-machine interfaces for remote control, augmented-reality monitoring, data collection, and safety, while adding energy, size, and wireless constraints.The wearable discussion focuses on requirements specific to wearables and augmented reality; remote-control requirements are covered in another scenario class.Source: A. Scenario Classes · p. 3

B. Deriving Network Requirements

The paper generalizes use cases into rough network-requirement profiles, relating scenario feasibility to real-time guarantees and network flexibility. It also situates industrial networking in the shift from relay-based control to PLCs.

B. Deriving Network Requirements

  • The feasible IIoT scenario spectrum grows with stricter network guarantees that do not limit flexibility, while wireless technologies are necessary for many identified scenarios.General IP networking is needed for external access or cloud computing, but IT/OT integration is not the enabler for most real-time-relevant scenarios.Source: B. Deriving Network Requirements · p. 3
  • The requirement profiles are rough generalizations because network needs depend on underlying real-time tasks and how messages enter the control flow.Greater distance among participants increases scenario complexity and limits real-time guarantees and criticality levels.Source: B. Deriving Network Requirements · p. 3

III. INDUSTRIAL NETWORKING

  • The first PLC, introduced in 1969, changed production-system design and replaced relay cabinets with compact devices.The Modicon model 084 had 16 inputs, 16 outputs, and 1 kB of memory; contemporary PLCs have hundreds of I/O and more memory and speed.Source: §III · p. 3
  • Table II connects industrial networks with use cases, framing the networking technologies discussed in the following sections.The passage does not provide the table's specific entries.Source: §III · p. 3

A. Fieldbuses

Industrial networking developed from specialized fieldbuses toward Ethernet and packet-switched networks. Industrial Ethernet and TSN adapt Ethernet for industrial latency and reliability needs, with TSN trading some flexibility for determinism.

A. Fieldbuses

  • Fieldbuses such as Modbus and Profibus emerged for industrial communication, while incompatibility across protocols drove standards efforts; CAN encodes message priority in its identifier.The IEC 61158 standard ultimately collected competing fieldbus protocols.Source: A. Fieldbuses · p. 3

B. Packet Switched Networks

  • Ethernet packet-switching routes frames individually and adapts to topology changes; TCP resends lost data, whereas UDP provides no indication that traffic arrived.Ethernet grew from 10 Mbps to 400 Gbps, according to the passage.Source: B. Packet Switched Networks · pp. 3–4

1) Industrial Ethernet:

  • Industrial Ethernet adapts Ethernet for applications with latency and reliability requirements; common real-time protocols include EtherNet/IP, Profinet, and EtherCAT.TTEthernet provides time-triggered, rate-constrained, and best-effort services.Source: 1) Industrial Ethernet · p. 4

2) Time Sensitive Networking:

  • TSN configures switched COTS Ethernet for bounded latency and very low packet loss on critical traffic, using static routes to reduce jitter and out-of-order delivery.Its fine-grained QoS mechanisms are discussed later in the paper.Source: 2) Time Sensitive Networking · p. 4

C. Wireless networks

Wireless networks can reduce installation and maintenance costs and may be necessary for some industrial uses, but their shared medium creates constraints. Wireless protocols make different trade-offs, while the review notes limited hard performance data for 5G URLLC.

C. Wireless networks

  • Wireless sensors can cost 60–90% less than wired solutions in some scenarios, and can be the only viable option for uses such as tire-pressure monitoring.The passage also cites tool deflection measurements as an example where wireless may be the only viable option.Source: C. Wireless networks · p. 4
  • Wireless network integrity depends on cooperative participants transmitting only in allocated slots, and wireless systems lack wired networks’ resilience to interference and predictable bandwidth.The passage also notes that wireless eavesdropping does not require proximity to wired infrastructure.Source: C. Wireless networks · p. 4
  • Wireless protocols trade among range, bandwidth, latency, host capacity, and interference resistance; improvements in one attribute can come at the expense of another.Longer range is associated with lower bandwidth and higher latency, while denser encoding for high bandwidth can increase interference susceptibility.Source: C. Wireless networks · p. 4
  • WirelessHART and ISA 100.11a are common industrial protocols; ZigBee appears in some industrial settings but is primarily intended for home automation and low-criticality systems.Bluetooth is described as low-power and short-range, with comparably high available bandwidth.Source: C. Wireless networks · p. 4
  • WLAN 6E adds 5 GHz and 6 GHz operation and improved clock accuracy for PTP, while 5G NPN supports industrial use cases including URLLC.The review says hard data to evaluate 5G URLLC performance is limited.Source: C. Wireless networks · pp. 4–5

D. Quality of Service

Industrial QoS must reconcile traffic treatment with variable routing, transmission delays, and finite buffers. The reviewed approaches combine stream reservations, traffic shaping, and scheduling, with different fit for regular, critical, or sporadic traffic.

