You’ve probably been hearing some talk about Time Sensitive Networking (TSN) over the past year or two but are likely not sure what it does, if products exist with the capability to do it, or how relevant it is to your operations. Based on what I heard at the ODVA 2015 Industry Conference, however, your awareness of TSN will be changing dramatically very soon.
But first, just what is TSN? To understand this concept, it helps to know the purpose of the IEEE Time Sensitive Networking Task Group, which is to “provide the specifications that will allow time-synchronized, low-latency streaming services through 802 networks.” Essentially, TSN is all about equipping industrial Ethernet networks with the level of determinism users came to expect from their legacy fieldbus networks, which has become increasingly difficult with the greater incorporation of AV (audio visual) streaming data on industrial networks. TSN, with its focus on handling AV traffic, is complementary to the Precision Time Protocol (PTP)—an IEEE 1588 standard which, according to Cisco, synchronizes with nanosecond accuracy the real-time clocks of the devices in a network.
Explaining application of TSN in EtherNet/IP networks during the ODVA conference, George Ditzel of Schneider Electric and Paul Didier of Cisco highlighted the current status of the technology.
Currently, you can’t buy any products with TSN capabilities, said Didier, as TSN is “not completely deployed today; it’s still in process of ratification. But you will be seeing products in the next year or two.”
Didier and Ditzel noted that TSN ratification will occur soon, adding that AVnu is working on making the TSN standard “interoperable and certifiable,” thereby making TSN “valuable for the ODVA community as an Ethernet enhancement.”
During their presentation, Didier and Ditzel highlighted three specific aspects of TSN: Time Aware Traffic Shaping, Frame Pre-Emption, and Stream Reservation Protocol.
Time Aware Traffic Shaping references the role of TSN as it works with PTP. Time aware traffic shaping sets up to eight different queues of priority for scheduled and reserved traffic on an Ethernet network, with a time aware gate determining how messages are sent. This means that periodic control and I/O traffic are delivered in specific traffic slots to provide determinism.
Didier added that work is being done now on communication ingress into the network so that devices will have an “awareness of knowing when control traffic should come in and be expecting it. Which means that if it (communication) comes later or earlier than expected, it could be an indication of something wrong with a device on the network or even a security issue.”
Frame Preemption means that, as traffic comes into a device, if an unknown frame arrives that could affect timing, that frame can be partitioned so that portions of it can proceed as needed to keep the scheduled time slot valid and then complete the other portions of the frame later. Another option is to use the guard channel, which sends the unknown frame at a later time without partitioning it. Ditzel explained that there are pros and cons to each option, but each can be a valuable method of managing CIP (Common Internet Protocol) traffic.
Stream Reservation Protocol refers to the establishment of class levels for multiple streams of TSN traffic. For example, you can have a Class A stream for control network communication, which will block outside device communication from being part of the control network’s TSN. “This guarantees control timing,” said Didier. In such a case, a Class B stream could be established to, for example, handle video camera data streams used in quality inspections. Didier and Ditzel explained that some packets could be thrown out in cases of severe network congestion in the same way that Ethernet can drop frames in such instances. But Stream Reservation Protocol ensures that, if you’re adding large streaming apps—such as audio or video—it won't take priority over TSN-based control communications.
Ditzel said this is simply a “reflection of the way Ethernet works now” but added that work is being done now to buffer that data rather than drop it. He added that if your network resources are adequate to handle the two streams, then both classes of communication will get through.
