An Introduction To Wireless USB (WUSB)

An anonymous reader writes "This technical whitepaper by Rafael Kolic, a technology marketing manager in Intel's Corporate Technology Group, introduces Wireless USB (WUSB) and explains how it will impact device performance and mobility. The latest iteration of USB technology, WUSB will offer the same functionality as standard wired USB devices -- but without the cabling."

Comments

Doesn't this already exist?

[JayDiggity]

Ummm.. don't we already have something for that called Bluetooth? Hrm.

From the Bluetooth SIG Mission Statement:
Develop, publish and promote the preferred short-range wireless specification for connecting mobile products...

Re:Doesn't this already exist?

[Agent Smart]

WUSB will replace Bluetooth

From an interview with Intel CTO
He indicates that it'll have better performance and lower power consumption. Eventually.

Re:Doesn't this already exist?

[Boltronics]

Actually, 802.11g can transfer up to 54Mbps (unless you don't have strong signal strength). But according to D-Link website 108Mbps wireless networks may be on the way in. I want...

Re:Doesn't this already exist?

[CowboyNick]

My point was, the cause of latency it totally different in both examples. Your Bluetooth headset only has to transmit data a few feet to get to your cell phone. If you actually stand next to the person you are calling you will notice there is about a 1/4 to 1/2 second delay between what you say and what is heard at the other end (part of this is due to the audio signal being encoded to digital and back). However your head set will not create a noticeable delay vs. using your cell phone's built-in microphone and speaker directly. When you are using your USB wired head set, there is a totally different transmission network that was not even designed for voice communication, so your results will vary depending upon network conditions. The PSTN and cellular networks are designed to transmit voice reliably. The delay that you are hearing in this case is not at all related to it being USB.

Re:Doesn't this already exist?

[fyonn]

actually, I think that the adoption of bluetooth has been pretty quick and fairly widespread. I'm in the UK and a large percentage of mobile phones come with bluetooth already and most of the ones due out soon include it. most of the current crop of pda's either have bluetooth on board or can be upgraded via a card to support it. you can buy countless BT headsets and hands free kits for your car and many laptops have it onboard.

thats pretty much bluetooth's market sewn up. while it's true that most desktop's don't come with BT as standard, some do and it's only a 17 usb dongle away. I got a dlink BT dongle for my mac and it works like a charm to speak to my siemens mobile phone, I just wish the prices of BT memory sticks for my clie weren't so outrageous.

sure it's not overly fast but it is easy and convenient and works. I thought it was doing pretty well.

dave

PS. saying that, has anyone else seen a bit of incompartibility between apple's bluetooth stack and a siemens mobile phone? I can sync over BT fine but sending txt's via it is problematical. it says everything is fine but alot of the time the txt's don't get sent. anyone else?

Uh...this could be

[iminplaya]

a problem.

bluetooth killer!!!!

[nil5]

According to the information published by different media, Wireless USB will support up to 480Mb/s transfer speed over 4 meters and up to 110Mb/s over 10 meters.

Wireless USB will be based on the multi-band OFDM technology backed by an industry alliance that includes Intel. It also blends in the common UWB radio platform defined by the WiMedia alliance. The UWB and wireless USB specifications are in the early stages of definition. Systems using wireless USB are not expected to ship until sometime in 2005, CommsDesign web-site notes.

Basic Specs from the Article:

[G4from128k]

Speed? 480 Mbps (USB 2.0 equivalent)
Security? unknown (better be good)
Power requirements?100-300 mW
Range? 10 meters
Cost? unknown
Number of Devices? 127

Highlights

[the JoshMeister]

Here are some highlights for those too lazy to skim the article (or in case it gets Slashdotted)...

• The Wireless USB Promoter Group consists of (of course) Intel and MS, as well as HP, NEC, Philips Semiconductors, Samsung Electronics, and Agere Systems
• WUSB hosts can connect to as many as 127 WUSB devices at a time
• Targeted bandwidth of 480 Mbps (same as USB 2.0)
• The biggest selling point seems to be that it "will offer the same functionality as standard wired USB devices but without the cabling"

Re:How does this excel existing standards?

[Crypto Gnome]

(sigh) RTFA duude!

• target 480Mbps (ie USB2) expect to hit 1Gbps at some point
• Looks like they're spec'ing minimal "basic security" for the connection itself, as they go on to say "Higher levels of security involving encryption should be implemented at the application level", implying that encryption will not be part of the WUSB spec itself.
• "For example, typical PDAs use 250-400 mW without a radio connection, while typical cellular phones use 200 mW-300 mW with the primary WAN radio. Adding a WUSB radio should not increase power requirements any more than existing wireless technologies already employed today. "
• "connectivity at less than 10 meters"
• Cost is a result of production, not (necessarily) something you can just spec in the design. Although they do say "Low-cost implementation of WUSB will also be important to the successful integration of the technology. Implementation will follow the wired USB connectivity models as closely as possible to reduce development time and to preserve the low-cost, easy-to-use model, which has become pervasive in the PC industry. "

Re:So...

[mcbridematt]

probably the standard wireless mouse. USB in it's own right eats up CPU power, which in turn, eats fps. WUSB probably won't be different. The usage for USB mice wouldn't be much, but any extreme gamer might take it as too much =(

Not to mention any lag factors with wireless.

