Was WiMedia UWB ever right for Bluetooth?
Recent murmurings have suggested that despite losing it's place as Bluetooth's high speed data channel of choice, Ultra-wideband technology (UWB) may not be dead after all.
What is puzzling is that the most strident support for UWB is coming not from the WiMedia Alliance companies, but from proponents of the alternative version of UWB technology, the companies that were once linked as the DS-UWB Forum. And, the suggestion seems to be that by going with WiMedia's solution, the Bluetooth SIG had backed the wrong horse.
Here's our point - at Incisor we don't know whether the Bluetooth SIG made the right decision, or not. We're not clever enough to be able to analyse the technical proficiency and fitness for purpose of the two erstwhile UWB combatants. Which was/is best for Bluetooth? We don't know. Fortunately, we are well connected with a lot of people that are better qualified than we are, and we are happy to provide this forum for discussion.
Kazimierz "Kai" Siwiak, the CEO of wireless consulting company TimeDerivative Inc., for example, has recently commented that the combination of WiMedia's OFDM-based UWB was never the right choice for the high speed version of Bluetooth - known as Bluetooth 3.0. His comments follow:
A while ago, the Bluetooth SIG made a tentative selection of the WiMedia variant of UWB as a candidate for a high speed option in Bluetooth. The choice is not surprising, since the key players in WiMedia and the key players in the Bluetooth SIG tend to overlap; it is thus natural that they should try to seek common ground.
However, that choice might be considered a very weak choice. The WiMedia variant of UWB is actually an OFDM system in which the 500+ MHz wide OFDM symbols can be hopped every symbol among channels that are 512 MHz apart. It is "UWB" as a consequence of occupying more than 500 MHz bandwidith by transmitting OFDM symbols comprised of about 120 sine-wave carriers spaced about 4.13 MHz apart. Efficient base band processing of this OFDM system, and in particular symbol-rate-frequency-hopping of this OFDM signal, has proven to be a very difficult problem to solve. This is especially true from the power drain point of view. In that regard, it is indeed a poor fit for Bluetooth, or for any solution which seeks to address high speed data needs in the hand-held battery powered market.
Run off from current consideration, but nonetheless much more technically and market-wise appropriate, are solutions that obtain their UWB bandwidth as a consequence of their short pulse duration. These systems tend to scale their battery power requirements in proportion to data rate, in contrast to the OFDM approach which always requires significant processing effort. The pulse-based systems can be built with power drain economies that are a good fit for high speed Bluetooth. In fact, the needed technology is already well defined and described in the IEEE 802.15.4a-2007 standard for wireless personal area networks, where low power drain is a key necessity. Perhaps that body of work might provide a viable source of technology that can be adapted to fit high speed Bluetooth needs.
These are controversial views, but are they correct? Is this just sour grapes?
Incisor invites parties from the Bluetooth community, WiMedia companies and those siding with Kai (perhaps members of what was the DS-UWB Forum), to comment here.
What is puzzling is that the most strident support for UWB is coming not from the WiMedia Alliance companies, but from proponents of the alternative version of UWB technology, the companies that were once linked as the DS-UWB Forum. And, the suggestion seems to be that by going with WiMedia's solution, the Bluetooth SIG had backed the wrong horse.
Here's our point - at Incisor we don't know whether the Bluetooth SIG made the right decision, or not. We're not clever enough to be able to analyse the technical proficiency and fitness for purpose of the two erstwhile UWB combatants. Which was/is best for Bluetooth? We don't know. Fortunately, we are well connected with a lot of people that are better qualified than we are, and we are happy to provide this forum for discussion.
Kazimierz "Kai" Siwiak, the CEO of wireless consulting company TimeDerivative Inc., for example, has recently commented that the combination of WiMedia's OFDM-based UWB was never the right choice for the high speed version of Bluetooth - known as Bluetooth 3.0. His comments follow:
A while ago, the Bluetooth SIG made a tentative selection of the WiMedia variant of UWB as a candidate for a high speed option in Bluetooth. The choice is not surprising, since the key players in WiMedia and the key players in the Bluetooth SIG tend to overlap; it is thus natural that they should try to seek common ground.
However, that choice might be considered a very weak choice. The WiMedia variant of UWB is actually an OFDM system in which the 500+ MHz wide OFDM symbols can be hopped every symbol among channels that are 512 MHz apart. It is "UWB" as a consequence of occupying more than 500 MHz bandwidith by transmitting OFDM symbols comprised of about 120 sine-wave carriers spaced about 4.13 MHz apart. Efficient base band processing of this OFDM system, and in particular symbol-rate-frequency-hopping of this OFDM signal, has proven to be a very difficult problem to solve. This is especially true from the power drain point of view. In that regard, it is indeed a poor fit for Bluetooth, or for any solution which seeks to address high speed data needs in the hand-held battery powered market.
Run off from current consideration, but nonetheless much more technically and market-wise appropriate, are solutions that obtain their UWB bandwidth as a consequence of their short pulse duration. These systems tend to scale their battery power requirements in proportion to data rate, in contrast to the OFDM approach which always requires significant processing effort. The pulse-based systems can be built with power drain economies that are a good fit for high speed Bluetooth. In fact, the needed technology is already well defined and described in the IEEE 802.15.4a-2007 standard for wireless personal area networks, where low power drain is a key necessity. Perhaps that body of work might provide a viable source of technology that can be adapted to fit high speed Bluetooth needs.
