Cable's Bandwidth Smorgasbord: More Speed and More Capacity

“Fast. Faster. Fios. ... breakthrough online speed ... Connection speeds up to 15 Mbps for only $49.95/month” (Verizon Fios website)

“Comcast Delivers New Ultra-Fast Speed Tiers - 8Mbps and 6Mbps”(7/12/05 press release)

All over the world, broadband providers are competing on the basis of speed. Some are deploying super-high-speed broadband systems: some pure "fiber to the home", others hybrid "fiber to the curb" systems combining fiber with DSL. All claim much higher speeds than current US cable and DSL broadband services -- some now approaching or exceeding 100 Mbps and a few at gigabit-per-second (Gbps) speeds.

This talk of ever-higher speeds appeals to potential customers and public policy advocates. As The New York Times put it on the Business page last Sunday (in the lead to an article on exchange-traded funds) “A 128-crayon box of Crayolas does not guarantee a better drawing than the standard eight-crayon version, but any child will tell you that more is better.”

Very few applications can take advantage of these high speeds today. But consumers find security in buying something bigger and better “you never know when you might need it”. And there’s a “virtuous circle”: the more customers sign up for higher speeds, the more applications will take advantage of it and prove the customers right.

More customers are signing up for broadband service, more applications are taking advantage of broadband speeds, more connected devices are installed in each home and more people are using these devices. This creates a multiplicative effect on the demand for broadband network capacity. Customers will be very disappointed if service providers fail to deliver on their “faster, faster” promises.

Thus service providers are working to address two problems: increasing peak speeds to win new customers, and increasing their network capacity to keep their existing customers happy.

Telephone companies are replacing their ancient twisted pair wiring with fiber, some all the way to the home. Will cable operators need to follow the same course?

Choices for Cable Operators

Cable operators around the world have spent the past two decades modernizing and upgrading the physical cable plant; in the US alone, their investments have been about $100 billion. Do they now need to start again with pure fiber, or can they address both problems--peak speeds and capacity—with their existing plant?

Fortunately, the modern "hybrid fiber coax" (HFC) cable infrastructure has a huge potential capacity. More than three years ago, we wrote about some of the ways that MSOs could expand their available bandwidth without a significant rebuild of their plant (see Sources and References below). Most of these technologies are now on the market and some MSOs have deployed them for customer use. We received an update on these technologies at the recent NCTA show and have seen further announcements since then.

The End of Analog TV

Most of today’s cable capacity is still devoted to analog television. Over the next few years, analog TV will be phased out and the cable capacity can be redeployed for IP applications.

At hearings currently under way in the US, the end of 2008 is being proposed as the "date certain" for the conversion of all terrestrial broadcasting from analog to digital TV (DTV). Similar discussions are under way in Europe, with 2012 proposed as the end date for analog.

To put this in context, a typical US cable system carries about 80 analog channels. These occupy a large percentage (about two-thirds) of the cable "spectrum"; digital channels occupy the balance.

Digital television is highly compressed. If all 80 analog channels were carried in digital form, they would take the space of about 6 analog channels--leaving 74 analog channels to be redeployed for other digital services. Each digital channel carries 30 to 40 Mbps, so this would provide 2 to 3 Gbps for new digital applications--even more with new technologies.

The impending "end of analog" also provides an opportunity for operators to address the highly-asymmetric nature of HFC and provide more “upstream” bandwidth. Upstream signals occupy about 5% of today's cable spectrum; this could be increased once cable systems no longer have to carry rebroadcast analog television.

Many Alternatives for Increasing Capacity

Once analog TV goes off the air--by the end of this decade, if not earlier--cable operators will recover lots of channels. They will reassign most of the analog TV spectrum to digital services, and rebalance the downstream/upstream symmetry to better match the projected application mix.

But they can't wait that long. They already use digital channels for broadcast digital TV, video on demand, high-speed data and VoIP telephony. Most are deploying high-definition TV, which uses much more capacity for each channel. Some are starting to deploy video telephony and others are offering premium business services. Some have started to simulcast analog and digital channels so they can start deploying digital-only boxes. These all need more bits per second.

