Ultrawideband: The latest chipset unveiled

Bruce Watkins, president and COO of Pulse~LINK, talks up a new chipset that promises wireless networking 10 times faster than the fastest available today

Californian company Pulse~LINK has been developing ultrawideband systems for four years, and has recently closed a $30m venture capital funding deal. It has just announced a new chipset that will add gigabit networking to cable modems and can also send data over the mains at 200 megabits -- all at the same time.

ZDNet UK spoke to Watkins about the physics and politics of creating new markets.

Can you give us some more technical information on the chipset you've announced today?
This afternoon we take possession of the first test silicon for our Radio Frequency (RF) analogue chip. The chipset is a two-chip solution, with the RF analogue chip taking it down to base-band. The digital chip does channel routing, the MAC and so on, but the RF chip is the more complex chip. We'll be doing everything we can to speed things up but at the moment we plan on getting the digital part by October or November, and we will have re-spun the RF part by then. We could get lucky and have no changes to make on the RF chip, but we're not betting on it. The digital chip is easier to simulate, and we can make test version from programmable parts and discrete logic.

And what's this new chipset designed to do exactly?
We've demonstrated ultrawideband in the past over wireless, over cable systems and over the power supply network -- powerline networking. The unique thing we haven't talked about in the past is that the architecture that we're going with will let all three happen simultaneously with the same chip -- and it will be a good platform for additional things in the future.

The design pushes digital as far into the transmit and receive chain as it can, with multi-gigasample analogue to digital and digital to analogue circuits on the chip.  We convert the radio signal to digital as quickly as we can when receiving, and leave the transmit signal as digital for as long as we can before actually transmitting it. This forms an architecture that can be updated in software -- cognitive radio, in other words. If we want to put in either of the 802.15.3a UWB proposals, we can do it in software. We can accommodate Wi-Fi and any additional functionality, pretty much as far as we can see into the future, through software.

We'll be able to do anything through the antenna up from DC [0 Hz] to 6GHz, that's the top design frequency of the RF chip. Within those ranges, Powerline is DC to 30 MHz, Cable is DC-50MHz upstream, downstream 50-750MHz, but we can add higher frequencies. Wireless UWB is 3.1 to 5GHz, but we can go to 6GHz, leaving enough space to do 802.11a, if desired. It's not in our initial plan but we can do it if we like, although not simultaneously with the UWB functions -- they'd have to be unloaded.

We can't say too much about the performance at the moment, but across the system the gain is around 50 percent higher than either of the UWB proposals require.


What do you think of the current UWB standards battle?
We've been pushing the concept of common signalling mode (CSM) as a way to let lots of different standards co-exist -- our work in that area is a little more than three years old now. We didn't start out by thinking it would be a good way for dissimilar UWB devices to be aware of each other. John Santhoff, our chief technical officer, said there aren't good co-existence models out there, and we needed them. We knew that whatever we came up with today, in five years' time we would definitely have found something better. John asked: "How can we ensure that future generations of our technology can recognise older technologies, maybe even using the UWB link to upgrade them to newer ideas?", which is the concept where the CSM came from. Let's devise a low data rate architecture that could always be a common denominator.

How are you going to get anyone to agree to this idea?
We'll be at the ITU UWB conference in Boston in June, where we'll demonstrate our shiny new test silicon. We've got no shortage of interested companies, including some very significant ones who've been into our offices on a regular basis, some weekly, and our objective is to go showcase these things in Boston and begin coalition based standards building with a couple of companies. We'll start that process this summer, initially with standardisation for putting UWB on cable, powerline and wireless. It's like Bluetooth -- you go to a trade association management firm who have all the legal templates, can manage forums, web sites, logos and all that sort of thing. We've hired someone to do this.

So you'll have a snazzy name and logo?

Where will we see this technology first?
Take UWB over cable, which we've already deployed for government and security companies. We see it going to cable companies next, then into consumer electronics. There's a lot that needs to happen on the cable side of things. Wi-Fi is becoming a mature technology, we've all been reading for seven years about home gateways -- so why don't you see Wi-Fi inside set-top boxes? It adds cost, which the cable provider has to pay for, and it doesn't add anything that the consumer wants to pay for. Stuff like 802.11 doesn't have quality of service, which makes it difficult to sell to consumers. They don't want their streaming video interrupted every time the man next door browses the Web.

So things develop very fast, and cable companies find the cost of upgrading crippling. But with our chip, not only will it do all the wireless functions and give extra bandwidth over the cable system, it can be upgraded over the system itself.