1394 interoperability still a no-show

1394 interoperability still a no-show
SAN MATEO, Calif. — The IEEE-1394 interface appearing in the current crop of digital electronics gadgets fails to provide the interoperability system makers promised. What's worse, industry experts say, the problems dogging 1394 today — such as mushrooming software complexity — could hold back other wired and wireless consumer interfaces as well, postponing the arrival of truly plug-and-play home networks for perhaps a decade.

"Plug anything — camcorder, PC, set-top — into Playstation 2 with 1394, and you'll basically get no functionality," said Mark Kirstein, vice president of research at Cahners In-Stat Group. "Similarly, only a few MPEG D-VHS-based VCRs can talk to DV camcorders. There is a chance of consumer backlash when things don't work."

Often a 1394 interface in one system will work only with another system that carries the same brand and is programmed in the same proprietary manner. Otherwise the system just sits there, looking sleek. "At best, today's 1394 products are badly interoperable," said Shiva Patibanda, chief executive officer at VividLogic Inc. (Fremont, Calif.), which develops 1394 software.

"People had somehow expected magic software that would miraculously make every device interoperable with everything else in the digital domain, but obviously that hasn't happened," said Greg Bartlett, president of Digital Harmony Technologies Inc. (Seattle), a developer of audiovisual software and device modules.

Some misjudged the bumps in the road to interoperability. "When 1394 was developed, a lot of people did not realize its profound implications," observed Scott Smyers, vice president at Sony's Interconnect Architecture Lab. The 1394 interface is bidirectional and point-to-point and also allows multichannels. Taking advantage of those features and implementing them correctly in systems, Smyers said, has proved "more difficult than people had imagined."

"We cannot deny the possibility that our 1394 systems do not talk to other companies' systems, if they were not available while we were developing our own products," said Thomas Asabe, general manager of the business and products planning group of the AVC Products Development Laboratory at Matsushita Electric Industrial Co.

The interoperability problems with 1394 can be attributed in part to divergent system-level engineering philosophies and implementations.

Digital difference

Consider the typical home entertainment cluster. Cables and physical-layer chips that meet interconnect standards won't work together, even if physically connected, unless each system's engineers have correctly implemented the control protocols and commands for AV devices. "In the analog days, if impedance and signal levels were correct, two systems talked to each other. That no longer holds true in the digital era," said Eddy Odijk, vice president of system architecture for Philips Consumer Electronics (Eindhoven, Netherlands).

A further issue is digital data format conversion. Unless the device receiving data can decode the information and display it correctly, interconnectivity accomplishes nothing. "Interconnect essentially only assures that messages are routed from one place to another so that two parties can exchange bits," Odijk said. That capability alone does not assure the device receiving the information can successfully read and display the bits as intended.

Many ascribe the interoperability problems to software complexity. Digital Harmony, for example, found it had to incorporate as many as 42 different protocols in its 1394 software suite to maintain compatibility. To address the problem, Digital Harmony, VividLogic, Philips and others are supplying their own off-the-shelf software stacks, rather than have each OEM write its own code.

But the digital interoperability problem threatens to cut across a wider swath than 1394. Other consumer-oriented interfaces, such as Bluetooth, could face similar hurdles. And OEMs must tackle the issue of building bridges among the interfaces themselves.

Each device interface may in turn link to one of the emerging home networking approaches, which use different protocols. Achieving interoperability within one home-networking architecture is one thing; pulling it off among two or more home network structures is quite another.

Industry opinions are sharply divided on what will be required to solve the interoperability problems and on how long the fixes may take. Some predict that a Dark Age of digital incompatibility and resultant consumer confusion could plague the consumer-electronics industry for as long as 10 years, slowing down the rollout of digital TVs, set-tops and digital recording devices. Others see help on the horizon.

Two separate ad hoc industry efforts hope to rein in the software complexity by specifying a software architecture for plug-and-play consumer devices: Home Audio/Video Interoperability (HAVi), backed by a group of consumer-electronics companies, and Universal Plug and Play (UPnP), spearheaded by Microsoft Corp. While the groups complement each other in some ways, observers say they remain largely competing efforts that will further muddy the waters.

Others cite a still-deeper divergence of opinion on the underlying question of whether home networks should be controlled by a single intelligent device, such as a PC, or whether distributed architectures will prevail.

