Ubiquitous data communications – synchronization & aggregation

Posted on December 18, 2009 by Anthony Magee

Pylon and sun

Image by vixstarplus via Flickr

I write this, my first blog, while sitting in a presentation on the ‘Smart Grid’ project which is underway in the US. This project aims to equip the domestic and commercial power grid in the US with tools to support two way power distribution (from and to utility companies), aid power monitoring via distributed meter reading and promote energy conservation. The plan is to deploy IPv6 to millions of Smart Grid devices over the next 3-4 years. Ethernet is acknowledged as a key component of the Smart Grid concept.

Recently I attended the International Telecommunications Synchronization Forum (ITSF) in Rome. A key topic of debate at that meeting was the use of Femtocells in the mobile communications market. Some proposals at ITSF suggested there would be billions of end devices eventually in the network many of which would require synchronization albeit to various amounts of accuracy.

Factory and industrial automation, intelligent traffic management systems, health care monitoring in the home, and mobile backhaul are other examples which add to the synchronization requirements.

So what are the common threads?

There is a clear trend towards multiple IP/Ethernet enabled devices at the edge of the network. Scalability, bandwidth and port aggregation are key requirements. In addition, many devices are likely to require some form of time alignment to aid in transmission without causing interference with neighbouring devices and also to allow correlation of events that occur at the same time. Time synchronization and frequency alignment are therefore key requirements.

The question is how accurate do the end devices have to be, and how will the network infrastructure evolve to support such wide scale deployment?

In my view Synchronous Ethernet, IEEE 1588v2 Precision Time Protocol are key technologies in this area. Accuracy requirements I’m aware of, start at about 1us (micro second) these days for mobile backhaul, but where this will end up is anybody’s guess.  Packet aggregation is also expected to be an enabling technology for ubiquitous data communications for applications such as mobile backhaul and Smart Grid, but what are the scalability and security requirements?

Please add comments on your thoughts or come back and follow my blog as I explore this topic as well as other Carrier Ethernet subjects in future…

Tags
  • Francesca

    Wouldn’t Time Triggered Ethernet be better than 1588v2 for low latency and determinism?

  • Anthony Magee

    Hi Francesca – many thanks for your question on this.  I’m not working on Time Triggered Ethernet regularly, and therefore especially given the nature of Time Triggered Ethernets use applications, would like to caveat my response – please take a second opinion on the appropriate technology for your application.

    Time Triggered Ethernet is work ongoing within the SAE - Society of Automotive Engineers, and is intended to support a deterministic and low latency path through a network, where the maximum delay experienced by a particular packet can be controlled depending on the message type and associated priority.

    An application area of this could be in safety critical systems such as vehicle control and embedded communications in the avionic indsutry.

    1588v2 on the other hand is intended to align the time of day of a Slave clock to a Master clock when the two nodes are connected together over a network which may be running Ethernet and IP and where delay variation can vary – often depending on traffic volume in the network.  1588v2 is used to measure the delay experienced by packets in a network, where the delay could be variable (although symmetrical delay is assumed) and adjust the Slave Time accordingly.

    There are many networks which require accurate time alignment but which do not need to reserve credits within the network in a similar manner to that acheived via TTE.

    In summary, if you need time of day alignment over a network, then IEEE 1588v2 is a good candidate, if you need determistic delay then you could consider something such as Time Triggered Ethernet.  However, there is an interesting article I found which proposes to make use of 1588v2 as a tool for aiding Time Triggered Ethernet and which may provide further insight into this subject area.

    http://www.emeraldinsight.com/Insight/viewContentItem.do?contentType=Article&hdAction=lnkpdf&contentId=874928

    I hope you find this response useful.

    Anthony