Range of communication between 2 XOs is limited only to 20 meters

[arjs]

Testing at pilot site at Khairat.

Conditions - A remote village, low density of trees, no other wireless network present there. Testing was done line-of-sight. Hardware version B4-6, Build 601.

The line of sight range of communication between 2 XOs is about 20 meters. This was done in an open field where both the XOs were outside the school server range. We did this by pinging from one XO to the other (after manually determining the ip address of each XO). This test was repeated many times to verify the results.

I have attached copy of iwconfig at the end of this email - taken when the 2 XOs have just lost contact (a few meters farther than about 20m). I observed iwconfig gives some information about transmitted power, received power, noise etc. in dBm hence am attaching it in case that might help pinpoint the reason for the limited range.

==================================== iwconfig log ====================================

msh0 IEEE 802.11b/g Nickname:"Mesh"

Mode:Repeater Frequency:2.412 GHz Bit Rate:1 Mb/s Tx-Power=18 dBm Retry limit:8 RTS thr=2347 B Fragment thr=2346 B Encryption key:off Power Management:off Link Quality=3/100 Signal level=-92 dBm Noise level=-93 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:54774 Tx excessive retries:763 Invalid misc:2661 Missed beacon:0

eth0 IEEE 802.11b/g ESSID:"olpc-mesh" Nickname:"MRVL-USB8388"

Mode:Ad-Hoc Frequency:2.412 GHz Cell: 63:2D:6D:65:73:68 Bit Rate:1 Mb/s Tx-Power=18 dBm Retry limit:8 RTS thr=2347 B Fragment thr=2346 B Encryption key:off Power Management:off Link Quality=3/100 Signal level=-92 dBm Noise level=-93 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:54779 Tx excessive retries:763 Invalid misc:2661 Missed beacon:0

====================================

[jg]

Are there other machines you can test with? It sounds like one of them has loose antennae, or shorted antennae.

[arjs]

===Posting communication with Michail here so that all discussion is in one place===

Michail

These are the serial numbers that I have at hand... SHF72500350 SHF72500356 SHF72500354 SHF7250033C SHF72500341 SHF72500316 SHF7240000E SHF7250034E SHF72500328 SHF72500327 SHF7250034D SHF72500347 SHF72500338 SHF7250032E SHF7250034F SHF72500306 SHF7250034A SHF72500352 SHF72500359

Carla - Hide quoted text -

On 10/8/07, Michail Bletsas <mbletsas@…> wrote:

Arjun,

From your description, it doesn't seem that electromagnetic noise could be an issue and you probably ended up with a batch of XOs with very bad antennas. I would appreciate if you could send me the serial numbers from the sticker in the battery slot.

20 meters is just too short.

M.

"Arjun Sarwal" <arjun@…> Sent by: arjunsarwal@…

10/07/2007 02:01 PM

To

"Michail Bletsas" <mbletsas@…>

cc

amit.gognaa@…, carla@…, "Manusheel Gupta" <manu@…>, "John Watlington" <wad@…>, "Walter Bender" < walter@…>

Subject

Re: Wireless test results - pilot project at Mumbai

Michail,

We did repeat the experiment many times and on different days. We were holding the XOs in our hands and were keeping them above our waist level and slightly near our faces.

We were in an open field. There are a medium amount of trees around but none that were even close to obstruct line of sight between two XOs.

School server was switched off. We just did a simple ping from one XO to other. When the ping stopped coming we did ctrl-c and started ping again just to make sure.

Any way to measure the noise level (associating with an Access Point will yield approximate measurements - mesh mode measurements don't produce very useful information)? Sounds like you operate in a very noisy environment.

Could you please let me know what kinds of noise affect such a range ? (EMI, presence other wireless networks ?)

The place where we carried out testing was at Pilot site. It is a village in the middle of nowhere where even the mobile network signal strength is low. The mobile band in India is 900Mhz. There are no wireless networks/access points till tens of kilometers away (Id say atleast ~20 kms)

Anyway we have brought few XOs back with us and would repeat tests near the hotel tomorrow.

The hardware version is B4 6 and the build is 601. What tests can we do to help zero down whether it is a hardware or a software problem ? We could install an earlier OS image and repeat tests, which one would you recommend ?

Thank you, Arjun

[jg]

We need to examine these for problems, such as Dane Roger's teardown report ASAP, and get samples to both Cambridge and Quanta ASAP for analysis.

Walter sent mail on this topic this evening.

[Quozl]

I've tested two B4 XOs to 1.6km repeatedly, but they were the same two, and there were no other XOs on. When I tried a mesh test three or four months ago with six XOs (3xB1 3xB2), if they all pinged at once the range was about 30m at 10% packet loss. The range increased as the mesh was more silent.

I agree, checking the units for problems would be good.

Some other ideas ... interference from military radar, extreme humidity, thick grass or vegetation on the ground, solar noise, obstruction in fresnel zone of path.

[arjs]

Some other ideas ... interference from military radar, extreme humidity, thick grass or vegetation on the ground, solar noise, obstruction in fresnel zone of path.

Humidity - yes that area has high humidity 60-80% RH average. The area also had thick vegetation but none in the way of line of sight while we were testing.

The best way to find out interference from military radar would be to actually try and find out the presence of a nearby military base ?

[carrano]

We received two XOs here at 1cc: SHF 725002FF SHF 72500306

Note that the first one is not in the list provided in #comment:2 The second XO was not bootable and was reflashed to 625.

Other than the above notes, they could ping each other at 175 meters in a parking lot near 1cc. They could possibly go farther, but we had no space.

Please confirm that during your tests, both XOs were in the same channel. If they were one or two channels apart the symptoms would be very similar to what is described above.

[arjs]

We simply powered up each XO and let it go to the default channel. At the same time we ensured that the school server was powered OFF during this so that the XOs don't go on the school server network.

Amit, can you please repeat a test at Khairat whenever you get a chance while ensuring that both XOs go on the same channel and let us know of changes in range, if any. Thanks. (please reassign to arnold when test done)

[carrano]

Replying to arjs:

We simply powered up each XO and let it go to the default channel. At the same time we ensured that the school server was powered OFF during this so that the XOs don't go on the school server network. Amit, can you please repeat a test at Khairat whenever you get a chance while ensuring that both XOs go on the same channel and let us know of changes in range, if any. Thanks. (please reassign to arnold when test done)

It would be extremely helpful if you capture iwconfig and ifconfig outputs for both XOs after and before the connectivity is lost. Thank you!