Remember the movie "Tron," where ordinary human beings were sucked up into the innards of their computers, becoming strange participants in a hidden, electronic world? While today's consumer of video and data services might not be ready to take that much of a plunge, subscribers are embracing better quality, more realistic pictures (a la DBS), as well as a more compelling, interactive entertainment experience (like that found on-line). But before cable subs start demanding to become one with the return path, there's work to be done.

CED recently polled three engineering executives about progress on activating the return path, as well as a status report on making the network reliable enough to support a plethora of new, advanced services. In addition, participants in the informal engineering roundtable offered their thoughts on which service—data, digital video or telephony—they are energizing first.


The participants are: Alex Best, senior vice president of engineering, Cox Communications Inc.; Jim Chiddix, senior vice president of engineering and technology, Time Warner Cable; and Tony Werner, vice president of engineering, TCI Communications Inc.

What follows is an edited version of their responses.


CED: How aggressively is your company installing and actually activating the return path of the network?

Alex Best: At the end of 1996, we had about 6,000 miles activated, costing us about $1,500 a mile to activate it. And we have budgeted in 1997 to do an additional 12,000 miles. In 1998, 9,000 miles. That's out of a total in our company of 50,000 plant miles.

Cox is somewhat unique, in that 85 percent of our customers are in nine large clusters. That is where we are focusing our efforts, on those nine large clusters.

Jim Chiddix: We have about 18 million homes passed across Time Warner right now, and of those homes passed, we have 12.1 million subscribers. As of the end of '96, we had upgraded 6 million of those 18 million passings to fiber-to-the-neighborhood, with, in virtually every case, activated two-way.

By the end of '97, we expect to have upgraded 9 million of the 18 million passings, and by the year 2000, we expect to have upgraded all of them.

Tony Werner: The company has lots of systems intermixed that have return path turned up or are in the process of turning it up. But for advanced services, we've really focused on three areas: the Bay area, part of the Chicago area, and Connecticut. We have spent a good portion of this year working hard to bring up advanced services in those three markets, with our objective being to offer core cable, digital cable, high-speed Internet service and wireline telephony. We have a pretty good-sized footprint in those three markets now that is going to grow to around a quarter of a million homes passed in the first quarter.

CED: What's the greatest obstacle to deploying a robust, working return plant?

Best: I'd say there are two. The first one is controlling ingress; the second one is proper levels set-up and alignment. To ensure proper alignment, the return amplifier must be adjusted correctly, the laser input and output levels must be adjusted correctly, and the "boxes" in the home must provide the correct level into the first active return module. If this occurs, then everything works properly. But if it's not properly set up, you can create an overload problem which can affect everyone in that node.

Chiddix: It's not our experience that there are serious obstacles to activating it. You have to train people, you have to align it, you have to set up a maintenance procedure, and you have to do a reasonable job of maintaining plant integrity. Those aren't obstacles; those are just the steps required.

Werner: Staff and training are probably the biggest issues that we have faced. While ingress is perpetual, it has not been a huge issue. A lot of these technologies have been robust enough to work in this environment pretty well.

CED: Are you considering using some of the more advanced modulation methods (such as S-CDMA and OFDM) to make the return work? If not, are you confident that noise and ingress can be overcome effectively?

Best: The answer is yes, but when we select a particular product, whether it be a digital video box, a telephone modem or a data modem, we look at the whole package. The return modulation method is just one of the issues. We will be deploying some boxes that use QPSK, and some that use more advanced schemes.

As for the second part of your question, Cox is initially putting a blocking filter on every active drop. The first thing we do is activate the return spectrum in the trunk, feeder and fiber node. Then, we put a blocking filter on every active drop at the tap, which essentially renders the system a non-two-way system. Then, if you take a two-way service from us, we remove that filter from your drop. A home that does not take a two-way service, therefore, will not contribute to return ingress, because the return path out of that home is blocked.

Most of the ingress occurs in the home. So this means that you don't have to "clean up" those homes until they take a two-way service. We think this saves money.

Chiddix: We are not depending on any of those. We have a significant number of telephony subscribers in Rochester today who are using our return path using QPSK modulation. We have a very significant number of Roadrunner cable modem customers in a number of cities, using the return path every day, with QPSK. These other schemes claim to be more robust, and/or to have more spectrum efficiency. If that proves out, those are good things. But we certainly don't need that in order to make effective use of our return path for new businesses.

Werner: We are interested in Synchronous-CDMA, or OFDM—anything that would give you a return path and allow it to be very robust, where you really don't have an increase in maintenance to keep the plant working. That's what I think everyone would like to see. With that said, even some of the QPSK systems that are using forward error correction and frequency agility are proving themselves to be quite resilient in that environment.

CED: Have you attempted to quantify your network's reliability levels? How close do you think you come to 99.99 percent availability—and how important is it that you achieve that benchmark?

Best: At the end of 1996, on a company-wide basis, we were at 99.986 percent reliability. We make all of our systems report in on a monthly basis: whether or not they have an outage, how long the outage lasts and how many people were affected. And so 99.986 means that, on average, in 1996, a Cox Cable customer was without service for about an hour and 15 minutes.

