Growing a spine

Developing the high-speed home backbone

By Matthew Miller, Special Projects Editor -- EDN, 6/26/2003

Entertainment-centric home networking is coming. Unfortunately, most homes aren't equipped to handle it.

For example, some observers expect that many consumers will soon want to distribute HDTV streams from a central receiver/storage device to multiple, far-flung displays. However, most homes lack the backbone—the built-in infrastructure of high-speed wiring—that makes such an application trivial. And currently available wireless and "no-new-wires" technologies don't cut it either. So apart from tearing up the walls, what's a would-be HDTV networker to do?

All the industry can offer at this point is a handful of developing home-backbone technologies—in other words, ink-still-wet specs, trade-show demos, vendor promises, and a lot of uncertainty. Will any of these technologies pan out? Come along as we survey the possibilities for this critical piece of the home-networking puzzle.

An HDTV stream can consume as much as 20 Mbits/sec. But bandwidth isn't the whole story. When multiple streams need to move over a network with tight bandwidth and a high number of nodes, a "best-effort" MAC (media access control) scheme, such as the one Ethernet employs, won't suffice. HDTV distribution requires a mode of transport with built-in QoS (quality of service) provisions—features that cater to data that absolutely, positively has to be there on time.

The ideal home backbone features category-5 cables, or some other high-bandwidth wires, radiating out from a central hub to wall plates in each room. This "home-run" setup provides ample bandwidth and can carry either Ethernet traffic or a QoS-enabled format such as IEEE 1394 (FireWire). For example, the 1394 Trade Association has demonstrated a system in which an HDTV stream in FireWire format travels over 100 meters of cat-5.

Needless to say, this ideal arrangement is an option only for new homes and those undergoing extensive remodeling. A fallback option involves stringing a single cat-5 cable across the longest span in the house. Though not as capable as a full-blown home-run network, this single-wire backbone can act as a bridge between separate islands of connectivity.

No new wires

Yet even this approach involves mess and expense. So technologies that can make use of existing wires remain enticing. For this reason, both the Home Phoneline Networking Alliance (HomePNA) and the HomePlug Powerline Alliance are pitching their technologies as media-ready backbones.

The HomePNA completed version 3.0 of its spec in early June. The technology offers a top data rate of 128 Mbits/sec and has deterministic QoS features, according to Rich Nesin, president of the HomePNA and vice president of marketing with CopperGate Communications.

None of these backbone technologies is close to real-world deployment in the quantities that would lead to standardization and critical mass.

Are such high data rates possible even in older homes, where phone wiring may be in poor condition? The alliance evaluated the technology against a test suite of 27 wiring loops, including some that were particularly hostile to communications, Nesin says. HomePNA 3.0 achieved data rates of more than 100 Mbits/sec in about 50 percent of the scenarios and more than 40 Mbits/sec in "just about all" the others, Nesin reports. The alliance expects silicon to debut in Q4.

Meanwhile, the HomePlug Powerline Alliance is working on its HomePlug AV scheme. Apart from saying that the technology will offer QoS and support multistream HDTV distribution, the group provides few details. That's probably because the details don't yet exist; the alliance issued a request for proposals in May and expects to complete the spec by mid or late 2004. It's worth keeping in mind that HomePlug 1.0 hasn't exactly been a rousing success.

Coaxial cable, which already runs between TV-equipped rooms in many homes, represents another existing-wire option. Players advocating this approach include Coaxsys, which recently introduced PureSpeed, a 100-Mbit/sec Ethernet-over-coax platform. As with HomePNA, the state of coaxial wiring in older homes may limit the usefulness of such schemes.

No wires at all

Turning to wireless possibilities, today's most popular networking technology, Wi-Fi (802.11b), is too slow, but its 54-Mbit/sec cousins, 802.11a and g, offer sufficient raw bandwidth. However, their achievable data rates drop off as distance increases, so their ability to cover large homes is questionable. Moreover, both technologies use an Ethernet-style, best-effort MAC.

Some proprietary wireless approaches exist. For example, Magis Networks has married the 802.11a physical layer with a proprietary MAC designed for media delivery. The company has demonstrated HDTV distribution using this Air5 technology, including one demo in which an Air5 link carried a 1394-based HDTV signal.

Finally ultra-wideband (UWB) wireless technology may be a contender as well. UWB can easily achieve the required bandwidth, and at least one vendor, XtremeSpectrum, claims that its UWB implementation can transmit over a range of 100 feet.

Wires after all

All of these backbone technologies bear watching, but none is close to real-world deployment in the quantities that would lead to standardization and critical mass. At least for the foreseeable future, true entertainment-centric networking will require not only high-tech equipment, but also equipment found at the local Home Depot.

Author Information

Special Projects Editor Matthew Miller (mdmiller@reedbusiness.com) would be happy with even one HDTV.

Dynamic duo

Plan combines Ethernet and FireWire

PCs speak Ethernet, which gives that networking technology a natural edge in home-networking applications. But IEEE 1394, better known as FireWire, has inherent QoS (quality of service) features that make it a better choice for consumer-electronics applications involving high-bandwidth video and audio. Now, rather than engage in a pitched battle, the shepherds of both technologies are working to combine them in a way that would be invisible to consumers.

The 1394 Trade Association is working with the IEEE (Institute of Electrical and Electronics Engineers) to develop a combined Ethernet/1394b PHY (physical layer). Once put into silicon, this PHY would live inside a hub that would connect all endpoints in a network regardless of what protocol those endpoints wanted to employ. That is, 1394 devices would think they were on a 1394 network, and Ethernet devices would think they were on a standard Ethernet network. At some point in the network, a bridge would allow Ethernet traffic to cross over to the 1394 "side" and vice versa.

The appeal of the concept is how it would change things for consumers. Instead of having to worry about differing network technologies, users would see a single RJ-45 wall jack that would simply work with whatever device they chose to plug into it. Inside the walls, an infrastructure of category-5 cable would carry both Ethernet and 1394 traffic.

The key to the effort is an autonegotiation protocol that would allow each port to select from various link-layer protocols, including 10- and 100-Mbit/sec Ethernet, 100-Mbit/sec 1394b, Gigabit Ethernet, and 400- or 800-Mbit/sec 1394. The 1394 Trade Association expects to see silicon vendors introduce the PHY later this year.