These are controversial views, but are they correct? Is this just sour grapes?
Incisor invites parties from the Bluetooth community, WiMedia companies and those siding with Kai (perhaps members of what was the DS-UWB Forum), to comment here.
posted by Vince Holton at
11:41
6 Comments:
The Failure of WiMedia with Bluetooth was not a failure of UWB. It was the failure of a specific implementation of UWB. WiMedia's architecture and implementation traded away all the unique benefits of UWB when they decided to go with the familiar but inappropriate use of OFDM for UWB. OFDM is great for narrow band applications such as WiFi, DSL and HomePlug. OFDM was a poor choice for WiMedia and definitely wrong for pairing with Bluetooth due to it's high power consumption for powering the Gigabit per second+ sampling rates required for the DAC's and ADC's that generate and recover the OFDM modulation. A true "Impulse" based UWB solution would be a much better pairing with Bluetooth consuming a fraction of the power that OFDM would. The choice of 802.11n to pair with Bluetooth is going to experience similar problems as WiMedia. OFDM is a great modulation for narrow band applications requiring tens of MHz, not so for hundreds of MHz.
The BT SIG decision to select MB-OFDM instead of some form of single carrier (i.e. true, IMHO) UWB based on DS spreading or CDMA was a political and business decision. The OFDM decision never went through the regular SIG technical review processes in the core BT technical groups. The decision also knowingly created an unprecedented situation where an outside group (members of WiMedia Alliance) controlled IP essential to future BT implementations. IMO, the MB-OFDM approach was technically inferior and, as we have seen, never resulted in commercially viable silicon despite 100's of $M in investment.
The FCC's original action to allow unlicensed intentional UWB emissions is still in place and this ruling has been demonstrated to allow very high rate, short range comms ideal for commercial multimedia applications. It remains to be seen whether the full potential of this ruling is realized. Concerns about interference to other devices that led to more restrictive regulations overseas were exaggerated and/or unfounded, IMHO.
-MLW
For those who wish a little light reading on this topic, there are 5 years of debate proceedings in IEEE 802.15.3a and more than 7 years debate in the FCC, CEPT and the ITU. There are meaty arguments to support just about any position that you wish to take. I trust that we all did what we thought best with the data that we possessed.
I personally wish you all the best of luck and an excellent holiday season. Focus on future opportunities, not past conflicts.
That is 2-1 to the DS- guys. Come on you WiMedia companies!
Posted on behalf of:
Jim Lansford
CTO, Alereon
Even though I'm CTO of a WiMedia company, I actually agree with Kai --- to a point. There was some good work in 802.15.4a, but the data rates are much lower than WiMedia; impulse radio is great for sensors and for positioning...but not for high speed. The fundamental problem is delay spread. Single carrier or impulse radio UWB is great at short range (<2 meters) or at data rates below 30Mbps, but the intersymbol interference at longer ranges and higher rates makes impulse/single carrier very difficult. We had many presentations in 802.15.3a that established the need for rake receivers in single carrier or impulse radios, and not just a few taps as in IS-95, but 20 to 30 or more to establish an equivalent link margin for 100Mbps at 10 meters. Ask the former Xtreme Spectrum guys how hard it is to build a rake receiver for UWB...and they were only doing 4 taps, but could never get it working well. A 32-tap rake for impulse/single carrier for these bandwidths would be a complex power hog (and BTW, would require gigasample ADCs at
3 bits or more of resolution, so you don't buy much power savings in the converters). For a selective fading channel --- and UWB is DEFINITELY a selective fading channel --- OFDM is a highly optimized technique for multipath mitigation. It works very well, and we have chips in production that go faster and farther than any equivalent impulse/single carrier UWB chipset ever did. If you want wide, fast, and long, OFDM is the best technique available. As Kai mentioned, the power consumption of the WiMedia PHY does not scale linearly as a function of data rate, but that merely says the technology is most power efficient when running as fast as possible for a given link. UWB as an alternate MAC-PHY for Bluetooth should be as an "afterburner" where you want transfers in excess of 100Mbps, and the WiMedia solution measures up better than anything else for that power-price-performance space.
In the end, the problems with UWB in Bluetooth have not really been technical, but political. There were a lot of hard feelings after the standards wars, and people had difficulty getting past that. It became a religious issue, and it's hard to overcome that mindset. There were also some growing pains with first generation chipsets that didn't help; these issues had nothing to do with the PHY, but the USB "wire adapter"
protocol, and those issues are gone; they aren't really relevant to the Bluetooth AMP architecture anyway. Today's WiMedia chipsets can deliver 200Mbps or more at the top of the MAC, and power consumption has dropped to about 1/3 of the first generation chips. It's a great high speed WPAN solution for video streaming and mobile sync. So the technical problems are largely solved, but the political ones? Time will tell....
Very, very interesting. And I have to tell you that I have learned an awful lot more off the record too.
I am aware that there are WiMedia companies that are still actively developing their solutions, and that there will be some activity at CES. The chosen modus operandi of these companies seems to be one of 'stay quiet and just get on with it'. Is this the right policy, or does there need to be ongoing positive PR in order for the potential customers to feel confident in buying into UWB?
Post a Comment
<< Home