How time-sensitive networking offers determinism

As work on time-sensitive networks (TSNs) for industrial applications is established in IEC / IEEE 60802 and TSN-compatible products become more and more available, now is a good time to get familiar with it. terminology related to TSN, associated devices, and their configuration.

Michael Zapke, industrial marketing manager at Xilinx (a provider of programmable logic devices) for industrial, vision, health and corporate science operations, offers a list of the technical methods used in TSN. In its list below, the terms at the end of each bullet define the corresponding element in IEEE 802.1Q-2018. All of these methods are essential to using TSN’s industrial automation.

Zapke’s list is as follows:

Centralized configuration

  • Identifies traffic relationships and bridges between them and calculates a schedule—Configuration of the core network
  • Defines the departure and arrival time—Central user configuration

Terminals / (end) Stations

  • All points are synchronized, that is, they operate with the same time of day—Generalized Precision time protocol
  • The network knows to which means of transport belongs which service and is aware of the valid timetable for departures—Flow identification
  • Sequence of components so that their individual schedules are adhered to—Waiting lines

Connections and bridges

  • Knowledge of schedules—Door checklist
  • The network can stop other traffic in an emergency—Preemption and intercalation of traffic
  • The network knows methods and algorithms to control unforeseen peaks—Filtering and policing by flow

“The future IEC / IEEE 60802 standard defines the function selection profile and their sizing for industrial applications,” explains Zapke. “The term ‘TSN-IA’ was introduced for this profile.

Michael Zapke, Industrial Marketing Manager, Xilinx.For a time-sensitive network like TSN, Zapke notes that a common schedule is essential. “The central network configuration contains information about the types of traffic and can retrieve information” from all points in the network (topology discovery), he says. “In addition, the travel times between network elements are known. There is exactly one active core network configuration per network.

A continuous connection between the central network configuration and the bridges exists and allows access to topology information and downloading of updated configurations to the bridges.

Zapke explains that the central user configuration communicates with every station in the network. In TSN, stations are nodes that send and receive units at connections. The central user configuration takes care of sending the right service at the right time, which may include: traffic with a strict arrival deadline, cyclic traffic with latency limits, AVB (audio video bridging) with bandwidth requirements, control traffic with strict priority and reliability requirements, and best-effort traffic that is sent when possible, but can be rejected.

A microcontroller embedded in a station will have a priority service called Generalized Precision Time Protocol. “Two adjacent stations can achieve an accuracy of the order of 10 ns; between all the stations in a network, you should not exceed a few hundred nanoseconds, ”says Zapke.

“Because stations are both a source and a destination of traffic, it is their responsibility to properly label transport units,” he says. “This label is a label in the Ethernet frame (layer 2). The most common is the use of Priority Code Point (VLAN) to identify the type of traffic. In more recent proposals, more options for tagging flows are also addressed so that any pattern in the Ethernet frame can be used (IEEE 802.1CBdb). Beyond that, there are also methods to use Layer 3 (IP packet) with IP intercept to determine the type of traffic.

Flows, queues and bridges

The term used to assign characteristics to a frame is flow identification. “Each image has to be scanned at each station for the label to determine how to handle it at the exit point of the station,” Zapke explains. “The stations sort the Ethernet frames and send the scheduled traffic at exactly the right time to meet the given schedule. “

Up to eight queues parallel to the egress point of a station can contain multiple Ethernet frames. Time-controlled gates allow data to exit the queue at the right time.

“All of this has to be done for each output packet and can result in a high processing load for the station,” Zapke explains. “The use of flow identification, queue management and time controlled gates is normally in a dedicated logic to offload software and increase time accuracy. Such network elements use programmable logic in FPGAs (field programmable gate arrays) or microcode programmable SoCs (systems on a chip) to perform this function.

Connections and bridges are the infrastructure used to connect stations. While the cabling between bridges is static, bridges actively manage the distribution of traffic among multiple connections. The identification of flows is compulsory there as in the stations. At each output port, Ethernet frames must pass time-controlled gates that follow a Door Control List (GCL). The intervals for checking the gates are of the order of a few hundred microseconds to a few milliseconds. Door opening times range from 50 µs to milliseconds.

“An open door for an unnecessarily long period of time reduces network capacity,” Zapke explains. “That’s why bridges normally know exactly when a scheduled frame is coming. Open doors with high timing accuracy are a hardware feature of modern TSN-compatible SoCs, as well as some FPGAs. “

Zapke says TSN is introducing “traffic preemption and interleaving” to “reduce this loss of throughput. Traffic with a lower priority can be cut into smaller fragments so that the guard band becomes very small. This allows traffic with a higher priority to be carried, even if a long frame with a lower priority is already using the connection.

This all works well, Zapke says, as long as there is no overload with high priority traffic on the network. An overload can be the result of an application crash that exceeds its traffic limits. To compensate for this, “flow filtering and policing,” which can apply metrics and filter traffic based on policies. “This is a network security feature that keeps the network operational,” Zapke explains.

Source link