Content for  TS 22.104  Word version:  19.1.0

In the closed-loop control use case for process automation, several sensors are installed in a plant and each sensor performs continuous measurements. The measurement data are transported to a controller, which takes decision to set actuators. The latency and determinism in this use case are crucial. This use case has very stringent requirements in terms of latency and service availability. The required service area is usually bigger than for motion control use cases. Interaction with the public network (e.g., service continuity, roaming) is not required.

  Several sensors are installed in a plant and each sensor performs continuous measurements. The measurement data are transported to a controller, which takes decision to set actuators.

For process and asset monitoring in the area of process automation, a large number of industrial wireless sensors are installed in the plant to give insight into process and environmental conditions, asset health and inventory of material. The data are transported to displays for observation and/or to databases for registration and data analysis Examples of sensors are temperature, pressure or flow rate sensors for process monitoring, vibration sensors for health monitoring of e.g. motors, or thermal cameras to detect leakages. Industrial wireless sensors are typically constrained in terms of size, complexity and/or power consumption. The operation for this use case can be in a wide service area, and interaction with the public network (e.g., service continuity, roaming) may be required.

  Sensors generating periodic measurements of a continuous value (e.g. temperature, pressure, flow rate sensors). The traffic is predominantly mobile originated. Sensors generating waveform measurements (e.g. vibration sensors). Even though the waveform measurement is continuous, it is expected that this type of sensors will buffer and transmit the data periodically (e.g. every second) to save battery by enabling discontinuous transmission. The traffic is predominantly mobile originated. Cameras (regular or thermal) for asset monitoring (e.g. for leakage detection). Even though the video recording is continuous, it is expected that this type of sensors will buffer and transmit the data periodically (e.g. every second) to save battery by enabling discontinuous transmission. The traffic is predominantly mobile originated.

To keep a plant running, it is essential that the assets, such as pumps, valves, heaters, instruments, etc., are maintained. Timely recognition of any degradation and continuous self-diagnosis of components are used to support and plan maintenance. Remote software updates enhance and adapt the components to changing conditions and advances in technology. The operation for this use case can be in a wide service area, and interaction with the public network (e.g., service continuity, roaming) may be required. In this use case, the assets themselves are assumed to be connected to the 5G system. The use case where sensors are used to monitor assets is covered in clause A.2.3.2.

  To keep a plant running, it is essential that the assets, such as pumps, valves, heaters, and instruments are maintained. Timely recognition of any degradation and continuous self-diagnosis of components are used to support and plan maintenance. Remote software updates enhance and adapt the components to changing conditions and advances in technology.

An Edge Computing use case example: Digital twin based production quality maintenance: A digital twin is a virtual representation of a product or production systems. Digital twins are used to simulate, predict and optimize products and production systems. In future, digital twin production control system based on augmented reality based will be used in the factories. In this usecase, a digital twin's digital and virtual model of a function combined with other physical data to simulate real-time aspects of how a system operates. A digital twin production control system can be automated using machine data and the AI/ML trained data after applying the AI/ML algorithm and further processing. The processed output can be translated to a control command back to the device by a control function running on the edge cloud. In another case, using telemetry data as input, the digital twin model's output may be fed to an AR server for sending low latency AR streams toward the manual operator in the factory production area. At the same time, it can further be utilized as input by an AI/ML model. A process control function can compare the machine data (example: position, rotation level, speed, sensor data, high-speed photography etc.) and perhaps a high-resolution video from the manufacturing line and if necessary, it can send commands for corrective measure. In this example, the process control functions reside at the edge infrastructure and the inspection related corrective input is sent back to the production system control function. Corrective actions/commands for misalignments from the processed output can be sent in two ways:

1. Manual process with AR server: In this case the service performance requirements should follow the Table A.2.4.2-1.
2. Automatic process with AI/ML: This usecase and service performance requirements are described below in Table A.2.3.4.

The general high-level service requirements of the edge computing usecase for the digital twin based production inspection:

• High bandwidth wireless data for both uplink and downlink
  • exact number depends on video encoding, frame rate and video-resolution requirements
• Timing accuracy and low latency ( ≤20ms)
• High availability of the communication network
• Security requirements: Data encryption, connection authentication, user authorization
• QoS methods to ensure quality of service performance over different UE to Application connection sessions (video streaming, sensor data, control data)
• UE Mobility and session continuity (optional)

  Use case one The periodic telemetry data and video images are used from the digital twin in the production system for analysis and then the processed outcome is sent back to the system for any adjustment of the machine components. The following diagram explains the above digital twin usecase steps to manage the production in a factory (both manual and automated operations)

• Low-latency AR overlays and incorporation of AI/ML techniques to identify manufacturing issues and improve product quality as well as to enable offline adjustments for optimization, adaptations, and preventive operations on the machines

