Wednesday, June 27, 2007

Recipe for a Wireless ISP

Sorting through hardware that I have accumulated over the last many years, and it looks like I've got everything I would need to start a Community Wireless Network right here in my apartment complex. I've got a WDS network set up between my WRT54G and my Airport Express, and I've got a $50 D-Link DWL-730AP that I can use in client mode to connect to the network. The WRT54G is the most capable device of the three, working as a WDS base station, where the DWL-730AP behaves just like you would expect if it was called a Wireless Modem.

What does this mean? WDS means range extension; I can set up independent wireless base stations, some of which are connected to the Internet, where some are only free-standing power saps with antennas pointed in the right direction. Clients can choose to associate with any access point on the network. In theory at least, that should bring Internet access to all clients, even if only some base stations are directly connected to the Internet.

In reality I have only one wireless client to test with, and all of these devices are crammed into my apartment. The DWL-730AP has an interface that allows to select a specific access point by MAC address, but there's no way to test that this has any effect. I won't know if WDS has the desired effects without stationing the unwired base station somewhere far away, where the wired AP cannot be reached by clients. The WRT54G can be hooked up to a yagi, or Cantenna, which has the effect of producing a small cone of boosted signal, so it could still reach the distant access point even if nearby clients could not.

The interface on the DWL-730AP is not ideal for this sort of testing. If I had more time and unused hardware to spend on this (another laptop), I would hook up a FreeBSD or Linux machine and change my association from one access point to the next, to test network conditions on either side of the half-wired/half-unwired network. If the tests go smoothly, I can start whoring out my cable modem to my neighbors immediately. Score!

1 comment:

Kingdon said...

Fine, so we know what hardware investment is required minimally to maintain an object-transaction safe exchange model. It could be profitable to pay $50/month for my own Time Warner internet service and just sell these devices pre-configured as wireless device cloud seeds. How about the infrastructure necessary to support partitioning or fire-walling across a decentralized system like that, to support authentication domains, and to enable digital social networking? All of this comes after the hardware.