Linksys WET11: Bridge 30 Devices To Any Wi-Fi Network

eggboard writes "The Linksys WET11 lets you bridge a wired network with up to 30 devices to any wireless access point that uses Wi-Fi. The utility is enormous: you could build a pseudo-mesh network by pairing cheap Wi-Fi APs with this cheap ($129) Wi-Fi bridge. Before this, the only generic Wi-Fi bridge was proprietary: you'd buy a bridge from Alvarion that paired with one of their hubs, and spend several hundred each. Even the dual-WAP11 bridge approach of last year was wonky and required extra gear (although it can handle more devices than 30 since it's a protocol bridge, not a MAC bridge). I review the WET at O'Reilly's wireless developer's site."

Comments

Wireless ISP Uses

[gabeman-o]

I hang out at the WISP message board on broadbandreports.com ( http://www.dslreports.com/forum/dslalt ) and it seems that many WISPs are using these to connect customers to their wireless network. The WET11's antenna is detachable, so you can use an external one. It can also be configured to use POE (power over ethernet) by changing two jumpers.

There have been some cool mods like:
http://www.dslreports.com/forum/remark,4220 109~roo t=dslalt~mode=flat

http://www.dslreports.com/forum/remark,4123612~r oo t=dslalt~mode=flat~start=0

Re:ok

[rnd()]

and the device is entry-level

You must be looking for Slashdot Pro. You can sign up for just $99/mo. You've reached Slashdot (lite) by mistake.

wonky is official jargon

[jukal]

wonky /wong'kee/ adj.

[from Australian slang] Yet another approximate synonym for broken. Specifically connotes a malfunction that produces behavior seen as crazy, humorous, or amusingly perverse. "That was the day the printer's font logic went wonky and everybody's listings came out in Tengwar." Also in `wonked out'. See funky, demented, bozotic.

Weather-proof?

[Quixote]

This does not appear to be ruggedized for outdoor installation, so I wonder how useful it'll be for bridging the 'last mile'.

I have been looking for a WiFi bridge to help a friend get cablemodem access: the cable company will not lay the cable to his house (cost > $1K), and it is about 300' from the road. One possibility I am considering is to use a pair of WiFi APs to bridge the gap. But they need to be rugged, to withstand temperatures from -30F to +100F, and rain/snow. None of the cheapo units comes in a weatherproof enclosure. Anyone know of any? I've looked around. Of course, the cost can't be more than $200 or so (each); the .com days are gone. :-(

I bought one

[DirkDaring]

It's a great little unit. Been playing around with it for about a month. It can use an external antenna (RP-SMA connector) also. It's also 70mW, which helps out on the transmissions.

There are a few current threads about the WET-11 for Wireless ISPs here:

http://www.dslreports.com/forum/remark,4123612~r oo t=dslalt~mode=flat

http://www.dslreports.com/forum/remark,3915738~r oo t=dslalt~mode=flat

Dirk

Odd behavior from the WET11

[JackAsh]

I was one of the early customers for the Linksys WET11. My hope at the time I bought it was to be able to hook it up to my Xbox game console and use it to play Xbox games online while running Gamespy or or XBConnect on my primary PC, and therefore not have to run any unsightly ethernet from my entertainment center into my computer room (I rent an apartment, and the landlord would appreciate me returning it eventually without too many extra holes).

Obviously, nothing is ever that simple. I discovered that the WET11 performs some wonky MAC addressing translations when forwarding packets. Seeing as Xbox System Link games depend entirely on ethernet addressing schemes, the WET11 proved to be useless for this - despite Linksys advertising it as a solution for Xbox gaming.

Correct me if I'm wrong - something labelled as a "Network Bridge" should pass packets from one side of the bridge to the other unaltered, and simply keep a table of what addresses are on what side so as to pass packets when necessary between two broadcast segments. WET11 converts all MAC addresses on its "wired" network into it's own address. The reverse is different - it'll reassemble incoming wireless packets based on what I guess was their original IP source, and place the MAC address it replaced with its own back in the packet. Go figure why they go through all the trouble. Of course this behavior is undocumented, so this took several hours of packet sniffing (so blame me, I don't deal with Layer 2 issues every day).