Chiddix: We have tracked it very carefully in Rochester, because we are delivering telephony services in a competitive marketplace, and we are achieving 99.99 percent availability there.

The conventional wisdom is, if we are going to compete in the POTS marketplace, we need to have very good reliability. Whether or not that needs to be at this somewhat artificial level of 99.99 percent or not is open to debate. But reliability is certainly important.

CED: What, in your opinion, is the greatest obstacle to achieving that level of availability?

Best: First of all, it's fiber deployment. We would not have gotten to an hour and 15 minutes if we had not deployed fiber. The next most important thing we are going to do is install central node power supplies with gas generator backup. It's powering issues which primarily affect outages in a cable system. The third is the deployment of route diverse fiber to 1,000-home nodes. With these three in place, we will be well below 53 minutes (Bellcore's recommendation) on an annual basis.

Chiddix: I don't think we have had a difficult time achieving it. Rochester was upgraded a few years ago with an HFC architecture. The amplifiers we use have a very good reliability record. And there's nothing that is extraordinarily different; although, for telephony, we are obviously installing standby power.

CED: What, if anything, are you doing from a powering and architecture point-of-view to improve availability?

Best: In the nine systems where we are going to offer telephony, we are going to go to central node power supplies—that means, one large power supply per 1,000-home-node, and that power supply has a natural gas-driven generator sitting adjacent to it. If you lose utility power, the batteries take over for a few minutes, until such time as the generator starts up and runs indefinitely. So essentially, it's going to an uninterruptible power source. And that, along with our fiber deployment, will get us to less than 53 minutes.

Chiddix: When a customer signs up for residential HFC telephony in Rochester, we install a new tap bottom plate. And we do that without interrupting the service by jumpering around the tap and then installing the new plate, but we also install a normal through switch, so that the next time that power-passing plate is removed, power is maintained automatically.

In a large number of locations, we are able to get feeds from the local power utility, from two different portions of the electric grid—two different primaries fed from two different substations. We do put in fairly short duration battery standby, and a transfer switch, so that if one of the power grids goes off, we switch to the other automatically. If they both fail, we have some standby power.

In locations where we aren't able to do that, we are doing one of two things. We are either putting in just standby batteries with a longer duration bank of batteries (or) we are putting in batteries and a local generator, usually tied to a local natural gas supply. That is the belt-and-suspenders method. In theory, you are power independent forever. The power goes off; the generator comes on; the batteries bridge the startup time for the generator. And then you run the generator forever, and you are hooked into the pipe natural gas system, so you don't run out of fuel. Now, that is not only the most expensive, it is also the most intrusive in terms of people seeing big boxes show up in their yards. So it's a tough sell. That's why we're exploring all three approaches.

CED: Have you built, or are you planning to build, any networks that feature a "ring" design that eliminates a single point of failure? Are you using Sonet technology anywhere, or do you plan to?

Best: As far as I know, we are the only cable operator who has deployed fiber optics in a route diverse or ring fashion, down to a 1,000-home-node level. And before we offer telephony, we would activate both sides of the ring . . . .

We are deploying Sonet in three ways. One is, around fiber rings in large systems from headend to hubs to interconnect the switch to all the hubs. Secondarily, in CAP networks, we use Sonet for voice. Thirdly, for large MDU complexes where we offer shared tenant services, we use Sonet to deliver the voice services to the MDU.

Chiddix: We use rings to what we call distribution hubs. Distribution hubs are points where we then feed an area of up to about five miles in radius. We do not run rings out to each node. We don't think that is justified in terms of cost vs. reliability. We are using Sonet rings, or are building Sonet networks in a fair number of our markets in order to provide telecommunications services, primarily to businesses. Alternate access to long distance points-of-presence, and other kinds of high-speed data services to commercial customers. Where we have such facilities, we take advantage of them to provide interconnection for our Roadrunner service, for example.

CED: Which of the new revenue opportunities (i.e., telephony, data, new digital video services) is the highest priority for your company?

Best: I would put data as number one, for several reasons. There are no regulatory hurdles. It offers a service—meaning high-speed—that is just unavailable anywhere else. The modems are available, and with the access issues that are occurring using dial-up modems, cable modems look especially attractive.

Digital video would come in second. Once again, no regulatory hurdles. The primary hurdles are programming, which of course, is now available from HITS, and boxes, which are now available from GI. The reason I put telephony third is, you have to have interconnection agreements; you need number portability sooner or later, or it will limit penetration; and you need to overcome the powering issues.

Chiddix: It's hard to say that one dominates the others. Residential telephony is something that we are not deploying in additional markets because of regulatory uncertainties. We are not sure when that will be clear. Having said that, we are proceeding aggressively with it in Rochester to learn all we can about that business.

The most immediate business for us is the cable modem business. And we are going to roll it out in a fair number of cities this year. We also think that digital video roll-out is extremely important. It won't happen quite as soon as the modem rollouts, but when it hits, it will be a much more broadly-based service.

Werner: All three are a priority with us, but I guess we think that digital TV is a huge opportunity. Telephony and data, we are still putting significant efforts into those. But they are going to be a slower market to harvest. In the short-term, we can get digital set-tops out in a ubiquitous manner, because you don't need HFC two-way plant to operate them, where for telephony and data, you do.