Log of Connection Tests

3 November 2010 - Regnart

Looked at spectrum facing Moffett from Regnart.

Regnart_Spectrum_View_of_Moffett.png

11 November 2009 - LGH to California St Garage rooftop, Mountain View

18 dBi panel to identical panel, 15 miles, 18 dBm out at one end, 26 dBm at the other. Unable to associate.

8 August 2009 - Layer 42, Santa Clara to SJCEOC

Layer 42 Photos

Ubiquity Bullet 2's using an 18 dBi panel on a 65 ft tower in Santa Clara and a 22 dBi grill parabolic on the roof of the SJCEOC. 20 dBm output power at each end.

The test went well and certainly as well as we dared hope. We were seeing no packet loss, and speeds were varying between 4-12 Mbps. Signal levels were between -45 to -61 dBm against a noise floor or -94 dBm. We had trouble executing our jitter tests later on, but when we measure it, it averaged 8.6 ms.

The antenna tower raised us to about 65 ft AGL. What surprised us in the end was that lowering the antenna closer to roof level did not lower our received signal strength. Instead, the level started to increase and then decrease and so on, fluctuating a significant 8-15 dBm. The only explanation that made sense to me was that we were finding peaks and nodes in the signal attenuation by the power mains we were forced to propagate through.

In any case, the signal was still easily strong enough to assure us a reliable link at least at some speed, and this means we have a green light for a permanent install.

To proceed, we will need to find or purchase an adjustable antenna tower and mount which will cost in the neighborhood of $100. We can then adjust it to peak signal height if we are lucky enough that our obstruction (whatever it actually is) does not vary capriciously.

Once we have that ready to go, the ethernet cord will be sent down a roof conduit into the colocation facility and to our servers where we can route it through to the internet and PoE.

Right now, the only thing this project is waiting on is a suitable tower. If you have one to donate, please speak up!

Special thanks again to Dan Fitzpatrick of Layer 42 for donating his time, antenna tower, generator, parking lot and roof access to this cause. We could not ask for more supportive partners in this project.

Credit also to every one else who participated today: Craig, Sean H, Rafael, Ian and Jim Oberhofer.

6 June 2009 - Los Altos Hills to Shoreline Mound

7.5 km, ~1 MB/s using Ubiquity Bullet 2's and a 17 dBi panel and 23 dBi parabolic grill.

Congratulations and thank you to the field team for a successful test today: Craig Anderson, Rafael Skodlar, Andrew Brown, Sean Head and Shu Zhang.

We got off to a rough start by discovering that the Shoreline Amphitheater had failed to check with us before scheduling a concert today. Consequently, after slogging through event traffic for much of Shoreline Avenue, we found the otherwise enormous empty parking lot had become a controlled event area with parking storm troopers running around everywhere. Fortunately, Craig was able to schmooze our way in using his civilian police volunteer badge and the parking attendants tried hard to accommodate us without really having much understanding of who we were.

We eventually got permission to park at the base of the hill where we set up to test our link. Once we saw pings, we packed one station into my car and Sean and I headed over to Los Altos Hills. We luckily had a clear shot through the trees to the mound where the signal mirrors were able to do their magic for pin pointing us once the clouds allowed us some sun.

With only eye ball alignment of the antennas, we were able to get connection pretty quickly. The lowest signal was -74 dBm but the guys at the mound were seeing better than that. We used a 22 dBi grill in Los Altos and the mound team tried a flat panel and then a dish.

Using the Ubiquity speed test tool, we observed pretty consistent data rates around 1 MB/s. Repeated 1KByte packets were able to ping with regularity, so the connection seemed pretty solid.

We took a stab at hooking up VoIP, but were not able to get the client configured in the time left. Perhaps we'll have to work the kinks in that out at a meeting sometime. It would be nice to have VoIP to add qualitative testing to our future field tests, as well as helping us develop more competency with VoIP.

Eventually, the police paid the mound team a visit and starting asking questions, followed by a friendly suggestion to move along. From what I understand, they were basically just in the area for the event and our group stood out as questionable. The antennas made them a little uncomfortable as well as the view we had of Moffitt Field (Were they worried about Larry Paige's jet?) But we still had time to wrap things up, minus the VoIP test. We'll try that next time.

The Bullet's worked as advertised and results continue to be promising.

17 January 2009 - Los Altos Hills to California St Garage rooftop, Mountain View

Successful connection using Nanostation 5's on their built in antennas by Bill Railey and Andy Brown.

