HomeGrid President John Egan was in ebullient mood on his return from Computex and a Chinese tour, this week, claiming design wins for G.hn, mounting interest from a wider community and shipments just beginning for first deployments.
He was chipper enough to fire warning shots at MoCA and virtually everyone that has doubted that G.hn was going to make it as a technology.
G.hn is one of the few areas that Faultline itself has been on the fence over, but we had our eyes opened back in February when Marvell (Egan works for Marvell too) showed its G.Now version of G.hn working at Korea Telecom. Here it will be used to offer a 500 Mbps broadband service over twisted pair that is installed around Korea MDUs, working with Korea equipment supplier Ubiquoss, a leading supplier to KT Corporation (formerly Korea Telecom). KT has some 18 million captive phone homes and many of them are based in MDUs. It has some 8 million broadband connections in these homes and 4.7 million IPTV customers, almost as many as AT&T.
G.hn was designed as a ‘no new wires’ home network, and was never meant to be used as an access chip in its own right, but this means that it now has possibilities in this arena, and we picked out the use case for G.Fast as a potential target for G.hn chips.
Egan told us that at this KT installation ‘Orders were firming up and that it will be a good sized deployment. Meanwhile we are looking at another huge deployment in Asia for this MDU technology.’
That huge deployment is largely thought to be at China Telecom, which we told you about in June last year, which has made it clear that G.hn would be its home networking installation of choice. It has had RFPs that cite G.hn as a core technology for years, but is still going through its testing cycle and will then choose suppliers and finally haggle over pricing with those suppliers, before anyone can claim the Lion’s share of that particular win. China Telecom has 172 million access subscribers, and adds 3 million broadband subs each quarter, so it could soon amount to a considerable number of chips and we suspect that the companies that win part of that order will get into profit on their G.hn chips almost immediately.
Egan said that in KT’s case, ‘The G.Now chips were simply dropped into a VDSL2 head end at KT and they worked virtually straight off the bat. And we demonstrated at Computex these interoperating with CPE which use other company’s G.hn chips.’
This whole issue of working with and yet competing with the other silicon vendors is central to G.hn’s success, says Egan, ‘Marvell initially wanted us to have the G.Now chips only work with other Marvell chips, but I convinced them that this was not the way to go. It is important at this stage that we open up opportunities for the other chip vendors ‘ Sigma, Metanoia and Xingtera ‘ so that none of the operators feel that they are tied in to a proprietary technology.’
He added that, ‘HiSilicon has given us permission to say that it has started working on a G.hn chip and that samples will begin in a few months, and that Triductor, also in China, is also beginning the process and would soon join the existing suppliers.’ Egan also said that he expected MStar, the leading global smart TV chip maker in the world to ‘join the market’ once it was fully established, saying ‘MStar always comes into a market late, so it’s waiting for volume.’
Our take is that there are suitable Chinese players in the market already and that MStar would be more likely to buy its way into the G.hn market than design its way in.
‘In China at both China Mobile and China Unicom, they have no plans to use WiFi to replace the entire home network. They are really worried about interference issues between different WiFi access points, especially inside their MDUs, and they will use G.hn for the heavy lifting inside a home and WiFi communication just for portable devices.’ He said pretty much that this was the same approach in Taiwan, and he had Mr. Shih Mu-Piao, President and General Manager of Taiwan’s Chunghwa Telecom at the HomeGrid Press Conference at Computex saying very clearly that he planned to deploy G.hn in the second half of this year, in much the same way.
Mu-Piao said, ‘With G.hn networking technology, completed commercialization is planned for the second half of 2014. Chunghwa Telecom will begin deploying then. G.hn is the solution to the problematic smart home networks issues. G.hn will help Chunghwa Telecom to expand the “light generation” fiber Internet and ICT technology deployment, and thereby strengthen the future of the Internet of Things.’
He added that the Industrial Development Bureau, Ministry of Economic Affairs and the Institute for Information Industry predicts that in the next four years the global wired home networking equipment market opportunities for G.hn will be around $667 Million, which he said did not include associated peripheral application opportunities for service providers.
The question on everybody’s lips is whether or not this technology can get out of purely a Chinese and Asia Pacific opportunity and move back across Europe into the US and Latin America. Egan insists that this is happening and cites 30 design wins that he cannot publicize, much as he would like to, in ‘all the major continents.’
