Media networking--using home networking to carry audio and video around the home--plays an important role in the strategies of the PC and consumer electronics industries.
High-definition video is the most challenging application for networking. HD needs data rates upwards of 20 Mbps per channel, sufficient range to cover the entire house, and "quality of service" (QoS) to assure that the video gets from one place to another without interference from other applications.
Until now, only Fast Ethernet over Category 5 or 5e cabling has been able to provide the speed and range necessary to support HD. But most people don't want to incur the expense of rewiring their homes for Ethernet. The PC industry has encouraged the development of networking technologies that could be used in existing home without running new wires.
That "no new wires" world is now unfolding. New networking technologies--operating over existing electrical wiring, coaxial cable and over the air--have all been designed with HD in mind. All promise sufficient speed, range and quality for several channels of HD video, plus SDTV, audio, voice and data. Some products are on the market now; others will arrive soon.
What is not yet clear is how well these technologies will work in real homes. It's easy to dazzle consumers with claims of physical networking speeds: "200 Mbps! ... 270 Mbps!! ... 300 Mbps!!!" What counts is real-world throughput at any place in the house you want to use it. Wi-Fi is notorious for having "null spots" where there are weak signals or none at all: microwave ovens and other devices can wreak havoc on wireless signals. First-generation powerline networking works well at some outlets, badly at others. Existing coaxial cables are full of splitters, poor connectors, and other potential impairments. The next year or two will resolve how well the latest technologies overcome these challenges.
Powerline Networking--HomePlug AV Finally Coming to Market
"Imagine: Entering your house, you unpack and plug in your newly purchased flat-panel TV. Simply and quickly - the TV automatically connects to the cable box, the DVD player, the Digital Video Recorder, the Home Theatre system, and also to the Internet." (HomePlug home page)
"No New Wires" (Intellon trademark)
The members of The HomePlug Powerline Alliance, the industry group developing standards for powerline communications (PLC), have long believed that if they can get the technology right, their chips would be built into all devices--you just plug everything in and it's networked. HomePlug 1.0, the first generation, provides performance comparable to the original 802.11b Wi-Fi. It works quite well, but never got much attention in a US market dominated by Wi-Fi; it did much better in Europe.
HomePlug AV, the second generation, has been in development for more than three years. It is targeted to networking high-definition television over existing electrical wiring.
Intellon's technology is central to the HomePlug standard, and it has produced most of the HomePlug chips. A few months ago, Intellon announced that it had shipped more than five million chips based on the HomePlug standard, and that its HomePlug run rate was more than one million chips per quarter. Some of these are a "Turbo" version of HomePlug 1.0 that pushes throughput up to speeds comparable to 802.11g.
Earlier this year, we interviewed Charlie Harris, Chairman and CEO of Intellon; a few weeks ago, we interviewed Andy Melder, Senior Vice President - Strategic Business Development, to follow up on a series of recent announcements.
Andy pointed out that HomePlug is being deployed by service providers all over the world. Both telephone companies and cable operators have adopted HomePlug 1.0 to extend broadband Internet service throughout the home. Recently, several large phone companies including PCCW and France Telecom have announced deployment of Intellon solutions for IPTV distribution in the home.
He said Intellon's HomePlug AV chips are now available and Homeplug AV adapters are in the final stages of testing. He said real throughput "will be between 40 to 60 Mbps on average except for a few outlying outlets."
He promised to ship us a pair of HomePlug AV adapters as soon as he had production models with finished software, and said he thought that might be as early as July. We've looking forward to running a set of AV tests in our home similar to the tests we ran on HomePlug 1.0 devices four years ago.
Other companies have developed different proprietary PLC approaches, but Intellon has long believed in building an ecosystem with many other companies through the HomePlug Alliance. With HomePlug AV devices starting to come to market, we will soon learn whether their long-held dream of "plug it in and it works" will come true.
Wireless Networking--Is "Draft n" Too Early?
IEEE 802.11n--the next generation of Wi-Fi--has been in development for more than three years. It aims to provide throughput of 100 Mbps or more, a range sufficient to cover an entire house, and the QoS necessary for high-quality video. Although the IEEE standard is not expected to be published until late next year, "draft n" chips based on a proposed draft of the standard have recently become available from several semiconductor companies, and products based on these chips are now available at retail.
This situation has some similarities to what happened previously with 802.11g--the current generation of Wi-Fi--but also some differences. "Draft g" products first appeared on the market in early 2003. Early adopters bought these early products to gain a three- to four-time improvement in network throughput compared with 802.11b. As users encountered some performance and interoperability problems, the industry was able to reflect the learning in improvements to the device software.
These "draft g" products were based on a nearly-final draft of the IEEE 802.11g standard that had already passed through many successful ballots in the standards process and was less than six months away from publication. The chip vendors were confident that any problems discovered in early products could be remedied by downloaded software fixes; some device vendors promised to replace equipment if it could not be upgraded by a software download to comply with the final standard. And in fact, downloading the updated software fixed most of the problems and these devices were upgradable.
By contrast, 802.11n is much further away from a stable draft. Many of the chip companies worked together during 2005 under the banner of the Enhanced Wireless Consortium (EWC) to create a proposed draft for the 11n standard. At the same time, several of these companies developed chips conforming with the EWC draft; these are the chips that appear in products claiming "draft n" compliance.
The initial draft of 11n--Draft 1.0--is different from the EWC proposal, and is a long way away from a final draft. It has not passed any ballot in the standards process--in the first ballot held in late April, it received less than a 50% "yes" vote with 75% required for acceptance. This suggests that many changes will be made between now and publication of the 802.11n standard, currently projected for October 2007; if it follows a timeline similar to 11g, a "nearly-final" draft will probably not be completed this year.
