HomePNA Blog

Gigle First to Claim the Really High Ground

By Rich Nesin, General Manager and Resident Philosopher, HomePNA

You’ve probably heard that Gigle Semiconductor has claimed the high ground; launching the “World's First Gigabit Home Networking Chip”. No doubt about it, they are good. So what’s Gigle’s all-important user throughput. I don’t know BUT I heard that Belkin adapters, the first products using the chip, are available so we’ll know soon. We’ll also know more about Gigle’s proprietary “mediaxtream”, the technology giving them the boost over Homeplug AV data rates.

The big question is whether Gigle will be able to convince large service providers to use proprietary technology (especially with all the service provider interest in ITU G.hn’s Gigabit standard). If they want it to be selected they really need an open standard. I remember way-back-when that Gigle proposed something to G.hn but it wasn’t included in the G.9960 PHY standard.  There are other standards groups in the world so if they are willing then who knows.

On to the small print. Gigle claims speeds up to 1000Mbps but there’s an asterisk and a note that 1000Mbps is the “ideal physical data rate” and actual data throughput and distance will be lower, depending on interference, network traffic, building materials, and other conditions”.

OK, we all play marketing games. That’s why I use phrases like “up to” when I write about data rates. It’s not my fault. Editors and analysts make me do it. Speed sells and it’s a lot easier than trying to explain user throughput. So … what it means is that the chip produces Gigabit speeds in the lab and maybe on a tester. In the real world you start with the real data rate which is pretty much guaranteed to be lower than a Gigabit and then start subtracting off the overhead.

The overhead depends on the encoding technology – the ways the digital data bits are “encoded” for transmission. Usually some user data bits are encoded into more bits for transmission on the media; be it phone wire, coax, powerline, Ethernet cable or air. There are other transmission taxes related to the way the technology deals with interference and other real world nasties. And then you subtract the MAC overhead – the gaps, address fields, error correction etc. Some MACs (like HomePNA 3.1) are very efficient. Others aren’t. What you have left is the real data throughput. And if you are lucky it will be stable. If you are unlucky it will vary all over the place depending on interference etc.

So the devil is in the details.  But this is old news – it’s the reason that major North American service providers prefer coax and phone wires over wireless Wi-Fi and powerline for distributing triple play services in the home. And while we’re on the subject, the ITU G.hn is meeting this week. What’s on the agenda? Cleaning up the draft and converging toward final consent in Geneva. Gigle is first but that may or may not be important in the long run.

Broadband and IP/MPLS Forums Get Married

By Rich Nesin, HomePNA general manager and resident philosopher.

If you use the Internet, and who reading this blog doesn’t, then you are using both the core and access networks (and you may even be using a home network). Five years ago it was two networks now it’s three -- home networks have gained respect in the telecommunications world.  The core is a very high speed mostly fiber central communication network that carries data from access provider to access provider. The access network connects users to the core network and the home network (my favorite!) connects devices in the home to the access network.

All of these networks are based on accessible practices and technologies of some kind; standards from standards groups such as the ITU or specifications and recommendations from industry groups such as the Broadband Forum and the IP/MPLS Forum.  Industry groups often offer certification testing to insure that products perform correctly (which is why, for example, it's so important to only use HomePNA products that have passed HomePNA certification testing, DSL modems that have passed Broadband Forum testing, etc.).

So where is this blog going?  If you're involved in telecommunications you probably heard that the Broadband Forum merged with the IP/MPLS Forum. If you haven’t heard, this is a good thing. There are a lot of industry and standards groups and they develop a lot of documents. Sometimes different groups will develop their own practices for the same application. A lot of effort goes into developing liaisons between the groups so they can share information freely. Sometimes the collaborations work and sometimes they don’t.

In this case there was a lot of interest and membership overlap between the IP/MPLS Forum and the much larger Broadband Forum. The standards world is a small one and you tend to see the same people attending meetings for different groups so merging them makes a lot of sense. Why it’s a good thing …. well that’s another story.

IEEE P1901 – Is the Opera Truly Over?

By Rich Nesin, General Manager and Resident Philosopher, HomePNA

Well the season finale of my favorite soap, how the IEEE turns, has aired. And what a season it was! Now it’s time for everyone to live happily ever after (at least they would if I wrote the script, three years to get to a draft spec is enough).

If you’re familiar with how the IEEE developed the elegant two PHY P1901 proposal, then you may not be surprised that the approved version includes not two but three PHYs. That’s right. Despite defeating a previous bid to use the ITU’s G.hn PHY for P1901, the final compromise proposal includes G.hn in addition to Intellon’s HomePlug AV Panasonic’s HD-PLC PHYs.

The inside story is that the P1901 proposals squeaked by and may well have been headed for another defeat. The addition of the G.hn PHY saved the day by drawing new supporters (including G.hn pioneer Coppergate Communications who voted yes for the first time).