There is one we know of in Portugal Telecom which Egan confesses has moved from the trial account it was back in September last year, to an RFP for deployment. Portugal Telecom has just 1.27 million broadband customers and the trial was held with Marvell rival Sigma, in a system that has ‘worked well.’
We think that Portugal Telecom was a fan of Spanish chip maker DS2, which has been at the heart of the powerline side of G.hn development, and which sold itself to Marvell. We understand that Portugal Telecom had early trials of UPA, the DS2 powerline technology which was absorbed into G.hn.
Portugal Telecom, a Microsoft Mediaroom customer, ran its triple-play service (MEO) over a G.hn powerline network for just 60-days trial, to an unspecified number of customer homes using the MEO service. Egan describes progress here as ‘they are playing with production units and are making their vendor shortlists,’ and he added, ‘Telefonica was the most advanced European buyer of this technology, but Portugal Telecom is now ahead of them and will deploy first.’ We suspect that most of the units being assessed are based on the Sigma G.hn chip since that was used for trials there.
Telefonica has been a long term advocate of G.hn, for much the same reasons, having wanted to work with Spain’s DS2 which Marvell bought. It will use G.hn probably alongside WiFi in the same way as China, but Egan says that he understand that each Telefonica region will make its own choice.
But this trend of using fixed line for a home backbone is certainly NOT being followed in the US where Comcast and AT&T are planning major WiFi ‘mostly’ installations using headless gateways, which will talk over WiFi to remote TVs, with a fixed line access provided only as the failsafe. At Comcast it is thought that backbone will be MoCA, while at AT&T the jury is still out. Our spies tell us that it is a straight fight between G.hn that AT&T has sponsored from the start and Homeplug AV2 powerline, and it may even take both technologies to trial.
But if our take on its gateway is correct, this technology will be less important for U-Verse than in prior generations, when it used HPNA, because it has selected the Quantenna WiFi 4 x 4 MIMO 80211n chipset and may even move up to the AC version. Quantenna 4 x 4 beamforming has the most penetration of any WiFi chipset, which is why it is the most expensive, and if AT&T invests in that it will not want to spend almost as much in a fixed line alternative.
To some extent it doesn’t matter, because if WiFi gets deployed for video more and more, then these fixed line home networking systems are going to be de-emphasized.
‘MoCA certain could be,’ says a mischievous Egan. ‘We’ve heard that some US cable operators want to shift up to use 1.75 GHz in their DOCSIS systems and that cannot co-exist with MoCA, so it will mean a forklift upgrade.’
That’s a bit disingenuous, in that DOCSIS 3.0 which is deploying now doesn’t go above 1 GHz, and MoCA can easily find the space it needs (100 MHz) above this. However the spec for DOCSIS 3.1 goes to 1.2 GHz with an optional extension to 1.75 GHz. Anyone using that will need a MoCA 3.0 chip which of course has not yet been designed, or as Egan suggests, it can switch to G.hn.
G.hn happily can work in 50, 100 or 200 MHz slices up to 3GHz, whereas MoCA uses 100 MHz just above the 1GHz line. We suspect this is a moot point, as fixed lines are supposed to be de-emphasized in the plans of the major US operators. We would not imagine that US cable operators would have trouble adapting to a new version of MoCA and it is probably under design now, but if it comes out late then Egan says G.hn would be happy to take the business. We’re sure it would.
We don’t think that any of that is likely, but at least one of the founders of Entropic which built the MoCA standard has since left the company, and it remains the province of Entropic, itself financially challenged, and Broadcom, which bundles MoCA into most of its set top chips.
There are other minor wins that Egan displays, the way the IEEE 1905.1a committee has come into being and he says it will have first draft at a meeting next month, and it could be ratified as early as September. This is an extension to the IEEE 1905.1 standard for communicating between WiFi, MoCA and Homeplug which G.hn was not invited to join, and he says that it was actually a Broadcom suggestion that was key to the first draft being so advanced, suggesting in his own way that maybe Broadcom is not so prejudiced against G.hn anymore and could one day embrace it. Egan also said that there has now been an interop between the MIMO versions of G.hn and the SISO version, and they can both be present in a single network now. This is important as China power arrangements are two cables, with Earth introduced further back, so it is unlikely to embrace MIMO.
That is typical of the mood around G.hn and Homegrid now, a feeling that the worst days are behind it and that anything is possible,. But we would not expect Homeplug or MoCA to lie down and simply let G.hn steal away all that they have achieved. So we first need to see Homegrid through the next six months without any slip ups, before anyone should get out the champagne.