Early adopters who buy these new "draft n" devices will probably gain a three- to four-time improvement in throughput and a substantial improvement in range compared with standard 802.11g. If they also expect that these devices will interoperate with each other, or be upgradeable to the final standard, they will probably be disappointed.
The Wi-Fi Alliance has made it clear that it will not certify interoperability until the standard is approved. Devices based on the same chips will most likely interoperate properly. Two of the "draft n" chip vendors--Broadcom and Atheros--have performed their own tests and claim their chips will interoperate with each other; while helpful, this does not guarantee that devices based on these chips will interoperate. We hear from several sources that devices based on Marvell chips will not interoperate with devices based on Broadcom or Atheros chips--they fall back to 802.11g. Responsible device vendors--such as Netgear, which makes two nearly-identical products using chips from Broadcom and Marvell--promise interoperability at full speed only with devices based on the same chips.
None of the device vendors are promising that devices will be upgradable to the final standard. Vendors we have talked with say that they do not expect them to be upgradable, and say their early customers are early adopters who are looking for high speed and don't expect upgradability.
Some of the draft changes currently in process relate to the behavior of 11n devices in networks that include 11b and 11g devices. Some early reviews report that some early "draft n" devices are not only not interoperable with others, but are not "good neighbors" when used in proximity to 11b and 11g networks.
In an exhaustive evaluation of these early "draft n" devices, Tim Higgins--one of the most respected reviewers--summarized his findings by saying "The industry had better stop hyping and start fixing this crap...and fast."
Our view is that if these devices are bought by early adopters with their eyes wide open, the lack of interoperability and upgradability will seem like a reasonable price to pay for huge improvements in speed and range. If industry hype leads many naive consumers to buy these products without reading the fine print, it may damage 11n in the eyes of consumers.
Metalink's View: EWC is a "Net Good"
Over the past year, we have been exchanging email and talking with Ron Cates, VP for North American Sales and Marketing at Metalink Broadband, an Israeli chip company that has been very active in 802.11n. About a month ago we again interviewed Ron by telephone to get his views on the prospects for 802.11n.
Metalink is a member of EWC and contributed to its draft. Ron said EWC's contribution would be viewed as "a net good" in accelerating the process of resolving the standards issues, and expected it would lead to an earlier draft as a basis for truly interoperable and upgradable products.
Ron said high-definition video is a very difficult wireless application. He said operation in the 5 GHz band is "critical for high-definition television", and HD also needs "MIMO, channel bonding, packet aggregation with block acknowledgement, and advanced forward error correction (FEC)--if you leave any out, you can't do multiple streams of HD."
Ron expects to see chips and devices specialized for different markets. Some companies might focus on HD--last week, Metalink announced a deal to supply chips to Philips for this application. Others might focus on Wi-Fi phones (which require long range but could sacrifice speed for low power consumption) or high-speed data (which needs speed and range, but could leave out other things needed for HD).
Ron predicted that interoperable and upgradable 802.11n products will reach the market by the first quarter of 2007, and will be shown at CES in January. He said consumers "will be very impressed with 11n performance."
Coaxial Networking--MoCA Gaining Speed
The Multimedia over Coax Alliance (MoCA) develops and promotes specifications "for the transport of digital entertainment and information content over in-home coaxial cable." Members of MoCA include technology providers and many video and data service providers such as Comcast, Echostar and Verizon. MoCA recently announced that AT&T had also joined, represented by AT&T Labs.
We have long believed that coaxial cable could play an important role in media networking. Coax connects all television sets in the typical North American home and is inherently capable of carrying high bandwidth (a year ago we wrote "if cable or satellite broadcast television looks good on the TVs, the coax will probably carry high-speed data just fine").
Entropic Communications, the primary company developing chipsets and associated software for MoCA specifications, recently made some significant announcements, and we took the opportunity to interview John Graham, Entropic's VP of Marketing, over the phone.
John said that Entropic has shipped more than one million first generation c.LINK chips compliant with the MoCA specifications, and has now introduced the second-generation of c.LINK. John said the new EN2210 chip lowers power consumption, and provides additional "useful interfaces" to "simplify integration into set-top boxes and fiber nodes," especially important for telephone companies as they roll out IPTV over DSL and fiber.
The Entropic press release for the EN2210 said it provided "270 Mbps throughput" and we questioned John on that assertion. In fact, 270 Mbps is the PHY or physical data rate of the chip, not the throughput. But John said the performance is very good: "the effective throughput is around 100 Mbps - guaranteed over eight nodes with full MoCA network at 98% of all F connectors in the house." In addition, the E2210 has a burst mode: "it can burst to specific nodes at 150 to 200 Mbps." It includes QoS since "operators want to use it with PCs as well as video" and QoS will give priority to video.
John thinks coax will be used to provide a backbone for media networking throughout the home, with wireless used to reach mobile devices. "You'll have twenty to thirty million homes with coax backbones. Operators will deploy entertainment networks in the home. They will run at least 100 Mbps, work in all the sockets and go 2x to 3x over time." He pointed out that Entropic has shipped a million chips in the past eight months, and its technology has been incorporated in a lot of products for the cable and telephone companies. Other chip companies have joined MoCA, and Entropic is in licensing discussions with several of them.
MoCA is not the only video home networking technology using coax. The latest version of Home PNA, which started as a phone wire technology, also runs over coax, and there are also several proprietary technologies. Only MoCA has attracted such a wide group of service providers.