Attack of the PODs

Point of deployment modules are almost here, and they want to take control of your TV set.

POD modules are the key to the architecture developed by CableLabs' OpenCable project to achieve interoperability of digital cable set-top boxes. PCMCIA cards resemble telephone modems for PCs and contain processing chips, memory and "conditional access" circuits. When plugged into a digital set-top bought at retail, the cards call out to the cable operator's headend, download code and wake up that empty box to activate service.

And they're due for mass release on July 1. That's the deadline the FCC set for cable operators to make provisions for permitting subscribers to go to retail stores to buy their own "navigation devices" - set-tops and cable modems - if they so desire.

The FCC mandate originated from the 1992 Cable Act but gained real momentum with the 1996 Telecom Act, which required that such navigation devices be made available to consumers from sources other than the video service providers or vendors linked to their multiple systems operator (MSO). Acting on that authority, the FCC ruled in June 1998 that consumers can attach any compatible equipment to the network as long as it doesn't compromise their systems for controlling who can tap into that network - known as conditional access mechanisms.

At that time, set-top manufacturers - primarily Scientific-Atlanta and General Instrument, which was bought by Motorola - routinely integrated conditional access into the design of their boxes, which most cable operators then leased to their customers. The FCC ruling literally meant the end of this tight integration of security and functionality. By 2005, MSOs will no longer be able to supply their subscribers with digital set-tops that contain embedded security.

But the same law that gave the FCC the authority to demand the functional split also forbade it to jeopardize the security of the cable network. Nor did the commission have the power to require that operators redesign their cable networks or to compel them to adopt a single, standard system architecture. And it couldn't compel cable equipment manufacturers to license their proprietary technology to competitors.

So the cable industry had to devise an equipment model that would let them comply with the FCC's rule - and quick.

Enter CableLabs, the R&D consortium composed of cable manufacturers and operators. Under the OpenCable project, CableLabs researchers began working on a design that would meet the government's requirements for modular security features.

They also had another impetus: economics. "If you think back several years, there was a sense of urgency among cable companies to try to spend less money and make more revenue, and one way to do that was to enable direct retail purchase of set-tops by customers and get that investment off the company balance sheet," said Don Dulchinos, CableLabs' vice president for advanced platform services and the director of OpenCable. "That goal has receded as the financial situation has improved, but the retail goal is still there."

The competitive onset of direct broadcast satellite has kept that pressure on MSOs. Customers can buy satellite dishes at the local RadioShack; cable operators want them to have the chance to buy cable set-tops, too.

As the process of writing specifications for interoperable cable hardware progressed, a parallel process was happening in regard to software specifications. "While that's not something that the FCC required, it's clearly something that our membership is interested in having to add service portability to the hardware," Dulchinos said.

Indeed, developing hardware that could work in any cable system became just phase 1 of OpenCable. Arriving at a software platform that allows MSOs and third-party developers to add services easily and lets customers move those services when they move the box now is the project's long-term goal.

"We want to get to where the standard software middleware layer allows those kinds of applications to be portable so a customer who's used to having a bundle of services - digital video, e-mail, a browser - in a given location can get the same suite of services in another cable system," Dulchinos said. He's hopeful that goal can be reached by the end of 2001. OpenCable issued a request for proposal for software specifications in September and currently is working on a draft version of that design.

Manufacturers are working on service portability - many of them in apparent testimony to the effect of the FCC ruling on set-top competition - but the boxes that will be available on July 1 are pretty much limited to a basic "watch TV" mode. And at the National Cable Television Association's Cable 2000 gathering in New Orleans last month, several of the largest companies now involved in the project presented their views of the technical challenges and conceptual missteps that they believe dog the pursuit of the truly portable navigation device.

For one, the software design underlying the POD module has some serious inherent flaws, said Mark Eyer, director of systems engineering for Sony Electronics' Digital Media of America.

"That retail box comes into your house empty," he said. "Once you get it home, you turn it on and magically software drops into it, and it becomes tailored to the look and feel of whatever cable system you're in. And if you move to another city, it will empty out and then get loaded again."

Actually, the set-top comes with a small resident application that supports the initial download; when the MSO's application suite is loaded, the resident application that came with the box goes away. That jeopardizes any attempts by the set-top manufacturer to differentiate a product from its competition, Eyer said.

"Sony would like to build some software into a set-top box that's better than the code from the local cable operator, with perhaps better features," he said.

Eyer also questioned whether MSOs could develop software that would run different digital devices - personal video recorders, consoles for networked gaming, a navigator that could control other audio-visual equipment and gateways for home networking.

Moreover, some of these service functions can't be managed by software alone. For example, some upcoming video formats such as MPEG-4 cannot be implemented via a Java applet; they require new codec hardware.

The POD design also introduces new complexity into the larger network infrastructure, particularly in the area of copy protection, which prevents the pirating of content from a network. The interface between the POD and its host device is vulnerable to perfect copying, said Mark DiPietro, senior director of systems engineering and technology for Motorola's broadband communications sector.

He outlined some of the security provisions that must be built into the new devices and the network they hang off. "First, you have to add encryption on the POD and decryption on the host," DiPietro said. "There's also a requirement for the host to take a certificate that says it's going to be a good citizen with the content it receives from the POD." The POD has to look at that certificate to make sure that it's legitimate, and the rest of the system must acknowledge that this has taken place, requiring certificate reporting back to the headend.

"This is the first case of copy protection where we've seen a requirement for headend support," he said.

Network management is similarly complex under the POD design. MSOs will need to know what kind of navigation host devices they have in their systems and what hosts and PODs are running together. They will have to know generic hosts' hardware configurations and what software is compatible. And they must decide what to do about outdated hardware that they don't own.

"Right now, since they own the set-tops, the MSOs have the option of taking an entire class of devices and swapping them out for an upgrade," DiPietro said. "That won't be possible when a consumer owns the device, so some creative way will have to be found to support boxes that may be getting old."

And there are other software concerns beyond middleware. How does the "empty" set-top get provisioned into the network? If it's talking IP, how does it get the proper IP address? How does it download applications? You need the middleware to do that, but where does the box get the basic boot-loading capability?

So there may be many different aspects of OpenCable still to be worked out. There's no doubt that the FCC's July 1 deadline will be met, said Tony Wasilewski, chief scientist for the subscriber networks sector of Scientific-Atlanta; but it will not be met with a richly featured product.

"Scientific-Atlanta and Motorola will both have POD modules with their respective conditional access systems in them, and both companies also will have a limited number of host set-tops that you can plug those PODs into," he said. "But it won't be anything earthshaking."

To Wasilewski, the biggest obstacle to portability of both hardware and software in set-tops are not strategic - they're tactical: encryption and copy protection. "From a technical viewpoint, they have been covered," he said.

Software constraints such as Eyer spoke of are a concern, but not overwhelming.

"It's a more political hurdle than a technical one," Wasilewski said. "I think the POD can do what you want it to do. The question really is, what will it be allowed to do?"