Would you like your message here? You can sponsor 802.11b Networking News for a week at a time and reach thousands of daily readers

The above would be a paid, sponsored link if it were an ad. Contact us for more information.

Today:

News

Bluetooth's Virtues

FatPort launches FatPoint: in a move long anticipated by their hardware focus (and some conversations I've had with them going back to November), Vancouver, B.C.-based WISP FatPort released three all-in-one packages under the name FatPoint. If you're in a Canada, you can get FatPoint Complete, and use FatPort's DSL leverage to get Internet connectivity and the box that ties you into their network. If you're anywhere else, you can just buy the box for US$525 and split revenue 50/50. (A subscription fee is folded into the one-year contract.) If you're the enterprising type, you can buy the OEM version, which is solid hardware, and customize to your own needs.

About sponsorship: FatPort will be sponsoring this site starting Monday for two weeks at the regular rates. I don't cover anyone just because they sponsor the site, nor does the sponorship confer any special status. As long-time readers know, or those who survey the archives can find out, I speak my mind.

Small Net Builder Product Review of the D-Link AirPlus Enhanced Cardbus Adapter (DWL650plus): a fairly comprehensive survey of the D-Link that uses the Texas Instruments chipset supporting 22 Mbps PBCC encoding as well as 11 Mbps backwards compatible 802.11b. D-Link won't be upgrading this unit via firmware to support 802.11g, however, as TI plans a different chipset for 802.11g support.

Bluetooth's Virtues

My dear colleague, David Weinberger, author of the mind-changing set of essays Small Pieces Loosely Joined has gotten embroiled in a debate over Bluetooth. I joined the fray, and he asked for some longer comments, which he abstracted into an article he's writing. I've edited my response and made it a little clearer, and abstracted his questions into the below Socratic text on why Bluetooth has a place in this Wi-Fi world.

Why is Bluetooth so slow (1 Mbps)?

Speed is a function of bandwidth, and usually there's a wire limit to bandwidth, like category 5 8-wire copper can handle 100 Mbps over X feet, or a physical (as in physics) limit to bandwidth, like "Ethernet nodes can't be more than X feet apart because you need within X milliseconds for any two receivers on an Ethernet connection to recognize that a collision has taken place and stop transmitting."

So you can't throw more bits on a wire than it will handle nor can you put things too far away for them to do their talking (although you can play games, like putting in switches that are, essentially, little Ethernet devices themselves).

In the case of Bluetooth, the designers had about 75 MHz of bandwidth to play with, or the full legal U.S. bandwidth for use in the 2.4 GHz band for unlicensed users with licensed devices.

Bluetooth designers were thinking about data sync and cable replacements, both of which are essentially serial technologies. USB runs at 1.5 Mbps and 12 Mbps (and USB 2.0 runs much, much faster: 480 Mbps), but it requires physical access and close physical proximity. Also, I’ve discovered that USB cables ain’t cheap. If you wanted longer cables, especially, you’re talking about a lot of money. USB hubs don’t work as well as I’d like and they require an outlay and often separate power supplies. USB may be simple, but it’s not as simple as it could be.

So the Bluetooth design goal was more or less: how do we make something that could eliminate the low-speed USB necessity and allow it to be more generic, requiring no separate drivers for each new device? The ancillary question was: how do we eliminate cables, hubs, and physical connections? So you have a single Bluetooth interface that can work with modems, keyboards, handhelds, printers, and so on, without additional cabling, network protocols, or what have you?

They chose a pretty simple approach. 1 MHz bands using frequency hopping which changes 1,600 times per second. This choice allows a lot of different Bluetooth at once (because of the lack of chance of random collision), and a reasonable speed for the design purpose.

Bluetooth further limited distance by specifying maximum power output to comply both with FCC regulations (they're well below those limits) and the kinds of devices they wanted to work with: USB stubs (USB-to-Bluetooth), cell phones, handhelds, etc. All places where battery life is a scarce commodity. Shorter range also reduces the chance of interference. If you had 100 Bluetooth devices across a 2,000 sq. ft. office space, you might have no interference. There’s no way to build that kind of ad hoc network with Wi-Fi: it has to be planned and analyzed.

A lot of Wi-Fi-is-god backers, which I used to be an ignorant member of, foresaw that Wi-Fi would replace many of Bluetooth's functions as Wi-Fi's curve for cost was plunging much faster than Bluetooth's. However, the power issue is a biggee. Almost all Wi-Fi devices, even those designed for small form factors, are designed to run 150 to 300 feet. This is part of Wi-Fi design spec.

