In our report on the 2001 Western Cable Show last December, we wrote about several new technologies designed to expand the digital carrying capacity of existing cable plants without massive new rebuilds. The past six months have seen further developments; there are now half a dozen different technologies designed to squeeze more revenue-generating services over the existing plant.
Because it was designed for television viewing, the HFC cable plant is quite asymmetric - much more bandwidth is available downstream than upstream.
Cable operators are deploying many services which use both downstream and upstream cable bandwidth. Some of these new services (such as VOD) are very asymmetric, while others (such as telephony) are symmetric. High-speed data services were designed assuming that most of the usage would be web browsing or file downloading - both highly asymmetric downstream.
The load on the cable network is growing quite rapidly as more and more homes subscribe to cable modem services, use the services more of the time, and download larger and larger files. To make matters worse, many users have installed peer-to-peer (P2P) networking applications; if they choose to share files (often in violation of copyright laws) their homes become highly asymmetric upstream.
Most cable systems have been rebuilt during the past decade: cable operators have replaced most or all of the physical infrastructure of optical fiber, coaxial cable and electronics which carry the services. As operators anticipate the need for more capacity to accomodate growth, they would much rather add technology to the existing infrastructure than start rebuilding yet again.
Many companies are working to address this need. In "Cable's Magic Trick: How Bandwidth Keeps Growing" last December, we discussed BigBand Networks, Narad Networks, and Rainmaker Technologies. We'll update these three, and add three more: Terayon, Pulse~LINK, and Xtend Networks.
The technological approaches fall into three broad groups:
Several of these approaches could be used together for a multiplicative effect.
A long-time player in the delivery of digital data over cable networks, Terayon offers two solutions to help MSOs conserve their bandwidth. Their CherryPicker product, deployed in many cable systems, allows the real-time adjustment of digital video bit rates to accommodate a given amount of bandwidth. At the recent NCTA show, they showed their newest version combining four HDTV streams into a single cable channel, and also combining fewer HD streams with multiple standard-definition streams.
Terayon invented and has long been promoting "advanced S-CDMA," an alternative physical layer technology for cable modems. S-CDMA permits more efficient use of upstream bandwidth, nearly doubling the upstream data rates possible with earlier versions of DOCSIS. The most recent cable modem standard, DOCSIS 2.0, includes S-CDMA, and Terayon is positioning itself to be first in the market with DOCSIS 2.0 certified CMTS and cable modems.
Terayon has established a subsidiary, Imedia Semiconductor, to provide DOCSIS 2.0 chips and software to other cable modem vendors.
BigBand Networks is focused on getting the most out of the cable bandwidth assigned to digital broadcast multiplexes. Since our last article, the company has announced that their BMR technology has been deployed in cable systems serving more than 500,000 digital cable subscribers. They have also announced the use of their system to carry three simultaneous HDTV feeds in a standard 6 MHz TV channel, which previously could support only two feeds.
We recently interviewed Seth Kenvin, VP of Corporate Development at BigBand, to get an update. Seth told us that their installations are "rapidly approaching a million subs". We asked about more aggressive bit rate reduction and he said that operators might be able to squeeze more than three HD feeds in a channel if they (and their subs) were willing to accept some degradation in quality.
We were especially interested in what BigBand calls "switched broadcast services," based on the observation that the increase in TV channels leaves a lot of them unwatched at some times and places. A typical cable node serves 500 homes, of which perhaps 350 might subscribe to cable. Suppose the cable system is carrying 500 channels, many of them digital. Pareto's Law suggests that many channels are not being watched by anybody in a particular node. BigBand's system can delete all of the unwatched channels from the broadcast stream, automatically switching a channel on only when a subscriber selects it from the EPG. This frees unused bandwidth for other services such as VOD and high-speed data. Systems that exploit this capability could add a wide array of special-interest channels without having to increase the system's channel capacity, since channels will be switched only to those nodes where viewers have selected them.
Seth told us that the switched broadcast services have been demonstrated but not yet deployed. He said that operators expressed a lot of interest at recent shows and expects a deployment soon.
Any approach to transmitting digital information -- whether video, voice, or data -- depends on a modulation scheme. In North American cable systems, the downstream modulation is either 64 QAM or 256 QAM, which operate in 6 MHz channels with a payload of about 27 and 39 Mbps respectively and a spectral efficiency of about 4.5 and 6.5 bits per Hertz.
Rainmaker proposes to make two changes to current standards: use wavelet modulation rather than QAM, and to operate in wider channels than the traditional 6 MHz allocation. They claim that this approach will provide 170 Mbps downstream in 18 MHz (combining three 6 MHz channels) and 150 Mbps upstream; this provides an efficiency gain of 1.5 to 2 downstream and 3 to 4 upstream. This near-doubling of payload could be applied to any digital service currently carried over cable: data, video or voice.
Since our December article, Rainmaker has become a Panasonic Digital Concepts Company with an investment from Panasonic Ventures. Their current focus is on adding wavelet modulation to the global standards for digital video transmission. This is still an early-stage effort, and is awaiting completion of Series B financing to move the technology from the lab to a field trial.
Narad Networks' Broadband Access Network offers two options for switched IP services: bidirectional 100 Mbps (Fast Ethernet) and 1 Gbps (Gigabit Ethernet) service. They say both will operate on existing cable plants without any effect on existing cable services.
While it doesn't require a rebuild, implementing Narad's system for an entire cable system requires a substantial investment in additional technology throughout the system, adding equipment at each fiber node and replacing the amplifiers and other boxes between coaxial cable segments.
Rather than advocating the complete upgrade of a cable system, Narad has been promoting its technology for cable operators who want to offer services to small and medium businesses. To address the needs of a specific customer, the Narad equipment can be installed only on the specific fiber node and coaxial trunk serving that customer. This provides a fast return on the capital investment.
