Five Minute Facts About Packet Timing
By Doug Arnold
Our next update on what’s in the 2019 edition of IEEE 1588 , delves into the mysterious sounding topic of special ports. A special port happens when a PTP network interfaces with a transport media which has a built in timing mechanism. Two examples are Wi-Fi and Ethernet Passive Optical Networks, or EPON. The idea is this: when transferring time over a time aware transport media, we want to use the built in timing mechanism, rather than sending Sync messages, and other PTP timing messages. So why do these networks get special treatment? The reasoning is that implementations of the native timing mechanism will likely include hardware time stamping at the Media Independent Interface (MII). Timestamping at the MII may not be available for a PTP messages which traverse these transport media. So just sending PTP timing messages over such networks might be less precise than using the built in time transfer of the transport media.
First why do transport media have built in timing? In the case of Wi-Fi time transfer is used to save energy. No, really! The Wi-Fi standard anticipates battery powered devices that are lazy, and want to sleep most of the time. The more accurate the device’s alarm clock, relative to the Wi-Fi access point, the later it can wake up and still be sure it receives the next update from the Wi-Fi access point.
EPON devices are typically not battery powered. They have another reason for time synchronization. This is illustrated in Figure 1. In this example three end devices, called Optical Network Units, or ONUs, all share a link with the Optical Line Terminator, or OLT.
For example, in a fiber to the home telecommunications network, the ONUs are built into the modem/routers, and the OLT is a neighborhood concentrator. Since the spliter is passive, with no queuing, the ONUs must take turns on uplink transmission to avoid talking over each other. This is accomplished using time division multiplexing (TDMA), i.e. each ONU gets a time slot for its uplink. Clearly, all of the ONUs must have the same sense of time to avoid collisions.
So now we know why, but how does PTP interface to a time aware transport media? A boundary clock can include both standard PTP ports, and special ports. This is shown in the example depicted in Figure 2.
In this example, a PTP network feeds time to an EPON network, by way of a Boundary clock which includes a special port. This way the EPON network has time in the timescale of the Grandmaster, not just relative to the OLT. It is also possible for an EPON network to supply time to a PTP network, or even for a Wi-Fi link to be in the middle of a PTP network. When the time aware transport media is in the middle of the PTP netwtork, Announce message are sent through the special ports to facilitate the BMCA in on the other side of the time aware transport media.
Here is what happens at a special port. The protocol stack includes a special port adapter layer, as depicted in Figure 3.
When timing is sent from PTP to a time aware transport media, timing information is translated from the way it is organized in PTP, via Sync messages, and Delay Request/Response messages or Peer delay messages, to the format of the transport media. This will be different for Wi-Fi than it is for EPON, or any other time aware transport media. When time goes from the transport media to PTP, all of the information needed construct PTP timing messages must be pulled from the media. Generally this means a timestamp, and the delay since that timestamp was created.
So far there has not been a need to create PTP special port adaptors for any transport technology other than EPON and Wi-Fi, but the next edition of IEEE 1588 makes PTP ready for this when the need arises.
If you have any questions about PTP, don’t hesitate to send me an email at doug.arnold@meinberg-usa.com or visit our website at www.meinbergglobal.com.
If you enjoyed this post, or have any questions left, feel free to leave a comment or question below.