Monday, May 05, 2008

Status of my WISP

Ahhh... So,... to report back in to the millions that are waiting with baited breath to find out how my wireless Internet service provider network is working... well... it isn't. At least not consistently. I put it up and take it down as I have the time to test it. This week will begin a new round of testing with the Oregon firmware.

The problem? Well, its been a combination of a few things, let's start with the most complicated first.

Software Complications
A deployment to allow for folks to pay for access is not available at this point because the software isn't ready. Two large hurdles still need to be overcome. First, access control using FreeRADIUS. Second, an open source billing system built on FreeRADIUS with the options I want (daily, three day, weekly, monthly, annual, bi-annual, biennial/EOY billing with a set payment schedule for the service period).

Hardware Complications
The hardware necessary to operate the wireless mesh network isn't 100% decided. I've had a bit of trouble with the Accton hardware originally distributed through "them." The devices seem to reboot unexpectedly if the power is the least bit unstable. For instance, I plugged in my outdoor unit at my in-law's house... ran it for nearly 3 hours and saw a constant reboot cycle. The only thing I can think to do on this is to build a very-mini UPS/power conditioner that I could have on hand for selected access points that have this sort of problem. If its a problem at my in-law's house... it is likely to be a problem elsewhere in the same power grid. Another option is that I select different hardware. Currently, my test units are old wireless units that were sourced from that one company that crapped on their customers last October.

Solutions Considered
I'm considering using Ubiquiti Networks equipment, but haven't yet decided that they would be a better bet at almost twice the cost. Ubiquiti seems like a very large company from their marketing. However, based upon their support and sales departments... my best guess is that they are a company of very few employees (<15). That isn't necessarily a bad thing. I'm a CIO at a company that is a scant few 7 employees. The only catch with a small company is that features are slower to come out. Since the hardware that I'm using needs to be flashed with different software to be mesh capable... the question is what about the hardware gets "lost in translation." Unfortunately for Ubiquiti, the four internal omni-directional antennas are the sacrificial lambs that head for the slaughter in the name of open source firmware to support a BATMAN mesh. That means an external antenna and sends the unit towards $100/node. Boo. If the Ubiquiti equipment had a driver that would work with OpenWRT for its automatic antenna switching... the UbNt equipment would be first on my list regardless of the $49 - $79 shift from the Accton to Ubiquiti. Another flaw with the Ubiquiti equipment is the shortage of flash and RAM. This is something, apparently, that some of the guys down in Argentina from LUGRo-Mesh have figured out... sort-of. Their firmware is fitting on the 4mb-flash/16mb-RAM configurations... but they don't have a central dashboard and they don't have user restrictions (splash page click-through only). D'oh! If only the LUGRo-Mesh group would coordinate with Vacouver and Oregon/Italy! This is something that I'm going to be pushing in the coming weeks is a solid effort to push these three groups together and hopefully combine efforts.

Range Tests
My initial range tests with my stock antennas on my old equipment was actually quite promising. I was counting out 4-5 city lots. Most of the city lots in my area are easily 120ft wide. That means an initial range of about 600 ft. Add to that the fact that I was not exactly line of sight (several cars, mailboxes, shrubs etc) and I was terribly impressed. I think it may have actually hit somewhere in the range of 800-1000 feet if I hadn't goofed up the experiment by accidentally rebooting one of the nodes with a power cycle. Whoops. I'll be testing more this week (hopefully the weather will improve again... I don't want to have to rain-shield my test equipment).

As you can see, its been a busy time. My initial goal of having paying customers by "this summer" is still on track. However, my hope that my first customers would be in June is quickly fading with the passing days of May. Its still a possibility, but is fading.


Post a Comment

<< Home