Today, there are many easier ways to do wireless networking. In fact, and this is the bad news, there are too many ways to make a wireless LAN work. First, we have 802.11b, known to those who love it as Wi-Fi. But, we're already seeing far faster 802.11a network interface cards and hubs powered by the Atheros chipset and by the third quarter of 2002 we should see 802.11g and ultrawideband (UWB). What's a network manager to do?

The answer is to look carefully at each protocol and see where it might fit into your overall enterprise plan. However, as fast as these might go--and 802.11a, 802.11g, and UWB can all theoretically get to 54Mbps--none are as fast as Fast Ethernet's 100Mbps. All but 802.11b are a big step up from normal Ethernet, but server-to-server communications and clients working on massive data files will still need Fast, or even Gigabit, Ethernet.

Let's start with 802.11b. It's popular and the devices are dirt cheap. It works reasonably well on low throughput LANs such as a branch office, where shuffling Word documents is the biggest bandwidth demand. In theory, it can go as fast as 11Mbps, but it usually checks in at two to 4Mbps. And that's shared bandwidth, so during busy periods or with, say, 50 or more users, that 4Mbps isn't going to go very far. Finally, as the whole world knows, 802.11b is notoriously insecure. For my money, I'd use it only in small branch offices where nothing of critical importance is likely to be sent between systems.

Next, there's 802.11a. It can reach up to 54Mbps, but in practice you're looking at about 20 to 22Mbps. While it also uses shared bandwidth, that's still five times more bandwidth than 802.11b can offer. Unfortunately, it has pretty much the same security problems as 802.11b. At first, it may appear more secure, but that's just because the cracker tools to listen in on it are still being developed. Even so, I'd deploy 802.11a in most offices.

Why? All the wireless companies are working hard to make Wired Equivalency Privacy (WEP) protocol a real security tool instead of a security joke. Gartner believes WEP will finally be secure by the middle of 2002. That's when I expect 802.11a to be widely deployed. Another reason I favor 802.11a is that the giants like IBM, Intel, and Cisco are also readying product for shipment. Technically, it's great and it has market support. What's not to like?

Unfortunately, 802.11a doesn't work with existing 802.11b LANs. Compatibility is where 802.11g comes in. That doesn't mean as much as you might hope it does. 802.11g hubs will be able to keep 802.11b NICs running, but you won't see any speed increase in them. To get the most from 802.11g, similar to 802.11a, you'll need to replace your entire wireless infrastructure.

Some people working on 802.11g tell me that it, too, will be in the 54Mbps range. Other 802.11g engineers tell me that it won't be quite as fast as 802.11a. Given that, like 802.11b, 802.11g works in the already crowded 2.4GHz range, I'm inclined to think 802.11g devices will be a bit slower than their 802.11a counterparts.

802.11g is a case of a technology that has a lot of money and emotion invested in it. As a result, its vendors are still trying to push it forward, even though 802.11a has a clear advantage. I can only think they're hoping 802.11a won't take off very quickly so they'll have a chance to try to outmarket products based on 802.11g, or get OEMs to bundle their chipsets into "universal" hubs that support 802.11a, b, and g. That may be the topic for another column.

802.11g is also a bitterly fought-over compromise between the wireless visions of Texas Instruments and Intersil. While the core encoding issue has been settled, there are many other peripheral issues to be squared away before 802.11g is an approved standard. But the bottom line is that you should consider 802.11g only if you one to keep both low-speed b and high-speed g NICs working with a single hub.

Vendors are on their way to preparing 802.11g devices for shipment by the second quarter of 2002. Honestly, though, can you see these early models being compatible with another vendor's hub or NIC? I can't.

If Intersil and Texas Instruments can bury the hatchet--and not in each others' backs--I still can't see 802.11g being fully approved until the third quarter of 2002. That means real 802.11g products won't be shipping until the end of next year. By that time, 802.11a will probably be rendering the whole question moot as to whether 802.11g is worth considering.

That may also be the fate for UWB. Until the FCC approves UWB for networking, it will remain only a potential standard. Unlike 802.11g, however, UWB may be successful if any of its vendors can hang on until FCC approval comes through. That's because UWB is inherently more secure than any of the 802.11 technologies. Because UWB requires exact key matches between hubs and NICs for data transmission, it's an order of magnitude more secure than anything WEP will ever be able to deliver.

I'm told by friends at both the FCC and the UWB vendors that a decision on UWB will be arriving shortly. I see a 70/30 chance that UWB will be approved no later than the first quarter of 2002, or even this month. If that happens, UWB will be the technology for anyone who needs security above and beyond other wireless issues like 802.11 compatibility. You'll have to wait longer for product. At the earliest, I don't see mass shipments until the third quarter of 2002, but for security- minded enterprises, it will be worth the wait.

If you have to plug a technology into your 2002 budget right now, go with 802.11a. I wouldn't plan on spending another dime on 802.11b. As for 802.11g, unless you can get NICs and hubs with dual 802.11a and 802.11g chipsets for one heck of a good price, I wouldn't bother with it. And, if keeping the competition out of your hair and network is job one to you, pray really hard that the FCC will bless UWB, and plan for end of 2002 deployments.

The one good thing about all these choices? They're all a lot easier than freezing over a microwave dish.