In-Stat's Kirstein said he expects home networking to evolve in multiple phases spanning at least 10 years. "Several interconnects will dominate in different clusters, so full interaction involves lots of protocols, bridges and physical interfaces," he said.

Phase one, in Kirstein's view, will be PC-based home networking with limited non-PC-based networking, primarily for Internet access. Phase two will add such applications as multi-room digital audio. In this phase, "non-PC-based networking becomes more important and residential gateways emerge," Kirstein said.

Phase three, he said, will occur when the market sees "limited interaction among different network clusters" — PCs, consumer devices, telecom and control functions. And phase four will see the realization of the fully interactive, "smart" home.

Battle lines

The interoperability battle plays into a larger turf war between the PC and TV industries — a conflict that has begun to engage communications OEMs as those companies build Bluetooth connectivity into wired and wireless gear. "The problem is that everyone wants to eat in each other's space," said Mark O'Brien, director of platform marketing at TeraLogic, a digital-TV chip maker. "And everyone wants to handle Internet protocols, quality-of-service and digital rights management" with an interconnect scheme that best serves their interests.

The key questions in devising solutions for the connected home of the future are where the ultimate intelligence will reside, whether control of the networked appliances should fall to one device or multiple devices, what those devices should look like and who should build them.

Microsoft, not surprisingly, has stumped for an approach — Universal Plug and Play — that makes the PC the intelligent master of the network. Some consumer electronics manufacturers, resisting the PC-centric master-slave notion, have banded together to support HAVi, which would allow consumers to connect any digital home appliance developed by any manufacturer in a non-hierarchical home network architecture. HAVi does not assume that a PC or any central controlling device will be charged with issuing command-and-control signals to the networked digital appliances.

Despite those clashing views, Sony's Smyers said he doesn't think "the battle is any longer about HAVi vs. Universal Plug and Play. I believe that they will happily coexist."

At a time when "nobody wants to make dumb and dumber products," it is logical to expect that "there will be multiple, incredibly intelligent devices at home," Smyers said. But "there needs to be a way to share the intelligence among those devices, and decide, depending on the circumstances, which device becomes the controller, which one the controlled, and which one serves as a proxy."

The HAVi architecture allows for two levels of interoperability. Level 1 supports a generic set of commands that lets one device to talk to another as well as a set of event messages that one device should expect from another.

Level 2 interoperability seeks to make the HAVi architecture future-proof. When a new device with a set of previously unseen features is attached to a home network, it can order the other devices, "Forget everything you used to know about devices like me." The HAVi architecture allows the device to upload device control modules as an alternative to embedded DCMs. The uploaded DCM will be written in Java; the HAVi architecture requires one device in the home network to provide the uploaded DCM with a run-time environment — a Java virtual machine.

For the consumer, Level 2 will hide complexity and ease plug-and-play. On the business side, HAVi will let consumer electronics vendors create and distribute a consistent user interface along with unique applets across devices in the network. "The money is in applications, not in hardware," Patibanda of VividLogic said. "If CE vendors don't want to yield to PCs, they need HAVi to succeed."

Some in the industry nonetheless reject the notion of a distributed computing architecture for home networks.

Faulty assumption

"There is a general idea in the industry today that centralized computing is bad and the distributed approach is good," said Peter Forman, chief executive officer and president of Ligos Technology (San Francisco). But the notion of building an autonomous home network structure wherein distributed applications can run on various smart devices by sharing one another's resources — i.e., processing power and storage capabilities — is "based on an unreasonable assumption," he said. "The concept only works if system application layers can be defined clearly and their applications can remain static."

Say a new streaming video capability suddenly became available that your Internet appliance couldn't handle, Forman said. "You would need a centralized controller that could figure out how to convert the data, process it correctly and communicate it to the other devices on the home network."

Indeed, many in the industry say they believe there is a need for an intelligent device — whether a PC or not — that can serve as a proxy, gateway or bridge for digital consumer devices running different home network protocols and delivering data in diversified media formats. Some envision the device as a plain digital cable set-top or a digital TV. Others see it as a TiVO or Replay type of personal video recorder, which TeraLogic's O'Brien describes "as a half-PC and a half-set-top."

Still others expect the emergence of a variety of residential gateways that not only will interface with broadband pipes coming into the home but also will serve as a bridge among multiple network protocols.