Re:Will die like bluetooth

[djupedal]

One of the biggest advantages of USB is the power that flows through it

FireWire, with 12V, is much more capable. Examples are pocket drives....the USB models all require a brick/adapter. It's a bit of a stretch to give USB too much credit in this department :)

Re:Neat

[Have Blue]

Actually, the previous article was really "Bluetooth is dying because WUSB will kill it", so it's not so much a conspiracy as just a glut of news surrounding Intel's announcement.

Re:"...without the cabling."

[gabebear]

These things are going to eat several times more power than a Bluetooth radio. The article says: "The power target for WUSB radio will be introduced at less than 300 mW and drive to a target of 100 mW over time."

That's the same as lower power 802.11 cards

Bluetooth chips generally eat less than 40mW, some as little as 20mW. I wouldn't put WUSB in a cell-phone.

Re:Ok...

[ummcdou4]

Interesting idea,

One problem might be that in order to drive an inductor or other type of charging mechanism you would necessarily need some friction from the mouse ball, the more friction, the more power generated.

I'm not sure how much friction a user would put up with using the mouse, certainly gamers would have none of it.

Supplying power

[Anonymous Coward]

About the problem of devices that draw their power from the USB cable, and not wanting to add cords or batteries to these items...
What about supplying power via a pad that you could simply set your devices on?

For example, something like at http://www.splashpower.com/

This would allow the devices to still not have a battery or power cable, they'd just have to sit atop one of the "power pads".

Eh?

Re:Wireless keyboard loggers, anyone?

[gnu-generation-one]

"I can only hope and pray that wireless USB will be very very secure."

WUSB is intriniscally insecure, just like TCP/IP. If you want a secure connection, you'll need to run a secure protocol over the WUSB connection, such as SSH, SFTP, etc.

Alternatively, you can use symmetric encryption if a paired set of devices happen to share a key. Imagine touching two devices together and pressing "generate key" on one, and "receive key" on the other. Instant high security.

It's probably good that it's not being included in the lowest-level transport protocol, because that'll mean that you can build a device without spending 6 years trying to understand the encryption provisions in the speicifcation. Look at WiFi for example -- it claims to have encryption, but you need to run SSL anyway because it's so weak. May as well make the transport protocol dumb, because anything that needs SSL will have to implement it anyway.

Re:Will die like bluetooth

[prator]

So here is the real issue. The USB spec only allows 500mA of current. At work we design boards with USB interfaces, and I can tell you that 500mA is not much.

I'm not familiar with Firewire, but Wikipedia says that a Firewire cable can provide 60W of power. That is a hell of a lot more than USB.

-prator

Re:Bill wants you to have it

[cHiphead]

actually its firewire that has the power supplying ability, USB only has a trickle of power. thats why the iPod can be charged by just pluggin it into the firewire port, but not with the usb connector.

Re:bluetooth meet wifi

[man_ls]

You can run absolutely anything to want over WiFi.

WiFi is the lowest two layers of the OSI model: The physical layer (the radios) and the datalink layer (CSMA/CA)

Anything else is software. Network layer, and so on.

Thus -- you can run anything that runs over Ethernet, over WiFi.

Re:Power

[dublin]

Digital Convergence While you *can* sortof achieve these things with WiFi and IP Streaming, the bottom line is that neither WiFi nor IP Addresses are trivial enough for Grandma to connect hreself.

No, actually, it *is* easy enough for Grandma, but the problem is that corporate egos are trying as hard as they can to keep this from really working. The problem, as usual, isn't that we have no standard, but rahter that we have at least one too many.

The bottom line: we (users and product designers) are all caught in the middle of a Mexican Standoff between Zeroconf/Rendezvous and Universal Plug-n-Play.

Zeroconf is a real, open, IETF standard, and also the basis for Apple's Rendezvous.

UPnP is the Intel/Microsoft altrernative, which is a SOAP-based "non-standard standard", and specifies far more than is necessary (and probably wise) for interoperability. ALthough it looks more "complete" on the surface, it is overweight, and that completeness may well turn into an unacceptable brittleness in years to come, where Zeroconf aims more to be a very basic platform upon which to build.

From where I sit, it's easier, cleaner, and considerably simpler to implement Zeroconf. It does all the things that matter, but since it skips the questionable value UPnP puts on forcing everything into SOAP, it's much cleaner, ligher weight, and thus far more suitable for embedded devices.

The problem, of course, is that Intel and MS are NOT going to support Zeroconf, although it works like maginc in the Mac world, hard-to-find Windows clients like Howl are required to use it with Windows. That pretty much quashes any possibility of Grandma's using it, if she's going to be using the PC she bought at CompUSA anywhere in the system.

Ultimately, this will be one of the most important battlegrounds of the next few years: It's hard to overstate how important it is to have this capability if we want to be able to move beyond PC-based devices to real, intelligent network devices that offer far more flexibility at far less cost. Intel and MS want that world to fail, while Apple sees little to lose there. Sadly, open source is mostly AWOL in this battle, although there are some exceptions.

There's only one answer: demand full Zeroconf support in *all* operating systems, and vote with your dollars!