D. Quality of Service

  • Ethernet’s multiple routes, non-preemptive packet transmission, and store-and-forward buffering can create variable delays, out-of-order delivery, or packet loss under high traffic.Packets may be dropped when bridge buffers are exhausted.Source: D. Quality of Service · p. 5
  • QoS describes how frames in a stream are treated relative to other frames; DiffServ scales through class-based service levels but cannot distinguish streams within a class.The passage notes that service for a class may vary between operators.Source: D. Quality of Service · p. 5
  • IntServ maps each stream to a service level, but its scalability is tied to stream count; a proposed hybrid uses IntServ at network edges and DiffServ in the core.The hybrid approach is presented as one possible approach of the IETF Deterministic Networking group.Source: D. Quality of Service · p. 5
  • TSN’s Credit Based Shaper groups streams by traffic class and shapes traffic to reserved bandwidth, addressing transient overload from bursty traffic.The passage says this overload mitigation has been shown using Network Calculus.Source: D. Quality of Service · p. 5
  • TAS schedules synchronized transmission windows for regular critical traffic, while ATS supports sporadic critical traffic with per-stream shaping; TAS requires central management.The passage describes TAS as complex and suited to the most critical streams, and reports that ATS does not increase worst-case delay.Source: D. Quality of Service · pp. 5–6

E. Common Network Requirements for Industrial Automation

The review adds bounded latency and time synchronization to established industrial-network requirements. It also frames determinism as a central challenge for packet-switched networks and surveys recent work against scenario classes.

E. Common Network Requirements for Industrial Automation

  • The paper extends four wireless-network criteria—security, dependable sensor-data access, interoperability, and an active research community—with bounded latency and time synchronization for industrial networks.The added metrics address deterministic transport requirements identified for both wired and wireless networks.Source: E. Common Network Requirements for… · p. 6
  • The review evaluates common industrial-network solutions along security, determinism, and bandwidth, grouping reliability, availability, and latency under determinism.It distinguishes security by external measures, integrity, confidentiality, and secure authentication and authorization.Source: E. Common Network Requirements for… · p. 6

IV. ONGOING EFFORTS TOWARDS REAL-TIME NETWORKS

  • The ongoing-work survey covers peer-reviewed papers from the prior five years and compares their scope and applicability to the review’s scenario classes.Packet-switched networks alone are described as inadequate for inherent real-time predictability because network load, transmission times, routes, and wireless packet loss are difficult to predict.Source: §IV · pp. 6–7

A. Network Architectures

Network-architecture research addresses IT/OT integration and heterogeneous IIoT requirements through SDN, IP-based reliability designs, and OPC UA integration. These approaches target manageability, security, and real-time communication in industrial networks.

A. Network Architectures

  • Recent architecture research responds to Industry 4.0 and IIoT demands for IT/OT integration and heterogeneous network management.The field includes SDN, IT/OT integration, and OPC UA.Source: A. Network Architectures · p. 7
  • A system architecture for heterogeneous wireless distributed SDN networks aims to improve parameter manageability and facilitate TSN integration with legacy networks.Source: A. Network Architectures · p. 7
  • An IP-based architecture combines zero-loss redundancy, TSN, and edge computing to address reliability and predictability for time-critical IIoT applications.Source: A. Network Architectures · p. 7
  • An SDN-enabled multilayer architecture targets layer-specific security and real-time properties, but architectural changes alone did not secure highly critical physical-machine layers during a DDoS attack.Source: A. Network Architectures · p. 7
  • OPC UA architectures and assessments show that the protocol can integrate real-time communication requirements into manufacturing Ethernet networks.OPC UA was introduced as an interoperability standard for reliable information exchange.Source: A. Network Architectures · p. 7

B. SDN

The reviewed work applies SDN to task offloading, traffic scheduling, and reconfigurable industrial Ethernet, while TSN research improves migration, scheduling scalability, and traffic handling. The cited work also examines TSN integration with OPC UA and DDS.

B. SDN

  • SDN-based task offloading prioritizes tasks using real-time parameters and sends them to fog resources in priority order.Source: B. SDN · p. 7
  • A dynamic SDN scheduler uses real-time network analysis, link QoS monitoring, and multipath routing to schedule delay-sensitive traffic.Source: B. SDN · p. 7
  • A PTP-synchronized, time-slotted software-defined Industrial Ethernet design targets scalability and reconfiguration without frequent infrastructure changes.It addresses poor scalability, limited self-configuration, and proprietary-hardware costs.Source: B. SDN · p. 7

C. TSN

  • Time-Aware Forwarding makes existing networking hardware compatible with TSN, supporting more flexible migration from Profinet.Source: C. TSN · p. 7
  • Hermes, a heuristic multi-queue scheduler, reduces scheduling complexity in TSN gate-control-list synthesis to improve network scalability.Source: C. TSN · p. 7
  • Egress TT buffers time-triggered packets only at the final network node, permitting non-TSN routing segments and reducing computation while somewhat increasing latency.Source: C. TSN · pp. 7–8
  • Studies report benefits from TSN integration with OPC-UA and improved reliability and reduced delay for DDS traffic protected and expedited by TSN.Source: C. TSN · p. 8

D. Real-Time Aware Wireless Technologies

The reviewed work explores real-time wireless methods, 5G approaches, and device-layer mechanisms for meeting industrial communication requirements. Across these areas, proposals address wireless packet loss and jitter, 5G service trade-offs, and priority-aware processing on embedded devices.