So this is great for everyone, right, a real win-win? I certainly hope so but there are nasty rumors circulating about it being a political ruse and behind the scenes efforts taking place to remove G.hn from the IEEE P1901 draft standard. The industry took a giant step forward when G.hn was added to P1901 and now it’s time for the IEEE to walk the walk and do the right thing for everyone.

ITU G.hn Delivers a Spec

By Rich Nesin, General Manager and Resident Philosopher, HomePNA

Have you noticed the press coverage that high performance home networking over existing wires is getting? It makes me nostalgic for the days when analysts dismissed us with a quick ‘the world is going wireless”. It probably is, but installations in millions of IPTV homes over the last couple years say that it isn’t always the best solution. Last week’s ITU announcement about completing the first G.hn standard seems to have gotten the most coverage yet (possibly due to how extraordinarily rare it is for a standards group to deliver a standard on time). So you may be wondering what actually happened and how?

If you have been following my ramblings about home networking standardization in the IEEE P1901; and ITU Q4/15 you may already have a good idea of what happened. If you’re not familiar with the ITU, standards can only be approved during special ITU “plenary” meetings. A plenary is a formal two week meeting of the entire SG 15 (not just the Q4/15 group that is part of SG 15 or the even smaller G.hn group that is part of Q4/15 and has been meeting every couple weeks for the last year to get the work done on time). Like the other meetings, the plenary was very technical with lots of informal working meetings during the evenings to refine the contributions and achieve consensus on the contributions (or in the words of one engineer, “lots of real work was done”). This is the “how”. This and a lot of hands-on direction from some of the world’s largest telcos.  Being the customer has its advantages and these customers see the advantages of a single open world-wide existing-wire home networking standard.

So what happened? I wrote a little about MACs and PHYs back in a June post about HomePNA's guaranteed QoS. Like most other home networking technologies, G.hn is built of two components; a MAC and a PHY. The MAC part, short for medium access control, manages the communication between devices on the home network. The PHY part, short for physical interface, encodes and decodes the data from the MAC – encoding it so that it can be carried efficiently over the media (which in the case of G.hn can be powerlines, phone wires or coax) and decoding the data that is received from the media so it can be processed by the MAC. Every product featuring a G.hn interface will have a MAC and PHY. So what is G.hn’s new G.9960 standard? It’s the new G.hn PHY. The companion G.hn MAC spec is scheduled to be completed in September.

More soon …

Powerline Home Networking Update - The IEEE P1901 Work Creeps On

By Rich Nesin, General Manager, HomePNA

Last summer I attended an IEEE P1901 standards meeting - the group standardizing home networking and broadband access over powerlines - and was inspired to post a couple blogs about the experience.

Well, half a year has passed and if you haven’t been following the play by play action you haven’t missed much.  If you recall, the current P1901 “compromise” proposals (there are 3) came about after HomePlug and HD-PLC supporters broke the three-way deadlock by combining their proposals at the expense of UPA. The new proposal, constructed by gluing together large chunks of HomePlug AV and HD-PLC along with some new “coexistence” circuitry, will allow them to coexist with HD-PLC and HomePlug AV as well as with other existing powerline home networking technologies (or at least the existing technologies that implement the new coexistence mechanism). It will coexist with some technologies – and you can guess which ones -- better than with others.  At least it will if they can ever get past the proposal stage and start working on the actual draft standard.

Anyway, six months later and the drama continues.  The joint proposal was voted down at the Madrid meeting which, in theory, should've sent everyone back to their drawing boards to draft a new proposal.  That’s the theory.  In fact what happened was that the same compromise proposal is being voted on again.

Now, and this is hearsay (but juicy hearsay), the rumor is that a certain party has been very active adding voting supporters to P1901. Some of these supporters, the rumor goes, have little to do with home networking or access over powerlines.  As we discussed in the past, a new member must attend a couple of P1901 meetings to vote on a proposal so this is where the rumor gets juicier.  The proposal supporters have asked for a telephone conference on December 9th, before the next face to face meeting in Japan on December 16th, to review the “no” comments made subsequent to the last meeting in San Francisco.  Speculation is that they have requested that this teleconference be considered a working group meeting so that it counts toward the new members’ voting qualification. 

Wouldn’t it be better if we all just tried to get along? 

What the Big Dig and Chunnel can teach us about IEEE P1901 and ITU G.hn

By Richard Nesin, president, HomePNA

As can probably guess we spend a lot of time discussing P1901 and G.hn. My colleague Michael Weissman, a source of really good marketing anecdotes, has a favorite one about the time he heard marketing guru Clayton Christensen, Harvard Business School professor and author of “The Innovator’s Dilemma”, speak at conference. The talk compared two major construction projects, Boston’s “Big Dig” and the “Chunnel” under the English Channel.

The story goes that the Chunnel took about 7 years to complete, is about 30 miles long, and cost around $15B in today’s dollars or $500M/mile. The Big Dig took about 15 years to complete, spans 3.5 miles (about half in tunnels), and cost around $15B in today’s dollars or about $4.3B/mile. As Christensen tells it, the moral of the story is that infrastructure made the difference. The Big Dig had to replace and connect to a very complex and heavily traveled transportation infrastructure. The Chunnel was green field – nothing existed previously so it was all new construction.

