During the first three weeks of February Terayon, an established vendor of cable modems, saw its share price increase over 250%. George Gilder, a longtime technology consultant and commentator with the power to make or break high-tech companies, ran a story in his newsletter (The Gilder Report) that likened Terayon to Qualcomm, a company that has seen it's own valuation increase by over 900% over the past year. This, along with the fact that both companies share a similar type of spectrum access technology in their products, contributed greatly to Terayon's moonshot.
The crux of the matter is CDMA, or code division multiple access, which the two aforementioned companies' businesses revolve around. CDMA is simply a method of spreading numerous signals across an available range of frequencies. It was originally developed for use by the U.S. military, as spread spectrum communications are inherently difficult to disrupt and intercept by non-intended receivers.
While CDMA's traditional usage has been in wireless communication systems, Terayon decided to adapt the technology to work over a coaxial infrastructure. Because CDMA spreads transmissions over the full available spectrum, noise or interference in any one particular area of that spectrum can be mitigated by leveraging other, cleaner areas of the available spectrum. This primary feature of CDMA is what allows Terayon's TeraPro to be exceptionally resistant to noise and interference. It's Terayon's greatest asset. Their TeraPro cable modem is certainly the best in its class with respect to being able to function under what would normally be unbearable line conditions.
So impressive is Terayon's product that the company was selected in late 1998 to co-author the DOCSIS 1.2 specification with Broadcom. Since that time, Terayon has committed its proprietary implementation of CDMA, called S-CDMA, or synchronous code division multiple access, to the royalty-free DOCSIS intellectual property pool.
Now, the above facts should be enough to make anyone jump with joy and bet the future of cable modem technology on Terayon. Rogers Cable did just that with their March 1999 deal with Terayon. The TeraPro has been exclusively deployed to all new Rogers@Home subscribers since mid-September of last year, in place of the Nortel (formerly Bay Networks) LANCity. The unfortunate truth, however, is that there are four glaring problems regarding the deployment of the TeraPro.
First off, the TeraPro yields poor network performance relative to the LANCity when both the upstream and downstream channels are utilized simultaneously. This loss in performance isn't readily apparent when doing web browsing, but is certainly noticeable when transferring files and using any time sensitive applications like streaming audio, or games. The biggest problem here is that even a very modest utilization of the upstream channel while performing a concurrent downstream action, will cause a performance hit.
Next, the average minimum latency (i.e. the time it takes for the cable modem to send and receive a reply from a local node) of the TeraPro is more than an order of magnitude higher than that of the LANCity. Indeed, probably the greatest asset that the LANCity possesses is the ability to yield extremely low latency and thus, maximum network performance under suitable line conditions. It's very likely that the ability to provide low and steady latency, not high bandwidth, will determine the leaders in the broadband arena over the coming years.
Third, the TeraPro will adapt its downstream rate depending on how noisy line conditions are. So there might be a number of subscribers living in one neighborhood who are able to achieve maximum downstream rates of 300 Kilobytes per second, while subscribers living in another neighborhood with a higher noise floor will only be able to achieve maximum downstream rates of 90, or 100 Kilobytes per second. Of course, both sets of subscribers receive identical monthly bills and the ones exhibiting the much poorer service aren't entitled to any kind of service credit. The TeraPros have the effect of sweeping ingress noise and interference problems aside and giving the MSO less of an incentive to improve their infrastructure in order to keep the noise floor to an absolute minimum.
Last, the TeraPros allow for much more aggressive loading of data segments than the LANCitys do. Utmost care and prudence should be taken by the MSO to monitor segment performance and split any segments that show even the slightest signs of overloading. One crucial point to keep in mind is that, regardless of the cable modem infrastructure involved, there is no free lunch with respect to loading segments. The TeraPro may allow for a more stable connection than the LANCity on a data segment loaded with 300 subscribers, but overall performance is still likely to be poor.
A good question to ask at this point would be, what in the TeraPro causes these faults and features to exist? A simple explanation would be that they are manifested as a result of Terayon's implementation of S-CDMA. The network performance and latency problems are almost certainly the direct result of this. CDMA, in general, is very accommodating to loading available spectrum with a great deal of low bandwidth traffic, without losing stability. It must be stressed, however, that this, coupled with the TeraPro's rate adaptive ability, can end up severely penalizing subscribers if not handled appropriately.
The LANCity is able to yield such high levels of performance because of its simple signaling mechanism, CSMA/CD (carrier sense multiple access with collision detection). This is the identical signaling mechanism that is used in conjunction with every Ethernet product on the face of the earth. It's cheap, it's ubiquitous, it's time-tested and performs extremely well provided that there isn't a significant contention for bandwidth. Yet, the LANCity is currently the only deployed cable modem that I'm aware of which makes use of CSMA/CD.
While it's true that the TeraPro has its performance faults, it's still one of the best, most full-featured cable modems out there. It's just that the aging LANCity, with its very impressive performance, ease of administration and simple architecture, seems to have been all but forgotten in this era of DOCSIS standardization. What's worse is the disturbing trend of cable modems which are being released that place less and less emphasis on the reverse path. Any MSO that has to rely on QAM (quadrature amplitude modulation) to achieve 10 Megabits per second in the 5-42 Mhz range, is just asking for trouble.
Is spectrum really that rarefied that 5 or 6 MHz can't be allocated to support the reverse path of an Internet service? All of these issues must be thoroughly resolved before cable can be declared the broadband Internet access medium of the future. My only hope is that this future holds some place for a cable modem design that is derived from the LANCity. It's not easy to refute the benefits of a cable modem which is based on technology that is present in roughly 90% of the world's local area networks.
Christopher Weisdorf is a Rogers@Home customer and technical director of the Rogers @Home Users Association (RHUA). Views expressed are solely those of the writer.