Pulse~Link wants partners in radio revolution
Looking beyond wireless USB
UltraWideBand start up Pulse~Link was showing us its latest chip set this week, which takes it one step closer to its goal of having the first Cognitive Software Defined Radio on a single chip. This is on track for the first quarter of 2006. Right now all of the function it says it needs for UWB communication reside in seven separate chips.
Last month the company filed yet another 27 patent applications, and continues to see UWB technology as completely bare of patent applications except its own. It now has 200 patents either applied for or granted in this space.
It has also come up with yet another way of using its chipset which it is currently researching and says that the Digital to Analog Converter within its components is good enough to manage HDTV delivery as a TV tuner.
Given that the company hopes to take its silicon to market in the $30 range, and that prices for HDTV tuners range from $200 to $300 right now, it might be yet another market that the company has unearthed.
HDTV sales to date are around 6m devices globally, but set to double each year for the next few years.
COO Bruce Watkins told us that the company was still intent on being a chip designer, and that it didn't think of itself yet as a pure intellectual property play, but it certainly seems headed that way.
Like most such businesses, it has to make a certain number of chips and get them deployed before anyone else is likely to believe in the technology enough to take it seriously. But after that it will almost certainly not be able to grow fast enough to keep up with the various separate markets it keeps unearthing, and it will have to license its designs and intellectual property. We imagine a company somewhere between Qualcomm, with its CDMA patents and ARM with its ubiquitous core design in 80% of mobile phones, as the outcome.
But the value of the patents are in some doubt at this time, due entirely to the restrictions placed on UWB communication by the Federal Communications Commission in the US. The FCC has placed stringent power limits on UWB, reducing it in many people's eyes to a high bandwidth short range communication, expected to lead as a wireless USB contender. Ofcom in the UK in the last few weeks has put in place even more stringent power controls.
"There are even issues with that," says Watkins, "and the second Thursday in December the FCC will make a ruling on whether or not it will allow a 3 meter transmission range for UWB." If it fails to agree that step, then the work of the two major consortia pushing UWB for wireless USB, the Multiband OFDM Alliance of Intel and Texas Instruments and one led my Motorola, will have to think yet again about applications.
In digital media, a technology of this type should easily be able to eject Wireless LANs from the home. If consumer electronics companies adopt one or other of these approaches across the board, programming can be sent wirelessly around the home at speeds far exceeding Wi-Fi, for instance 4 or 5 HDTV signals, each taking up 22 MBPS, at once, with room to spare.
Right now Wi-Fi is pretty much established as the technology of choice for building a home network for entertainment, but UWB is a fundamentally more promising technology.
In its purest form it relies on brief, low power pulses of radio activity, and components that detect either its presence or absence. This concept is frightening to most radio frequency engineers as it denies the need for a carrier wave.
Setting up a carrier wave and modulating it has been the basis of radio communications for the past 100 years, and UWB promises to overhaul it, although it will have to get ratified by government regulators and in that sense it must get past a lot of vested interests in some very large companies.
Even the variation that the FCC has licensed is one that uses a particular frequency and it has had strict power and interference requirements placed on it, which have hamstrung the technology.
Watkins has lots of humorous stories about raising capital for the company and one meeting with executives of Qualcomm that demurred at having UWB explained to them (because of course they understood it), and then asked questions about what carrier frequency the product used and what power requirements it had.
The stunned Pulse~Link team tried to explain that it didn't use a carrier wave and that its power varied depending on how far the devices sent how much data. They were thrown out of the Qualcomm offices with a flea in their ear saying "Come back when you're prepared to share those secrets with us." Naturally Pulse~Link never went back and found funding almost entirely outside the US.
Watkins is adamant, and always has been, that Pulse~Link cannot go up against TI, Intel and Motorola. "They are all aimed at the wireless USB market, so we are aiming at distances far beyond that."
So far in the laboratory Pulse~Link has achieved 1.3 Gigabits per second at 3 meter distances which translates to 130 Megabits per second at 10 meters and 2 MBPS at 100 meters. It says it can deliver HDTV at 40 meters, so anywhere in a residential complex with wireless HDTV. "The requirement from the industry is picture in picture HDTV (two signals) over 25 meters, through at least one wall," says Watkins and that's certainly not a requirement Wi-Fi will meet any time soon.
