Friday, December 26, 2008

3.17 mi field test, Ping, Distance, 2nd STA

Today Kerry and I set off for a 3.17mi field test--a link longer than any we expect to make on this trip to Jalalabad.

First we determined that the AP->STA ping problem we had last time was a firewall issue on the router. For the moment, we just disabled the firewall by removing the S45firewall file from the /etc/rc.d folder on the STA and AP (though I suspect we only needed to remove it on the STA - will try later). For those who aren't familiar, the purpose of the files in this folder is to call startup scripts in a particular order (each file links to a file in init.d). By removing S45firewall, we kept the firewall from booting, leaving the link wide open.

With that issue solved, we moved to the field and I found this beautiful spot at the fells end of the 3.17 mi link:
pointing at the wrong parking lot... I guess I only know my way around in the dark. It wasn't long before I realized my mistake and shuffled over to the southwest side of the fire tower for this setup, pointing the right way:We then proceeded to make links with both the small and large reflectors, seeing a -70 dB signal strength with the large reflector and -82 dB with the small. Extrapolating from the signal strengths at 2.17 mi, the ideal expected signals would be -67 dB and -79dB, respectively, but our line of sight was much narrower on this link and there was more moisture in the air than on the last test.

Initial throughput tests for the large reflector yielded .58 Mbps. After setting the
option distance 5200
setting in /etc/config/wireless (same section as the channel), our link speed improved to 2.62 Mbps (x5!!). Switching to the small reflectors, we recorded a throughput of .79 Mbps, suggesting lots of packet loss--plenty worse than the ideal case numbers I posted here--maybe this is a result of the WiFi Signal a couple channels over? (see trace below)

Below is a trace from the Wi-Spy on the big reflector during the throughput tests. You can distinctly see the (local) STA and (remote) AP.
We attempted to get a reading on the Wi-Spy by adding a second STA on the AP side and sending data to it as mentioned in the last post. Note, to add a STA you need to add a route to it on the AP. Kerry added the following route to the boot section of /etc/init.d/network. New STA is .133.1, AP is .127.1:
route add -net 192.168.133.0 netmask 255.255.255.0 gw 192.168.127.3
though it can probably be added to /etc/init.d/custom-user-startup and have the same result. At any rate, our three router scheme didn't work as planned, as we didn't see anything on the other end of the link. More testing is required in a controlled setting. A complete set of the config files on the STA and AP for this test can be found here

A final interesting tidbit: the Motorola Talkabout radios worked passably over this distance when mounted in the reflectors. This is 1.5-3x (I have two models) their intended range.

Improved backpack selection made heaving 40 lbs of gear up the hill much more comfortable, BTW:

Send this thing halfway across the globe already!

2 comments:

xsdg said...

"New STA is .133.1, AP is .127.1:

route add -net 192.168.133.0 netmask 255.255.255.0 gw 192.168.127.3"

What device is .127.3? Does it have routes for both .127.0/24 and .133.0/24? Alternately, does the machine you were adding this network to have a default route that it may have preferred to the network-specific route?

The Wrench said...

Off the top of my head I'm not sure. Gotta look at the configs. All devices are routars, so they all live on the subnet boundaries.

I think I was a bit unclear about the actual results. We were able to connect the third station, but our hope was that we could scream some data between the two nearby stations (one of which would be mounted in the reflector) and then see the RF signal at the remote reflector with the Wi-Spy. For some reason this didn't work, and I wasn't at the 2 device end, so I can't say for sure if I'm missing something with the way the radio works (as in it's smart enough to dial down the power when talking to a nearby device) or if the data transfer just wasn't working as expected. We did get the two local stations to see each other.

Post a Comment