D. Real-Time Aware Wireless Technologies

E. Fifth Generation Mobile Networks

  • 5G slicing can separate virtual networks and support connection-specific latency control, while private 5G implementation remains feasible mainly for large companies with networking backgrounds.For critical mMTC, URLLC supports excellent real-time behavior but relatively few devices at its cost; mMTC connects many devices but its individual QoS does not fit sensor networks.Source: E. Fifth Generation Mobile Networks · p. 8

F. Device-Layer Technologies

  • Device-layer proposals prioritize real-time packet processing through early classification, rate limiting, priority-aware multi-queue NICs, and hardware MAC filtering.Source: F. Device-Layer Technologies · p. 9
  • TSN-extended Timed C treats network channels as language primitives and reliably forwards critical traffic with minimal delay and jitter under large network loads.Source: F. Device-Layer Technologies · p. 9

V. DISCUSSION

Industry 4.0 increases demands for flexible, dynamic industrial networks, while wireless scenarios remain difficult to support with current technologies. The review identifies wireless predictability as a potential gap-filler but emphasizes radio uncertainty, implementation complexity, and unresolved standardization and validation challenges.

V. DISCUSSION

  • Industry 4.0 and IT/OT convergence require industrial networks to become more flexible and configurable, with tighter connectivity and more dynamic node membership.Source: §V · p. 9

A. Bridging the Gap between IIoT Scenarios and Network Requirements

  • Wireless requirements for Scenario Classes I and VI are especially difficult: existing standards fall short on latency predictability or multimedia bandwidth, and current wireless networks do not scale sufficiently for high reliability and latency demands.WLAN, particularly WLAN6E, is described as an attractive alternative for bandwidth, latency, timing accuracy, and COTS hardware.Source: A. Bridging the Gap between IIoT… · p. 9
  • Wireless TSN could help address the gap by improving latency guarantees through real-time-aware packet switching, but radio remains inherently unpredictable.Source: A. Bridging the Gap between IIoT… · p. 9
  • Privately managed 5G may contribute to solutions, but its complexity and cost make it impractical for many companies and use cases.Source: A. Bridging the Gap between IIoT… · p. 9
  • Wired real-time networks remain more static and carefully planned than traditional IT networks, and high flexibility and reconfigurability remain largely unsolved despite TSN.Combining TSN with centralized controllers and SDN is presented as a promising avenue for WLAN and 5G, though it adds substantial system complexity.Source: A. Bridging the Gap between IIoT… · p. 9
  • The review identifies missing standards, holistic large-scale implementations, and validation in complex engineered systems as open challenges for safety and reliability.Proposed approaches need integration into system architectures spanning embedded devices, network devices, and compute nodes.Source: A. Bridging the Gap between IIoT… · p. 9

VI. RELATED WORK

Prior surveys address IIoT, wireless real-time communication, and open challenges, while this review compares networking technologies against industrial use-case criteria. Its aim is to overview challenges and solutions from a real-time networking perspective.

VI. RELATED WORK

  • Xu et al. cover IIoT systems without real-time requirements, Kim et al. focus on real-time wireless sensor networks, and Sisinni et al. identify real-time performance as a future challenge.The surveys also cover control, networking, and computing; open challenges include energy efficiency, coexistence, interoperability, and security.Source: §VI · pp. 9–10
  • Park et al. note that wireless interference limits isolation, so control loops must tolerate packet loss and variable delays; joint network-and-control design is a promising approach.The surveyed benefits include easier installation and reduced operating costs.Source: §VI · p. 10
  • Zunino et al. report that existing Industry 4.0 technologies fall short in flexibility, scalability, reliability, real-time behavior, and security.Future networks must provide real-time guarantees to some nodes while supporting other QoS levels for different nodes, sensors, and systems.Source: §VI · p. 10
  • Chen et al. argue that wireless multi-robot manufacturing requires configurable systems and networks that adapt to varying loads, alongside adaptable AGVs.Their discussion centers on flexibility in production type, quantity, and quality.Source: §VI · p. 10

VII. SUMMARY

  • The review identifies common and anticipated industrial-network use cases, derives criteria for their requirements, and reviews established networking technologies, protocols, and current academic work.It compares technologies against the criteria and use cases to survey IIoT real-time networking challenges and solutions.Source: §VII · p. 10

Related papers

Loading 2311.08852v1…