Right now Pulse~Link makes its money from US Homeland Security projects, although you get the impression that this is short term expediency. It's pretty hard to intercept a UWB signal if you don't have the Pulse~Link chips, and if it's encrypted it's impossible. Also secure and military frequencies have different controls applied to them than those allocated for home digital media. Most radio devices would miss the signal entirely because it operates at such low power levels. "Airports find they can't use wi-fi for LAN communication because of the security elements, but they can use our communications," said Watkins.
The next phase of commercial exploitation for Pulse~Link is to offer a signal on a shielded digital cable TV loop that operates in its original pulsed format up and down the spectrum, but beneath the noise floor of the cable modem. On a 860 MHz cable this can deliver a spare 1 Gigabit per second and the cable companies can use this either in cable divided between the homes it reaches or sell spare capacity in the form of commercial T1 lines, to local businesses off the loop.
The costs of doing this, reckons Watkins, will be in the order of $100 per connection once he has his technology in a single chip and every one of the top 6 cable companies has offered him a standing invitation to come and try out the technology in their networks.
Pulse~Link took up that invitation on a small Californian cable TV company on a network of 4,000 homes using a hybrid fiber coax network. "Nothing unexpected happened in the trial," Watkins said "and we delivered 40 Megabits per second of spare capacity over the cable," which of course is not the 1 gigabit that he has promised which is why he cannot push this out tomorrow and make a fortune? Well development of new technology is so rarely like that.
"We are currently using components that were made to the FCC specifications and are using a fixed frequency range. So they can only operate in part of the spectrum of the cable and that piece of the spectrum needs to be free. In fact a fully populated cable set up means that we can't get the signal to work at all."
"We need another chip run, where we don't fix the signal to a frequency or we would end up having to implement a different set up for every different cable operator." So once again the FCC limitations have made life difficult for the new technology.
"The chips we are making now are only being made in samples of around 200 and we've had 3 foundry runs so far. We never expected these chips ever to have commercial applications, but the more we talk to potential customers the more we find new applications for UWB and in the end we may have to license this 7 chip set to a manufacturer and let them sell them, while we get on with integrating the design onto a single chip," he said.
But for now the 200 samples are being put into larger board level boxes and offered for trials with development kits.
One of the fascinating possibilities with the Pulse~Link Cognitive Software Defined Radio approach is that one chip can be used to offer UWB over cable modems, over powerline, any of the wi-fi frequencies or bluetooth ranges of radio and yet again UWB over twisted pair. Once Pulse~Link can get to 9 bit resolution (currently it's at 6) it will also be able to sense and send WiMAX and Cellular signals.
"The physical layer is the same, but there is only so much space to put the software onto the chip to define the profile. We could do two of those types of communication with the same chip at the same time, but not more."
There is also a limit to how many different complete application layers that a start up's software engineers can write, and Watkins is only too aware that Pulse~Link cannot do everything at once.
The core thrust of Pulse~Link is to drive the technology further. And if the FCC has placed strict power requirements on the technology, the only way forward is to make the components more sensitive.
"At 45dB to 50dB of gain on CMOS or Germanium the traditional oscillation that it sets up from the gain amplifiers kills the ability to see the signal. And yet for every 6 dB of gain you find, you can double either the distance that you can send data or double the data rate," he said.
Watkins goes on to explain that the Pulse~Link variable gain arrays have to rake through the core signal plus any bounced multi-path signals that it picks up, in order to extract a clearer and clearer picture of any given signal, and says that using these techniques it has managed to take the gain to 90dB.
"That gets us from the 10 meter level to the 100 meter level," he said, "and takes us into an entirely different business from the wireless USB guys."
Finally, the standards organization that Pulse~Link previously told us it is putting together to get behind all of these technologies is gradually coming together, according to Watkins. What he is after is a standards body cum implementers group, backed by some of the big CE companies. "We haven't got a name for it yet, but it is gradually gaining support and we should be able to announce it in February or March of next year."
Just how powerful the names of the allies in that group are may determine not just the future of Pulse~Link, but the future of radio technology exploitation on our planet for the next 20 years. So we await the announcement with bated breath.
Copyright © 2004, Faultline
Faultline is published by Rethink Research, a London-based publishing and consulting firm. This weekly newsletter is an assessment of the impact of the week's events in the world of digital media. Faultline is where media meets technology. Subscription details here.
Sponsored: Customer Identity and Access Management