SCTE Member Since 1996 Title: President/CEO Broadband Physics Broadband Background: Laubach previously served as CTO of Rainmaker Technologies, a predecessor to Broadband Physics. He co-founded cable modem company Com21 in 1994, going on to serve as VP and CTO. Prior to Com21, he held management and engineering positions at Hewlett Packard. In 2004, the SCTE honored Laubach with senior member status. You spoke on Dan Pike’s panel at NCTA. What was the purpose of your paper? Sub-band division multiplexing’s (SDM’s) sub-band independence property makes it more tolerant of the interference caused by cable’s composite noise environment. Composite noise manifests itself as very narrow band spectrum interferers (10 kHz to 30 kHz wide) that persist for some to many tens of microseconds. The beats are caused chiefly by the intermodulation between analog TV channel video carriers and are called composite second order (CSO) and composite triple beat (CTB) effects. Intermod beats are also created by dispersion effects in fiber-optic cables. Is this one of the dilemmas: Higher order QAM is more susceptible to noise and impairments? Higher order modulations are more susceptible to noise impairments—that’s physics. However, modulations have their unique inherent characteristics (that is, the math used) that cause them to react differently to the same noise impairments when trying to get more bits per second through the channel. Based on the inherent properties of its math, SDM can go one to perhaps two orders of modulation depth better than QAM in the same cable environment. Is it simply the nature of sub-bands to be immune from impairments? Immune is an absolute, which physics doesn’t permit. Better to say that SDM is inherently more tolerant of certain well-known cable impairments as compared to what is in use today. How close to Shannon’s Limit do you want to get? Shannon’s Limit is a very harsh mistress. From a purely theoretical view, SDM can be pushed and band-aided to get close to that 20+ bits per second per Hz holy grail. The reality, however, must always have a consciousness that continually asks, "Yeah, but at what cost?" Cost isn’t always measured in dollars. It’s a multi-headed beast that includes price, information rate, silicon die size, interactive performance, power consumed by CPE, amplifier price, plant cable price, and price of servicing to maintain the cable plant performance, etc., as some of the more popular heads. SDM’s value is that for the same approximate price of QAM CPE, SDM CPE will deliver twice the data capacity as compared to 64-QAM with the ability to turn on 6 Mhz of spectrum up to 18 MHz of spectrum at proportionally higher information rates and without requiring any changes to the cable plant. The plant stays the same; it’s just the dongles on the end that change. What were you demonstrating in CableNET? We demonstrated SDM running at 12 bits per second per Hz delivering HDTV MPEG2 video between a headend transmitter and a pedestal receiver over approximately 60 feet of coaxial cable. We were moving just under 70 Mbps in a 6 MHz RF channel centered at 12 MHz. We call this our "baseband" configuration, and the frequency space is within the cable upstream RF spectrum. Twelve bits per second per Hz is equivalent to 4,096-QAM. We showed that SDM works and that the SDM engine will be capable of being turned up to a modulation depth never before seen on cable. Typically, most cable deployments will run at 10 bits per second Hz (equivalent to 1,024-QAM). Once having deployed SDM, there will be one more turn on the same hardware if noise conditions permit. Did you see anything especially intriguing at CableNET or elsewhere at NCTA? I was tethered to our pedestal at the CableNet pavilion and didn’t get around much. However, I do know SDM was the only new physical layer technology being shown for "last-mile" cable plant. How does sub-band division multiplexing compare with the DOCSIS 3.0 wideband proposals? DOCSIS 3.0 wideband is an inverse multiplexing solution that uses today’s physical layer technology. It creates a larger logical channel, but does not increase the underlying efficiency of the cable plant. SDM is a physical layer technology. Deploying SDM first would hold off the need for deploying channel bonding and would maximize the efficiency of the plant. After that, a layer of channel bonding could be added in the future if needed. My personal view is that it is much better holding off pouring concrete for the new dance floor until the physical foundation is the firmest one can build. Would an SDM deployment be targeted toward particular applications? SDM is both a downstream and upstream technology. With sufficient resources and industry acceptance, SDM could be ready for wide-scale custom semiconductor deployment in a year-and-a-half to two years. That schedule would meet a DOCSIS 3.0 timeframe if SDM were selected for specification. A noncustom semiconductor deployment could be ready much earlier for commercial services. It really depends on how the cable industry and cable product vendors would like to roll out SDM. Many options are available. Are you still racking up patents? Or are you moving beyond R&D mode? Patents rack up on their own in parallel if you let them. We are moving from R&D mode to demonstration mode. Time to get SDM on cable speaking for itself. Editor’s note: To order a copy of this year’s NCTA Technical Papers in either CD-ROM or book format, visit the NCTA publications site, http://www.thenationalshowstore.com.