Fortunately, operators do not have to wait for the end of analog TV. They can "mix and match" from a smorgasbord of techniques to get more out of the physical cable plant. Their choices fall into four broad categories:

• Make more efficient use of the channels assigned to digital services
• Increase peak data rates
• Extend the operating spectrum of the plant
• Accelerate the digital transition

---

Improve the Efficiency of the Digital Spectrum

A significant portion of the cable spectrum is dedicated to digital services. Several techniques make more efficient use of this "digital spectrum":

• Spatial reuse multiplies the effective carrying capacity of the digital spectrum. Many services--such as VoIP, video on demand and high-speed data--are shared by many homes, with central resources such as CMTS ports assigned to meet expected peak demand. As more homes subscribe to these services, and more homes use them simultaneously, peak loads increase. Operators can add central resources and reduce the number of fiber nodes that share them--thus reusing the same space in the spectrum to serve more homes. When individual resources can no longer support the users in a single fiber node, operators can subdivide nodes into smaller groups of homes.
• Advanced modulation schemes such as 1024 QAM and Sub-band Division Modulation (SDM) provide higher data rates from existing digital channels. Compared with 64 QAM or 256 QAM modulation used today on digital channels, these provide 40% to 60% higher data rates. These advanced schemes require set-top boxes and cable modems equipped with the advanced schemes, and corresponding head-end equipment. Since they are more sensitive to noise, they require operators to further control noise in their plants.
• Advanced video compression schemes such as MPEG-4 Part 10/AVC/H.264 and WMV/VC1 promise compression ratios about twice as high as MPEG2 used today. This would double the capacity of digital video channels used for broadcast and on-demand video, and will be especially important as more and more channels are carried in high definition, which requires much higher bit rates than standard definition. These advanced schemes require set-top boxes equipped with appropriate decoders, and corresponding head-end equipment.
• Switched digital video allocates slots in broadcast video streams only to those video channels viewers have tuned to. By deleting unwatched channels from the broadcast stream, operators can assign less spectrum to broadcast video--or alternatively carry more video channels in the same spectrum.

---

Increase Peak Data Rates with "Channel Bonding"

The fundamental design of television forces the cable spectrum to be partitioned into fixed channels. In North America and Japan, with NTSC-based television, these are 6 MHz wide. These fixed partitions limit the peak data rate to about 40 Mbps today, perhaps 60 Mbps with advanced modulation schemes. (Europe, with PAL-based TV, uses 8 MHz channels and correspondingly higher peak data rates.)

The push is now on to combine several 6 MHz channels. This "channel bonding" would enable higher peak data rates and more efficient use of the channel capacity. There are two broad ways to do this:

• Combining several adjacent channels into a single wider channel would permit added efficiency by relaxing the need for a "guard band" between channels. Since existing user devices are only capable of tuning in "traditional" channel widths, this approach would require new user devices capable of tuning the widened channels, and would limit the bonded channels to use by new devices.
• Using several independent channels simultaneously is not as efficient, but could permit existing and new devices to use the same channels simultaneously. For this reason, it is likely to be the favored approach.

ARRIS recently announced a "FlexPath Wideband" product based on the latter technique. Its initial implementation embeds four cable modem chips in a new Wideband cable modem. Each chip can be tuned to a different upstream and downstream DOCSIS channel. The CMTS multiplexes data across up to four downstream channels, and the modem combines packets sent through the four channels to obtain a single higher-speed data stream. The same technique is used for upstream data. Using DOCSIS 2.0, the modem is capable of bi-directional operation at close to 160 Mbps. Interesting, the new modem requires a 1000Base-T ("gigabit Ethernet") computer interface since the conventional 100Base-T interface would limit the maximum speed to 100 Mbps.

ARRIS says the new modem will work with existing ARRIS CMTS equipment after a software download to support channel bonding, and that several MSOs are already running field trials of the new equipment. The new software supports existing single-channel modems on the same DOCSIS channels, using bonding only for the new wideband modems.

Although the announced product is limited to bonding four channels, ARRIS expects later models to bond up to 32 channels, permitting peak speeds of 1 Gbps.

CableLabs has been working on DOCSIS 3.0, the next generation of cable modem technology. CableLabs recently announced that it has selected the same channel bonding technique as part of DOCSIS 3.0.

Channel bonding gives cable operators "bragging rights" against competitors deploying fiber much deeper in their networks. It is also another technique to improve the efficiency of the network.

---

Extend the Operating Spectrum

The operating spectrum of cable plants is limited by the highest frequency passed through the optical and electrical portions of the network. In today's cable systems, this is typically 750 or 860 MHz.

Extending the spectrum provides more capacity without disturbing the existing spectrum allocation. There are two broad ways to do this:

• Extending the upper limit of the downstream spectrum, typically to 1 GHz, opens up additional channels for digital services. 1 GHz technology has been available for a long time, and some cable systems have been built with it. Upgrading an existing system to 1 GHz requires replacing nearly all the electronic equipment, and sometimes replacing part of the outside wiring as well. Most user devices are not capable of tuning above 860 MHz and would have to be replaced.
• Overlaying "out-of-band" spectrum adds equipment to the existing plant to carry bidirectional services above 860 MHz. Several approaches have been proposed for this--one by Narad Networks based on bi-directional Ethernet, the other by Xtend Networks based on analog block-shifting.