As we know from physics, when you combine speed and distance, you need power and bandwidth. You can tweak different aspects to control or expand those parameters. Wi-Fi uses 22 MHz channels, non-frequency hopping (direct sequence, a different technique) and to get the Wi-Fi stamp have to meet certification tests that mean that any Wi-Fi device can reach the same minimum distance and conform to the same standards.

You can't, at the moment, make a Wi-Fi Lite because there's no way to differentiate. You’d need an entirely different spec. Hmm.

Further, the low bandwidth isn't a hindrance for the appropriate activities. When Bluetooth is harmonized with Wi-Fi (802.15.2 and a new Bluetooth revision), you'll be able to have possibly a single radio that talks Bluetooth and Wi-Fi using different MAC chips. They'll serve different purposes, but be bundled in the same wrapper. There are already companies selling Bluetooth/Wi-Fi products, and we’ll see more of those.

Wi-Fi, as I've said many times, has a lot of network costs: you have to build a network connection, use TCP/IP or a similar protocol that runs over Ethernet, and transmit a lot of noisy information, just to exchange a few bytes. The authentication protocols are computationally expensive, as well as tedious for smaller devices in which entering 128-bit WEP keys (15 hexadecimal two-digit numbers) is ridiculous.

Bluetooth was designed to allow security in a computationally efficient manner for the kind of data being transmitted, down to the simple passphrase handshake to establish a connection between two devices.

A device needs to be certified before it is Bluetooth compliant?

Certification is as much an issue for Wi-Fi as it is for BT, and this argument is incorrect.

The IEEE 802.15.1 task group has passed a very full subset of Bluetooth in coordination with the Bluetooth SIG, the industry consortium that owns the patents and defines the technology. You'll be able, with minimal IEEE appropriate license fees, to implement Personal Area Network (PAN) devices that are Bluetooth compatible using 802.15.1 and advertise them as such. (I’m confirming this via the 802.15 public relations task group.)

If you want a device that says Bluetooth on it, which is a trademark and a certification, you will have to pay other fees, just like the members of the Wireless Ethernet Compatibility Alliance (WECA), the $25,000 or more a year membership fee trade association that controls the Wi-Fi certification process and trademark of interoperability.

You will be able to make digital cameras that conform to 802.15.1 if you pay the license fees (IEEE requires them to be reasonable and uniform), but you won't be able to call it Bluetooth. Today, you can make 802.11b equipment and not call it Wi-Fi, too, but you find very few manufacturers who don't want the marketing advantage of hopping on the certification and mark.

The weakest argument against Bluetooth is that I shouldn't need both Bluetooth and 802.11b/a. Isn't that like saying that I shouldn't need both USB and a special cable to hook up my monitor?

This picture from the 802.15 working group tells it best, showing the relationship of WPANs (Wireless Personal Area Networks), WLAN (wireless LANs), WWANs (Wireless Wide Area Networks), and WMANs (Wireless Metropolitan Area Networks). Note that as you increase distance, you also increase speed to allow a bigger pool of bandwidth from which to draw.

On the more practical side, Bluetooth is ad hoc from the ground up. Two people meet, don't know each, and can exchange data with minimal fuss. (Critics of the current generation say that's not so, but I've been doing essentially that with various Bluetooth devices in my office for WEEKS.) It'll get easier than it is today when it's in the Mac and Windows OS like Wi-Fi.

Yes, there are specialized reasons to need different protocols. We don't all view Web pages over FTP; we don't send email via carrier pigeon; we don't take pictures with our monitor. Maybe we will (shades of AT&T), but there's no good reason to.

The folks who are anti-BT like the fact that Wi-Fi does so much, but there are degrees of trust that are extremely difficult to embed in small, low-power, simple devices when you're working with Wi-Fi. I run into the problem of stuffing the wrong problem in the wrong sack all the time when moderating mailing lists: well-run mailing lists answer some questions so effectively that people get mad when the moderator (moi) won’t allow them to ask all questions they want, even those far off topic.

The most idealistic folks I hear from say, sure, but you can just change Wi-Fi: make a low-power alternative, change the spec, add another kind of protocol that runs over Wi-Fi, etc.

Yeah, yeah, and it's taken almost two years to fix the broken WEP (wireless equivalent privacy) encryption layer in Wi-Fi, and we're still looking at maybe March 2003 for ratification and probably six months before the firmware updates for OLDER devices roll out.

So anyone who says, just change the spec, hasn't read the committee meetings from the IEEE, which I have. They need to go and read these public documents from several meetings and understand the number of players and the degree of collegial and non-collegial accommodation that happens to make even the tiniest of change.

Wi-Fi works because of this great, long process, and it can't turn. Works like a fish, steers like a whale, to paraphrase Doug Adams.