Back to the Future: New Wi-Fi Bridges Use 1999 Standardby Glenn Fleishman
It's a proverb that there's nothing new under the sun, and the recent surge in inexpensive Wi-Fi wireless gateways that build clouds of wireless access that bridge to each other without a wire in sight proves the principle. The standard that allows this new integrated feature in cheap access points appears way back in the revised IEEE 802.11 specification from 1999, which helped start this current revolution.
Wireless bridging comes in two forms: the one we're accustomed to talking about generically, in which a Wi-Fi signal is bridged via an internal router in an access point or gateway to a wired network. The wired network can be as plain as a connection to an external or internal dial-up, DSL, or cable modem. Although this is often called wireless bridging, it's really wireless-to-wired.
The second form is the "new" flavor, however: wireless-to-wireless bridging using the Wireless Distribution System (WDS). WDS allows packets to pass from one wireless access point to another, just as if the access points were ports on a wired Ethernet switch. WDS bypasses the kind of magical kludgery that Linksys offered with its WET11 bridge or with the WAP11 in pairs or as a bridge/AP combination.
The original WAP11, and similar bridge/access points made by other companies for the consumer market, could provide point-to-point or even point-to-multipoint bridging, but had to be locked into a bridge-only mode. They couldn't function as an access point and a bridge at the same time, a limitation that newer devices with WDS support have avoided. (It's still an option if you just want to bridge without offering client access.)
The WET11 and even the newer WET54G work singly, by contract, and they can connect to any access point. They achieve their ability to bridge a number of wired devices—from 30 to 50—by masquerading the wired devices' MAC (Media Access Control) addresses as their own. This can cause some network confusion, as MAC addresses are meant to be unique, typically.
WDS doesn't use MAC masquerading nor does it require supporting devices to lock themselves into a bridge-only mode, although some makers are still choosing that option. WDS-enabled access points can service local clients while also bridging traffic to other, similarly enabled access points in a hierarchical tree. One unit can act as a master gateway or the "point" in a point-to-multipoint network; the master is the unit that is typically configured to be connected to the Internet or some sort of robust backbone. Other WDS-equipped gateways all point to each other in a star configuration.
WDS relies on a previously underused element of the Frame Control Field for IEEE 802.11b. With the distribution system (DS) bits both set to 1, the client can provide not just its source address (SA) and the destination address of a frame (DA), but also the addresses of the two intermediate wireless gateways (transmitting and receiving or TA and RA) that are bridging the connection. The WDS-enabled access points broadcast MAC messages across all connected nodes allowing all clients to see all adapters on all connected wireless and wired networks.
Let's take a typical example. Jim is using a laptop in Building A, while Jane administers a cable modem in Building B. Building A and B are linked by Access Point 1 and 2 via WDS. Access Point 2 is also a relay, which connects with Access Point 3, which is connected via Ethernet to the cable modem that Jane runs.
Jim tries to access the Internet. His wireless adapter transmits a frame with the destination address of Jane's cable modem. Access Point 1 knows that Jane's cable modem is on the other side of its bridge based on its MAC table. It rewrites the frame to include itself as the TA and Access Point 2 as the RA.
When Access Point 2, the relay, receives the frame, it knows that the destination is really on Access Point 3, so it rewrites the frame again, turning itself into the TA and Access Point 3 into the RA. The frame arrives at that final access point and hits the cable modem.
In a star configuration, each access point has a list of other WDS-equipped access points that it's bridging with.
The excitement about this technology comes from three factors: a WDS gateway can act as an AP and a bridge in a single box; WDS is standard, although interoperability isn't being tested by manufacturers yet; and WDS is cheap, with Buffalo offering its WLA-G54 for as little as $100 street price. In the past, expensive enterprise devices either involved standalone bridges or pricey combined units that worked only with other proprietary gear.