Narad and IBM recently announced an alliance to work together to help MSOs address the SMB market. This combines Narad's service delivery platform with IBM's software and services. To learn more about the alliance, we talked with Peter Gaucher, Director, Segment Strategy, IBM Pervasive Computing Division, and Ahmet Ozalp, VP Strategic Marketing at Narad.
Peter and Ahmet told us that many SMBs are struggling with IT complexity, and are looking for solutions to compete more effectively. Their solution is for the MSO to deploy Narad's technology selectively and use IBM's products to provide core and premium services to the target market - the "sweet spot" is businesses with 10 to 300 or so employees.
The MSO would provide core services such as Internet access, VPN, and T1 equivalents for voice and video communications. IBM has a growing portfolio of outsourced services through its "Manage IT for me" program, and would provide such premium services as web sites, email, application hosting and management, CRM as well as traditional back-office applications like payroll. These premium services could be co-branded with the MSO or sold under the IBM brand.
The Narad/IBM offering includes TDMoIP technology from RAD Data Communications to permit the MSO to offer T-1 and E-1 services in competition with the local telephone company. Unlike T-1 lines with fixed capacity, the TDMoIP approach provides bandwidth on demand and can pay back in 12 months.
We received some literature from Xtend Networks and arranged a telephone interview with Hillel Weinstein, their CEO, at his office in Tel Aviv.
Xtend's product line superficially resembles Narad's - it uses the bandwidth above 860 MHz to expand the capacity of a cable system without replacing any of the fiber or coax. The difference is that Narad's technology dedicates the extended bandwidth to a specific service - switched Ethernet - while Xtend's simply increases the available cable bandwidth which can then be used for any service or combination of services. Narad's technology is mainly digital, while Xtend's is analog.
Xtend's technology extends the cable bandwidth from the conventional 860 MHz all the way to nearly 3 GHz. The additional bandwidth can be used either symmetrically (about 1 GHz in each direction) or split asymmetrically. The added bandwidth could be used for additional digital multiplexes (for additional broadcast channels and/or VOD), for additional DOCSIS channels, and/or for switched Ethernet.
From our conversation with Hillel, the best way we found to think about the Xtend approach is as a "virtual dual cable" system. During the 1980s, the most advanced cable systems were built with two parallel cables, each operating up to 450 MHz (the bandwidth limit of technology at the time); half the TV channels were carried on each cable. Special set-top boxes were used to switch between the two cables as the subscriber selected a channel. Since a "cable ready" TV did not have the switching function built-in, the cable operator would provide an "A/B" switch so the subscriber could manually select the appropriate cable; many customers preferred to use the special set-top box on each TV.
The Xtend system can be used in a similar way. Xtend's XTB module, installed at the customer premises, acts as a "block converter" to shift the extended frequencies into the conventional TV band. The effect is identical to having a second cable, except that the Xtend cable has much higher upstream bandwidth. As with the "dual cable" approach, the subscriber can use a manual "A/B" switch. Hillel told us that Xtend plans to work with cable manufacturers to build the block convertor and switching functions into set-top boxes, so the "dual cable" is invisible to customers. Hillel projects that adding Xtend's technology to an existing system has a capital cost of about $200 per subscriber today, and will come down to about $100 per subscriber once XTB functionality is built into set-top boxes.
The net effect of Xtend's approach, once proven in the field, will be to give the cable operator more than twice the capacity at a much lower cost than a complete rebuild. But the operator does have to add equipment throughout the cable plant to gain the full benefit.
While Narad and Xtend add bandwidth at the upper end of the cable frequency range, Pulse~LINK adds in within the standard frequency band. They do this with ultra wideband technology, a novel technique approved in February by the FCC. While many companies have talked about applying UWB to wireless networking, Pulse~LINK is the first to announce plans to use it to carry data over existing HFC cable systems.
Pulse~LINK has just gone public with their plans, and we interviewed John Santhoff (founder and CTO) and Bruce Watkins (President and COO) to better understand their approach. John told us that their current prototype uses a very narrow pulse (1.5 nanoseconds) across the entire downstream frequency band to tranmit data at 400 Mbps. The pulse is transmitted at very low power and does not interfere with any of the programming already carried in over those frequencies. They say they've already tweaked the data rate above 1 Gbps and expect to get yet higher with a production product.
Since the UWB data is carried in the existing frequency band, no changes are required to the cable infrastructure. An MSO would need to install suitable equipment at the headend. In customer homes, a "UWB to Ethernet dongle" might be the first product before suitable chips were built into digital set-top boxes and cable modems.
Pulse~LINK is assembling the patent portfolio for UWB over cable, and expects to licence its technology to cable equipment manufacturers. Many companies will await field trials to see whether the technology works as promised.
Summing It Up
The companies making the most ambitious claims are all at early stages of their development. They need funding to prove that their approaches work as well in the field as on paper and in the lab. This is not the most propitious time for funding, and some of them may not survive even if their technologies are sound.
Nevertheless, we feel that cable operators are in a good position to fulfill the promise of broadband. All MSOs have made heavy long-term investments in rebuilding their cable plants; most are fortunate that they are well along and their investments are winding down just as the market has turned sour on spending for long-term rewards.
As many of the new applications -- especially high-speed data, video on demand, and telephony -- prove themselves in the market, cable operators will need more bandwidth than they have. They would like to pull back analog channels but will have to move slowly unless digital TV sets start selling much more quickly.
All of these new technologies suggest that MSOs will be able to wring a lot more bandwidth -- probably many gigabits per second -- out of the existing plant. Taken together, they promise to postpone the need for additional massive capital investments for many years.