Page 26

Industrial Ethernet Book 105

Technology Application layer OPC UA is employed on the application layer, including support for the Client/Server and Publish/Subscribe communication models. OPC UA servers on all devices should support the Embedded Server Profile. For very resourceconstrained devices, only a publisher feature for providing data and a TCB client for network configuration can be utilized. Client/Server: Communication model used for device configuration, browsing the information model, registering e.g. for diagnostic information. For secure applications, the device configuration shall provide data integrity (signature) and optional confidentiality (encryption). Publ i sher /Subs c r iber Pub/Sub) : Communication model for cyclic transmission. Optionally signed and/or encrypted using OPC UA message-based security. A header profile with static dataset offsets can be used for efficient dataset extraction in end stations. Additionally required features The ISO/OSI reference model provides a quick overview of the protocol stacks involved in OPC UA TSN technology. To satisfy the requirements of industrial communication systems, however, the following additional features are needed: Device roles New features are required to orchestrate booting and operation of a network of OPC UA TSN devices. The roles are (almost) independent of the system hardware. State machines: End stations in an industrial network must have uniform behavior defined according to a state machine. This makes it possible for a central instance (i.e. a network managing node) to orchestrate the behavior of the entire network. Many industrial Ethernet solutions implement a state machine based on the ideas of CiA. Topology detection: Scheduling of real-time traffic requires detailed knowledge about the topology of the network. Topologies can be detected (using LLDP29) and imported or created offline in a configuration tool. The CNC uses this information to compute the configurations for Qbv and Qav. Cut-through switching: The cycle-time performance that can be achieved on a switched network depends heavily on the latency of frame transmission. In particular, long line or ring topologies pose challenges. Thus, cut-through switching (forwarding a frame as soon as the address information has been decoded) constitutes an indispensable feature of 3-port switches in field devices. Device profiles: In industrial communication systems, interoperability needs to be ensured on each OSI layer. The lowest layer that violates interoperability constitutes the highest layer for the interoperability of the entire system, independently of any higher Centralized User Configuration User/Network Configuration Info RESTCONF OPC UA OPC UA Centralized Network Configuration Management NETCONF over TLS Listeners Talkers The fully centralized model of Qcc (with OPC UA applications). layers. Legacy Industrial Ethernet systems share only the same physical media. This fact has caused a lot of customer dissatisfaction, because the original marketing message was that Ethernet is Ethernet, so they all should be compatible. To prevent OPC UA TSN technology from falling into the same trap, the goal is to use common implementations of all seven OSI layers (for communication between devices) and moreover to have both a standard device profile and type-specific device profiles. Today, standardized profiles for safety, drives, IO and controller to controller communication are under consideration. Device description files In the realm of OPC UA, a device is represented by its server instance, whose features can be browsed online “at any time.” While online browsing suffices in some industrial use cases, those with a high degree of repetition, such as serial machine building, require an offline method for configuring and programming devices. Hence, all relevant features of a device (OPC UA, application and networking capabilities) need to be described in files, substituting online access to the device. Configuration and boot-up Almost all fieldbus systems existing today, based on real-time Ethernet or not, provide mechanisms for network management. These mechanisms do things like boot a network device by transitioning it through a series of states into an operational state; enable a device to detect, handle and signal errors during runtime; or implement procedures necessary to replace faulty devices. States and state-transitions comprise functions such as network device identification (ensuring that the device can be reached on the network, matches the expected vendor/ model, etc.). They are also used to perform any necessary configuration/firmware updates and subsequently notify the device to transmit valid process data (if the application on the device is ready to do so) and evaluate received process data (if a central network instance controlling the network decides to do so). Many existing implementations of network management in the various fieldbus systems combine all of this functionality in one device (i.e. the PLC). The explicit goal in this work is to logically separate and decouple these functions into so-called device roles, such that each could theoretically be implemented on a different device within the network. Multiinstance and device-role redundancy shall be addressed as well. Role management For machine networks, a number of network functions are required in order to reach defined states in the network during start-up and operation. Those functions can be grouped and allocated to device roles. The following is a list of well-known device roles for IT and OT systems as well as new ones for OPC UA TSN. The section is concluded with a list of user roles for developing and running the network. Currently required device roles TSN switches: They constitute the network infrastructure of an OPC UA TSN network. Multi-port switches are used for setting up the network topology from a bird’s view, while switches with two external (and one internal) port reside in switched end stations to allow for efficient cabling in a line topology. The state machine of a switch adds states to prevent message storms in case of loops in the network. DHCP (server): DHCP is a mechanism to allocate IP addresses from a pool and assign them to unconfigured devices. Furthermore, most DHCP server implementations allow static binding between Layer 2 MAC addresses and Layer 3 IP addresses. The combination of these features makes it possible to boot unconfigured devices (with unknown MAC address) using a temporary IP address and, after successful identification (and probably authentication), 26 industrial ethernet book 4.2018


Industrial Ethernet Book 105
To see the actual publication please follow the link above