Apple also offers WDS in its AirPort Extreme Base Station ($200 or $250, depending on features); Linksys is using WDS in its revised bridging AP, the WAP54G, but the unit has to be locked into bridge mode even with WDS turned on. (Both the Buffalo and Apple units can be set to ignore local clients and just act as bridges.)
You might want to use WDS in a variety of common scenarios that would otherwise require expensive or unnecessary Ethernet cabling. If you needed two access points to serve your house, for instance, and didn't want to run Ethernet between them, WDS is an able substitute.
It's also a cheaper and simpler way to provide blanket coverage. Attach some sectorized antennas on opposite ends of an open courtyard or park and use WDS to avoid having to put backhaul or backbone between the two units, even while both of them can offer the optimum strength for client connections.
Remember, however, that the stronger the antenna's signal, the more directional it becomes. If you wanted to use WDS to span a mile between your home and that of a friend, you could certainly do so, but the WDS devices wouldn't be able to also serve up access to clients in their vicinity. WDS is more about building a cloud than a long-haul system, and stronger omnidirectional antennas could be part of that.
Still, a pair of Buffalo WLA-G54's talking WDS might be easier to manage than a pair of WAP54Gs, and more easily swapped in and out if equipment problems emerge.
All of the current home gateways are single network interface devices, which means that the WDS backhaul subtracts from your pool of bandwidth. Even worse, WDS bridges and master units must work over the same Wi-Fi channel, which increases interference and reduces throughput. With enough local devices, you're sacrificing a lot of bandwidth for routine traffic.
The 802.11g factor does help out, though. If you have the usual DSL speeds on your Internet backhaul, then cutting 20-odd Mbps of net throughput in an ideal 802.11g environment down to a third or fourth or a fifth still won't hold up traffic upstream or downstream. It may be less ideal in offices, homes, or dorms in which local data needs exceed the equivalent of 10Base-T. (In that case, you could use back-to-back bridged connections in which you use a short wired backbone between two access points and set the back-to-back set to a different channel.)
Still, Wi-Fi traffic is usually in bursts, so even a solid stream of data that's being relayed from one gateway to another via WDS and starting and ending at a wireless client should be able to achieve 5 to 10 Mbps.
A looming question with WDS is interoperability. WDS isn't part of the Wi-Fi certification program, and it doesn't appear as though access points made by different companies use the identical approach for configuration, nor is any maker guaranteeing their gear works with anyone else's.
I was able to turn an Apple AirPort Extreme Base Station into a master WDS node (see Figure 1), and it automatically discovered a Buffalo WLA-G54 on the same network that had its WDS mode enabled (Figure 2). I entered the AirPort address into the Buffalo list of WDS access points, and selected the Buffalo on the AirPort's list. After reboots, they both spoke to each other with no hiccups.
Mac users might find this information especially appealing, as they can use the WLA-G54 as a cheap alternative to the more expensive AirPort Extreme Base Station. The WLA-G54 is an access point with no gateway features like DHCP service, making it the perfect WDS extender.
WDS can pull some of the complexity, expense, and copper out of any new installation, and it's worth thinking about whether when you're building a home, office, or neighborhood network whether a WDS access point, bridge, or relay could fit the pieces together with less stress.
The next generation of WDS devices clearly need two radios, and some enterprise equipment has already taken that tack. Vivato recently announced a two-radio access point meant to fill in the gaps or edges of service that it can offer with its phased-array antenna design; they use WDS for the switch-to-access point communication. Tropos Networks has a mesh system for backhaul that relies on WDS, while an optional second radio offers client wireless access.
It might take a leap of faith for the consumer companies to add two radios, but with the cost of a gateway with WDS in the $100 range, a $175 device with two cards and separate antennas for back haul could be part of the next big expansion of wireless networks.
Glenn Fleishman is a freelance technology journalist contributing regularly to The New York Times, The Seattle Times, Macworld magazine, and InfoWorld. He maintains a wireless weblog at wifinetnews.com.
Return to the Wireless DevCenter