Now, Xbox Live is expected to work at the IP level, but that is'nt out until December. Existing System Link game like Halo only work on a local broadcast segment, using ethernet for addressing while sending out some horribly mangled non-RFC compliant ethernet packets that look like UDP but aren't.

The ethernet mangling caused problems with Gamespy and XBConnect, but I was able to get in touch with the developer for XBConnect and over a nice weekend hacking session we were able to cobble together support for the WET11 in his program - essentially it now has the option to look for the MAC address of the WET11, and retranslate that to that of an Xbox. The funny bit is on the return path from a remote Xbox, it needs to again translate the address of the local Xbox back to a WET11 so your average Access Point knows who to retransmit your Xbox packets to. Every Xbox game needs to go through four translations: two on the WET11 and two on XBC.

http://www.xbconnect.com

http://www.apoxx.org/community/viewforum.php?f=1 3 - this is the forum for XBC. Look for anything in there labelled WET11.

-Jack Ash

Re:Odd behavior from the WET11

[hey!]

Correct me if I'm wrong - something labelled as a "Network Bridge" should pass packets from one side of the bridge to the other unaltered, and simply keep a table of what addresses are on what side so as to pass packets when necessary between two broadcast segments.

Close. A network bridge does not separate broadcast segments, but passes all broadcast packets through to all available interfaces.

WET11 converts all MAC addresses on its "wired" network into it's own address. The reverse is different - it'll reassemble incoming wireless packets based on what I guess was their original IP source, and place the MAC address it replaced with its own back in the packet. Go figure why they go through all the trouble.

Hmm. I don't know much about 802.11 but Part 2 is easy; it is a result of part 1. If the WAP didn't put the real destination MAC address back in, the real destination will ignore the packet unless it is promiscuous mode.

The interesting question is why change the wired MAC addresses in the first place. I've loooked at the 802.11 header format, and it would appear the behavior you describe is standard. There is a "ToDS" flag bit, that if set means the packet is intended for an AP for forwarding. The standard then says that in this case the 802.11 destination address field should have the AP MAC address rather than the recipient.

I think this may have something to do with ad hoc vs. infrastructure mode. In ad hoc mode, each wireless station talks to each other wireless station. In infrastructure mode the packets always pass through the WAP. Each of these modes have their advantages. For infrastructure mode to work, every packet transmitted by a wireless node needs to have the WAP's link address. Cards should not try to handle packets, even if they can see it and are the ultimate destination, because in this mode they are supposed to wait for the WAP to retransmit it with their MAC address in the destination field.

Of course this issue is moot when we are talking about connecting wireless nodes to wired nodes, but the fact remains the WAP is going to ignore any wireless packets that aren't broadcasts or targetted to it's MAC.

The behavior which you desire might be accomplished by setting up an ad hoc network, and putting a box on it with bridging software which talks to both the wireless and wired network. I wonder if a Linux box could be configured this way, using the Linux kernal's bridging features. It might not work since the wireless network is somewhat different from the wired, but it would be worth looking into if there were a need to do some protocol other than IP (note that WAPS can't be protocol agnostic like bridges can be because they have to do a network address to MAC address translation).

Re:Odd behavior from the WET11

[JackAsh]

Re: "Broadcast segments" - You are absolutely correct sir. My mistake. I unfortuantely have a tendency to use the word "broadcast" when in reality I mean collision. Broadcasts should indeed be passed to all interfaces, and the WET11 handles this appropriately. What I originally meant was that I thought bridges were supposed to separate collision domains - packets would only be passed through the bridge if the destination address was on the other side.

Re: behaviors: Indeed part 2 is a direct consequence of part 1. If you send packets out with a modified source MAC, the reply's destination will have to be that same modified MAC. However, I still fail to see why the WET11 needs to change MACs in the first place. D-Link has a device similar to the WET11 in function (sorry, don't know the name) that reportedly passes packets on transparently. From what I can tell, on my BEFW11S4 access point, the Wireless connector acts simply as a switched port. It keeps a table of attached MACs in order to minimize traffic effects.