The Narad and Xtend approaches are similar in some ways, different in others. Both operate above the standard upper cable spectrum limit, and install new equipment in the cable plant to support bi-directional communications. Operators can deploy the new equipment selectively to provide specialized services--typically business services competing with telephone companies--and later extend it to cover more of the cable plant.

The approaches differ in their ability to be combined with other emerging cable technologies. The Narad system is specifically based on symmetrical Ethernet—-at either 100 Mbps or 1 Gbps--and is well suited for the delivery of IP services running over Ethernet.

The Xtend system adds more analog spectrum carrying many more 6 MHz channels in each direction, which can then be allocated to any application—-analog or digital video, DOCSIS, video on demand, etc.—-using single or bonded channels.

Cablevision Systems recently announced the completion of a successful trial and targeted deployment of a high-speed data service for business customers in Oyster Bay, Long Island. Using Narad's 100 Mbps technology, Cablevision is offering 50 Mbps committed information rate (CIR) services.

Cox Communications has been using the Xtend technology to support business applications in Pensacola, Florida.

---

Accelerate the Digital Transition

US homes have an enormous installed base of analog-only TV sets and VCRs, and many have recently-installed digital set-top boxes equipped for today's digital services. Any plan to change the cable plant is constrained by this installed base. The end of analog is coming soon--probably within four years--and most of those devices will still be in customer homes.

To avoid obsoleting customer equipment, operators may choose to offer a digital-to-analog signal converter box for each customer TV. Several companies have projected that such boxes could be in the $35 range at high volume.

Operators could choose to reclaim spectrum by phasing out analog TV before the official cutoff date, providing digital-to-analog converters for customers that want them. Since operators will probably find it necessary to provide these converters in the not-too-distant future, they may choose to accelerate the transition and get a huge gain in digital spectrum for new services. This is much more feasible economically for a relatively small operator, like GCI in Alaska (which has followed this path) than for a huge operator like a Comcast or Time Warner.

---

Spectrum is like Real Estate

Spectrum is the real estate of the cable industry and the MSOs are the developers of that real estate. Right now, much of the MSO’s real estate is being taken by widely-scattered single-family dwellings (analog TV). And there is some land still not built on (beyond the limit of today’s downstream spectrum).

The good news is that there are lots of perfectly-feasible ways to make better use of the real estate, as the demand for more and different kinds of housing increases (more users, higher bandwidth applications, more devices using the network).

These include:

• Sub-dividing larger properties to hold more single-family houses (“spatial reuse”);
• Building apartments (MDUs, flats)—which essentially stack houses on top of one another to increase occupancy (channel bonding);
• Building on the unused land (extending the upper limit of the downstream spectrum); and
• Providing incentives to move existing homeowners into new and better homes (transition from analog to digital to free up the old space).

Land developers look at the market needs and competitive situation and decide what mix of housing to build and which land to use. Cable operators have the opportunity to pick and choose the optimal combination of techniques described above to increase plant capacity--both to meet tomorrow's growth needs and to increase peak speeds to satisfy their most demanding customers. By doing this, they can create the technical underpinnings for a powerful marketing message.

---

Sources and References

We found An Evaluation Of Alternative Technologies For Increasing Network Information Capacity, by Ron Shani of Xtend Networks and David Large, a well-respected cable technology leader, to be very useful in framing the alternatives discussed here. The Shani/Large paper, presented at the recent NCTA show, goes into depth and provides an economic evaluation for each choice.

We have written extensively about the future of the cable plant in this newsletter and other publications. For further reference:

• In our report on the 2001 Western Cable Show, Cable's Magic Trick: How Bandwidth Keeps Growing (BBHR 12/19/2001) covered Switched Digital Video (BigBand Networks), Narad Networks, and Sub-band Division Modulation (Rainmaker Technologies, now called Broadband Physics)
• Our article More Solutions for Expanding Cable Bandwidth (BBHR 7/9/2002) covered Xtend Networks and updated BigBand, Rainmaker and Narad.
• Our article Broadcom -- Next-Generation Video Compression described the development of MPEG-4 Part 10/H.264/AVC -- three names for the same next-generation video codec.
• Our cover article for the April 2003 Communications Technology magazine Planning for the All-Digital Future--The End of Analog Television described the choices facing cable operators as analog TV comes to an end, including modifying the asymmetry of the cable plant.
• In our report on the 2004 National Show, Low-cost Digital Set-tops described several set-tops priced under $40.
• Our recent article Toward 100 Mbps discussed what we termed the "religious coloration" surrounding 100 Mbps.

( www.arrisi.com ) ( www.cablelabs.com ) ( www.naradnetworks.com ) ( www.xtendnetworks.com ) ( www.cablevision.com ) ( www.cox.com ) ( www.bigbandnet.com ) ( www.broadbandphysics.com )