A great anecdote but what does it have to do with P1901 or G.hn? When G.hn began it was understood and agreed that no attempt would be made to make it interoperable with existing standards. This was painful for HomePNA members since we have always successfully standardized our technology under the ITU but we agreed for the greater good. G.hn is a green field standard, free to select the best and newest technology available. IEEE P1901 started with several existing standards (the infrastructure) and has evolved into an awkward multi-MAC, multi-PHY architecture with several implementation options. The result is slow progress and the probability that a consumer will buy two IEEE P1901 devices that won’t be any faster than today’s products or work together. We’re not sure that is progress.

Comparing IEEE 1901 and ITU G.hn

By Richard Nesin, president, HomePNA

I just returned from my first IEEE P1901 meeting. If you follow home networking (and who doesn’t?) you probably recognize the IEEE as the group that developed and continues to evolve the Wi-Fi 802.11 and Ethernet 802.3 standards. They do a lot of other things and some of those things are being done in P1901. The P1901 group was formed a couple of years ago to standardize networking over powerlines. The group started with three non-compatible industry technologies: HomePlug, UPA and HD-PLC.  According to the project authorization request or “PAR” approved by the IEEE, P1901 is working on a standard to enable coexistence and interoperability over powerlines for both home networking and broadband over powerline (BPL) applications.

The meeting was well attended with representatives of over 40 “entities” present. I counted at least one copy of Robert’s Rules of Order for every three attendees; a very respectable ratio. (An exaggeration but not far from the truth). The meeting was very formal and Robert’s Rules were quoted frequently.  There were several conflicting agendas being pursued and a good deal of the meeting was spent discussing what motions were and were not allowed.  Almost all of the technical work seems to have been done behind the scenes. Having attended G.hn meetings, where most of the time is spent presenting and discussing technical contributions, this surprised me.

I was also struck by the very different policies and procedures followed by the IEEE P1901 and ITU-T G.hn – or for that matter between P1901 and other IEEE groups. Unlike some IEEE groups, P1901 only allows one vote per entity (which is a good thing).  An entity can be a company, industry SIG, university, etc. As discussed back in the March blog about the differences between Special Interest Groups or SIGs and standards organizations like the ITU-T and IEEE, SIGs are usually run by the “privileged” members.  Allowing SIGs equal voting rights can provide companies or groups of companies with additional influence on votes potentially making the process more political. The ITU-T is more selective and doesn’t allow SIGs to be members. Now that you know more than you ever wanted to know about entities, we’ll just call them members.

Members must join the P1901 group and attend a number of meetings to be entitled to vote (they lose voting privileges if they miss too many consecutive meetings). It’s not possible for a company to send a lot of employees to a given meeting to force through a favorite motion (although I’ve seen this done in other IEEE groups operating under different rules).

As far as the work itself, it’s taken a while and will take a while longer. The current voting isn’t on a draft standard, it’s on three individual “clusters”; one for home networking called “IH” for in home, one for BPL called access and one for coexistence called coexistence. The clusters must each be approved by over 75% of the groups members. They will then be combined into one working document that follows the IEEE style rules and voted on again. After 75% of the members approve, the document will be considered mature enough to be a draft standard. Another 75% vote will put the draft before the sponsor (ie the IEEE Communication Society) where it will undergo an new iterative review and approval voting process by that group per their own policies and procedures to make sure it meets the PAR objectives and IEEE requirements before it is released as a standard. As you know if you have been following this blog, ITU-T is a consensus process.  When a draft standard achieves consensus, it is released. 

Two standards groups, two very different operating modes.

QoS part 2: What is Priority QoS

By Richard Nesin, president, HomePNA

In the June 4th post QoS part 1 we jumped into the middle of a discussion on Quality of Service (QoS). If you recall, QoS is a mechanism used to provide reliable and predictable delivery of real-time data such as IPTV and VoIP streams over a home network that is carrying other “best-effort” data like porn (you are enabling those parental controls, right?). To techies, home network Quality of Service (QoS) used to means priority-based QoS, the original QoS provided by Wi-Fi home networks.

So where does the “priority” in Priority QoS come from? It comes from the IEEE

Data sent over a network is broken into same-size pieces called “packets,” which are like the way the words in a book are broken into pages. Information is attached to each packet to tell the network equipment things like the size of the packet, where the packet comes from, and where it’s going – this is called the packet “header.”

Header information helps network equipment to route the packet from its source, say a server at You Tube, to its destination, say your PC. Included in this header are three “priority bits.” (You’ve probably guessed where this is going, right?) Three priority bits enable eight levels of priority although sometimes only four are used. Real-time data such as video is usually given a high priority and non-real-time “best-effort data” is given low priority. Network equipment uses various techniques to insure the delivery of high-priority data over low priority. Priority QoS, also called Class of Service (CoS), will prioritize one type of data over another but typically won’t help when two same-priority data streams are being sent simultaneously.