Personally, I suspect Linksys is using the MAC translation in order to allow more devices behind WET11s to connect to a single Access Point. I mean, imagine your accesspoint has an address table of 5 MACs (just a sample number for my point). A single WET11 is designed to have up to 30 devices behind it, so passing each one's MAC address would overload the Access Point and make it not work. However, if a single MAC is passed for the whole network the Access Point would be able to handle up to 150 devices with 5 WET11s! NOTE: This is pure conjecture on my part.

I don't know if there's a difference between Infrastructure and Ad-Hoc modes. All I've worked with is Infrastructure, so I've no information on Ad-Hoc. However, Linksys has some documentation indicating that two WET11s, linked to each other in ad-hoc mode would allow XBoxes to communicate with each other, which might indicate no MAC changes in ad-hoc...

The way I see it, the WET11 should act as a pure bridge, and simply keep tabs on which MACs are on which side of the bridge and forward accordingly, and of course transparently. This would also allow the WET11 to forward protocols other than IP - one of the objectives you suggest...

I guess what I'm saying is if you're gonna build and market a Network Bridge, make it just a BRIDGE, dammit. :)

-Jack Ash

Re:ok

[liquidsin]

It would be a much more interesting read if someone setup a linux (or any other OS) box with a wifi card in it and a wired nic that feeds a hub/switch and NAT'd a bridge.

Go for it. Or wait for someone else to do it. In the mean time, I'll be thankful for the review because it's putting ideas, just like yours, into peoples heads as to what they can do with this new toy.

What am I missing?

[FreeLinux]

Ok, the price is pretty cheap. No argument from me there. But, I still can't see the "technical greatness" of this device. Symbol, Cisco and probably Lucent have always had bridges capable of both bridging and AP mode at the same time.

Also, they can all be configured to talk with each other so, I don't understand the proprietary reference. Sure, Symbol et al don't speak Cisco's enhanced WEP but, the Cisco can be configured to use standard WEP so it will speak with the others.

So, besides cost, whats so special about this device? Something else that I am missing is the power output of the Linksys bridge. What is it? Historically, the cheaper Linksys Wi-Fi products have had a lower power output that the others. Is this also true here or does this latest Linksys bridge put out the full 100 milliwatts, as the others do?

Re:What am I missing?

[eggboard]

We're talking two issues here: cheap and simple. I don't think the average consumer would spend $1,300 to get the wireless bridge + AP that the Cisco gear would cost, and I don't think the average consumer could figure out how to configure it!

The WET11 is pretty straightforward, just as hard as hooking up a client to an AP, and it's dirt cheap.

WAP11

[NetJunkie]

If you are using WAP11 APs already it is actually cheaper to buy another WAP11 and put it in "client access mode". That's what I did. The WET11s were going for $130 *IF* you could find one, but the WAP11s were going for $105 shipped from buy.com.

I just bought another WAP11 and put it behind my home theater, set up bridging mode, and plugged in a hub. Works great with my Rio Receiver, XBox, PS2, etc.... So, it was cheaper and can be used as a full AP later if needed.

Re:I'm confused.

[eggboard]

Actually, an AP bridges from wired to wireless. The WET11 can take a pod of up to 30 wired machines (or rather, machines attached to a single Ethernet segment) and bridge all of their traffic to ANOTHER access point. The WET11 is a wired-to-wireless bridge that hooks into a wireless-to-wired bridge.

Put 30 machines in a room on one side of the San Francisco Bay, plug in a WET11, point an antenna to an AP on the other side, and presto: you're running a large (very latent) Ethernet network.

Re:I'm confused.

[Gumber]

What's missing from this explanation is consideration of the WAP11 which can also bridge a wired segment wirelessly to another wired segment via another WAP11.

The problem is that neither WAP11 can serve wireless clients at the same time. With one of these in place of one of the WAP11s, the WAP11 can both link individual wireless client machines AND bridge to the remote wired segment served by the WET11.