Okay, first off,
this article seems like a reasonable overview, and demonstrates where Intel itself is pitching the tech.
Since it's only "wideband" in the microwave range, I might be a little optimistic about propagation; I'm not sure how well any of those signals make it through, say, drywall, or what receiver characteristics will be like. One potential benefit of a 'wideband' signal is that, theoretically, whatever particular splotch of spectrum makes it through can carry the bits; in practice, this depends on how well the whole "modulation technique" can take advantage of that fact.
(That's probably a slightly skewed way of looking at it versus the way the RF engineers themselves think, so for the record, I'll note I'm way behind on the 'theory' -- meaning the actual physics -- but probably out in Floidspace when it comes to what people want to do with things; graceful degradation is going to be hot in an all-digital, all-IP world, and UWB techniques seem like the most efficient way of getting it... except for their nasty risk of raising the noise floor for all other users of the spectrum.)
Now, to the meat:
Dan wrote:
Floid wrote:
The main thing is that WUSB will probably make Bluetooth look long-range -- and indeed, Bluetooth at least has a long-range spec if you're crazy enough to invest in making a WLAN with it.
Most longrange Bluetooth connection is replacment of serial connections AFAIK. I have seen Bluetooth Wlan hubbs but that was insanely expensive.
Well, if you're calling Bluetooth point-to-point, USB certainly has
even less features to handle multicast/anycast or 'ad-hoc' transmission. Of course, this also (to my knowledge) makes it relatively suited to UWB, because as far as I can tell, UWB is "hard enough" to implement (well, to implement with its presumed benefit) in point-to-point form.
[In other words, USB seems to be great for this because there's explicitly no allowance for inter-'peripheral' communication, all communication flows back and forth from the root of the tree. So using the new miracle modulation technique becomes an easier hack along worse-is-better lines.]
I could be talking out my backside, but that's how I see it... and yes, the 802.* camp are probably quietly working on the harder problem already.
I've heard of Bluetooth WLAN equipment (and looked specifically at one little home gateway device, back before 802.11 made it the equivalent of 5mbit SCSI-1 vs. cheap and prevalent UDMA) but I've lost track of how the links are actually managed, so maybe I shouldn't ramble on about that.
It's very neat, and will hopefully convince all the cool peripheral classes Bluetooth was supposed to bring us to become cheap and commodity, but if I understand the technology to be used, there are some fairly sane regulatory reasons why the power will not be goosed.
Bluetooth will probably become the lowcost alternative in the future.
Hmm, the "low cost" part is specifically not happening in the US-of-A; I'm not sure if the rest of the world is faring better. I want to like Bluetooth, but it's very much an added-cost thing here, while normal American humans don't know if their cellphones have it built-in (and thus don't even know to ask their cellphone vendors for one of those nifty earpieces that still bear a huge markup because the phone shops can pretend they're proprietary).
It's true that you can buy a Bluetooth dongle with only a little expense and trying, but you need to know it exists and have a good excuse -- USB, in contrast, everyone needs the moment they buy a gamepad or mouse, so people are forced to be aware of it (sort of like they're forced to be aware of IDE).
This article over here thus seems like some very wishful FUD to me, as much as I might want to believe. For instance, the author intentionally misinterprets Intel's stated goal:
First applications Despite what an Intel executive might believe -; he allegedly stated at the company's recent developers conference that WUSB would effectively kill Bluetooth when launched later this year -; the initial application for WUSB will be aimed at replacing the wired USB cable (which is perhaps where the Intel employee was becoming confused, albeit an alarming admission from a company attempting to make inroads into the wireless arena).
...but Bluetooth is so effectively (and sadly) sidelined in the American market that it's entirely true from his perspective. That commentator goes on to state:
What will also keep WUSB from competing with Bluetooth will be the implementation cost. Insiders say that to integrate WUSB into a host, such as a PC, will cost 3x more than Bluetooth given the complexity and the extra components required. This price differential will also be seen with WUSB dongles when used to replace basic USB cables, primarily due to the need to interface the WUSB technology into the existing interface requirements dictated by the wired equivalent.
Well, this is like saying x86 will never, ever beat RISC. Sure, the parts cost might be high at first, but so was it with Bluetooth, and getting people sold on wiring up a mouse is strangely easier (on this side of the planet) than getting them sold on a really expensive headphone to work with something they're already paying a monthly bill for. With all the usual Eastern companies leveraging their existing USB junk into wireless form, for use with cheap, commodity computers, palmtops, and game systems vs. "expensive," pay-to-play phones... and with most of the Bluetooth know-how unfortunately stuck in the darn 'expensive' West, Intel's packing a suitcase nuke, and this can only be bluster before the duck-and-cover.
If that camp emerges to mount a return volley (say, actually proving Bluetooth can exist at low cost), I'd love to see it... but the other problem is that WUSB throughput will mean devices can be dumber (browsing the web over Bluetooth will require a browser; over WUSB, you can surely make a
640x480 VNC display pretty dumb and cheap), so what gets spent on throughput can be saved on local grunt. "We've heard that before," you might say at first... but take a look at the steaming rubble of the embedded space, and it's pretty clear that licensing and hacking an OS and components is one of the big expenses; pushing the software problem back to the 'big' desktop OS that at least everyone's familiar with would probably a be good idea, at this point.
[Unless companies like, say, our pony were to suddenly emerge from hibernation and prove it reasonable -- we've just heard not to expect that in the near term* -- or people would wake up and notice OPIE or QNX6 are pretty darn good-enough... but none of the
cheap engineers seem to understand anything beyond Windows, argh!]
... however, when the "peripheral" is a Palm or an iPaq, someone's certainly going to find a way to do it. For the moment, most of the networking geeks have congregated around things called "ethernet," but that's just because competing link-layers haven't been sufficiently cooler or cheaper.
Certainly seems like a way to bring wireless to my old PocketPC thats incapable of using SDIO.
Indeed, though I'd expect that end will carry the same markup as other cable-replacements at first, "because they can." The peripheral end isn't the crack they need to sell to position the tech, but for actually being "USB," the bridges should be relatively simple, and I'd assume reach the "canned ASIC, $5 for a no-name adapter on Pricewatch and you can even expect it to work" level a lot faster than most of what we've seen before. (Software is expensive, and flexible technologies have the unfortunate habit of leaving more room for software issues...)
*No insult meant, considering I just outlined why it's near impossible to do anything 'right.' ;-) I just hope whatever route they
do take to is sufficiently cool and open-ended enough that I don't have to bugger off and wish I'd been wasting my time learning Java, instead.