Control panels are crucial devices for the interaction between people and production machinery as well as for the interaction with moving devices. These panels are mainly used for configuring, monitoring, debugging, controlling and maintaining machines, robots, cranes or complete production lines. In addition to that, (safety) control panels are typically equipped with an emergency stop button and an enabling device, which an operator can use in case of a safety event in order to avoid damage to humans or machinery. When the emergency stop button is pushed, the controlled equipment immediately comes to a safe stationary position. Likewise, if a machine, robot, etc. is operated in the so-called special 'enabling device mode', the operator manually keeps the enabling device switch in a special stationary position. If the operator pushes this switch too much or releases it, the controlled equipment immediately comes to a safe stationary position as well. This way, it can be ensured that the hand(s) of the operator are on the panel (and not under a moulding press, for example), and that the operator does—for instance—not suffer from any electric shock or the like. A common use case for this 'enabling device mode' is the installation, testing or maintenance of a machine, during which other safety mechanisms (such as a safety fence) are deactivated. Due to the criticality of these safety functions, safety control panels currently have mostly a wire-bound connection to the equipment they control. In consequence, there tend to be many such panels for the many machines and production units that typically can be found in a factory. With an ultra-reliable low-latency wireless link, it would be possible to connect such mobile control panels with safety functions wirelessly. This would lead to a higher usability and would allow for the flexible and easy re-use of panels for controlling different machines. The cycle times of the control application depends on the process/machinery/equipment whose safety has to be ensured. For a fast-moving robot, for example, end-to-end latencies are lower than for slowly moving linear actuators.

  Periodic, bi-directional communication for remote control. Examples for controlled units: assembly robots; milling machines. Aperiodic data transmission in parallel to remote control (use case one). Periodic, bi-directional communication for remote control. Examples for controlled units: mobile cranes, mobile pumps, fixed portal cranes.

Operation and monitoring of machines, mobile robots, or production units via a mobile operation panel provides higher flexibility and comfort for human operators. A single mobile operation panel can be used to manage more than one production system due to its mobility in the factory. The mobile operation panel provides relevant information for configuration, control of industrial machines as well as monitoring of relevant data generated during the construction of a product. The monitoring data is generally considered to be less time-critical subsequently requiring non-real-time communication. On the other hand, the mobile operation panel supports safety-critical functions such as emergency stops or enabling or changing the position of robots and other machines. These functions are generally considered to have strict ultra-low latencies and reliable transmission requirements that must follow strict safety standards making them time-critical (real-time communication).

  Emergency Stop with periodic-deterministic communication for connectivity availability and aperiodic-deterministic communication for emergency stop events. Safety data stream with periodic deterministic communication. Visualization of Control with periodic deterministic communication. Motion Control with periodic deterministic communication. Haptic feedback data stream with periodic deterministic communication. Manufacturing data stream with mixed traffic.

It is envisioned that in future smart factories and production facilities, people will continue to play an important and substantial role. However, due to the envisaged high flexibility and versatility of the Factories of the Future, shop floor workers should be optimally supported in getting quickly prepared for new tasks and activities and in ensuring smooth operations in an efficient and ergonomic manner. To this end, augmented reality may play a crucial role, for example for the following applications:

• monitoring of processes and production flows;
• step-by-step instructions for specific tasks, for example in manual assembly workplaces;
• ad hoc support from a remote expert, for example for maintenance or service tasks.

In this respect, especially head-mounted augmented-reality devices with see-through display are very attractive since they allow for a maximum degree of ergonomics, flexibility and mobility and leave the hands of workers free for other tasks. However, if such augmented-reality devices are worn for a longer period (e.g., one work shift), these devices have to be lightweight and highly energy-efficient while at the same time they should not become very warm. A very promising approach is to offload complex (e.g., video) processing tasks to the network (e.g., an edge cloud) and to reduce the augmented-reality head-mounted device's functionality. This has the additional benefit that the augmented-reality application may have easy access to different context information (e.g., information about the environment, production machinery, the current link state, etc.) if executed in the network.

  Bi-directional message transmission between an augmented-reality device and an image processing server.

In factories of the future, there are needs to perform remote access and maintenance to devices and entities, for instance, by remote control centres. The devices and entities might be installed at geographically distributed locations. These devices typically have firmware/software which needs to be updated occasionally. Maintenance information also needs to be collected and distributed from/to these devices periodically. The devices can be both stationary and mobile. Device maintenance may happen in parallel to the actual production process and other communication services performed at the device side without any negative impact on these production communication services.

  Transmission of non-deterministic messages in parallel to other interactions. Example applications: software/firmware updates and exchange of maintenance information.