• Antenna Gain: 14 dBi
• Polarization: Vertical
• Measured signal: about -76 dBm
• Measured data rate: tested inconsistently between 0.5 and 2 Mbps
• Distance: 4.5km (2.81 mi)

Use of 28 dBi parabolic again failed to connect. The suggests something wrong we aren't yet aware of. Even the high loss of the RG174/U patch coax shouldn't account for that much loss.

11 January 2009 - Los Altos Hills to El Camino Hospital rooftop, Mountain View

Using 28 dBi dishes on 5.8 GHz Nanostation 5's, we eyeballed from both directions using a signalling mirror. One radio was fried, believed a result of using a cross over cable on the PoE portion. Using a back up radio, the best signal strength observed was only -90 dBm. This was far below expectations and the experiment is still under analysis. This was our first attempt using these radios, antennas and cables.

23 September 2008 - Cambrian Area, San Jose

A connection was established from a member's house in the Cambrian Area of San Jose to the Los Gatos hilltop router with an 18dBi panel antenna. This link provides Internet access to the EWN over a shared DSL line. The EWN routing was adjusted to make use of this connection.

21 September 2008 - San Jose EOC and Los Gatos Hilltop Installation

13 members of SVWUX and San Jose RACES came together to install a 10.4 mile Wi-Fi link from the San Jose EOC to a hilltop above Los Gatos. Though many hickups it finally worked. This was the culmination of a lot of work by a lot of SVWUX members.

At San Jose EOC, a 22 dBi solid dish was installed pointing towards the Los Gatos location about 17 km away. A Linksys WRT54GL was flashed with Kamikaze 7.09 and successfully linked to the mountain top. Ping times were impressive, averaging somewhere around 8ms. No lost packets observed.

Ian, Sean and Warren, risking life and limb, replaced the transmitter at the Los Gatos hill top location and installed an 18 dBi panel antenna pointed towards the San Jose EOC. The second antenna pointed towards Cupertino was installed, but had to be taken back down due to running out of sealant tape for the connectors.

Photos below courtesy of Bill Moffitt. Photos not available of those working indoors on the router or on the hillside in Los Gatos.

• The nice line of sight from SJ EOC roof top towards Los Gatos Hills
  

• How many engineers does it take to screw in an antenna?
  

• Andy, Heather and David supervise Craig while Kiavash ponders the future.
  

• -
  

• -
  

• -
  

• High precision alignment process... :
  

• -
  

29 April 2006 - San Jose OES

Ian Kluft and Chris Verges used a 21 dB parabolic dish antenna connected to a Linksys WRT54G (OpenWRT whiterussian rc5). We were able to see the ESSID coming from the SVWUX 802.11b/g access point, but were not able to associate to the access point.

Summary of discussions:

• Standard 802.11b/g defines RTS/CTS timers that restrict the operational range to less than 5 miles. 802.11 equipment that works over longer distances redefines the RTS/CTS timers.
  • Can we redefine the timers on the SVWUX repeater (a Linksys WRT54G running OpenWRT)?
  • If not, can we add a wireless bridge to the SVWUX repeater site that handles the proper RTS/CTS for long-range wireless shots?
• Supporting multiple protocols (802.11a, 802.11b/g, 802.11n, 802.16) would prove beneficial as a long-term goal of the group in bringing together all manner of users.
• San Jose R.A.C.E.S. needs a way to define preferential traffic in the event of an emergency situation. One possible solution discussed includes the following components:
  • Multiple ESSIDs, each ESSID associated to a separate VLAN
  • Quality of Service (QoS) defined for each ESSID/VLAN.
    • Highest QoS value (7) reserved for system management
    • Next three highest QoS values reserved for emergency operations (6 = medical, 5 = fire/police, 4 = all other emergency traffic).
    • All other QoS values can be defined as different service levels for non-emergency traffic (3 = voice, 2 = interactive sessions, etc.)

27 May 2006 - West Valley College, Saratoga

I found a good spot with a view of the repeater just over the top of some trees:

http://maps.google.com/maps?ll=37.261408,-122.013119&spn=0.006302,0.014355&t=k&om=1

in the south west corner of the parking lot just to the north of the baseball diamond.

Using a 14.5 dB yagi (Hyperlink HG2415Y) and a Linksys WRT54GLv1 .1 running OpenWRT WhiteRussian RC5, I was able to associate, obtain a DHCP lease, and maintain a stable connection.

• Signal/Noise: -64/-83 dBm
• Ping SVWUX repeater (100 packets): 0% packet loss, round-trip min/avg/max = 2.8/4.7/21.8 ms
• Ping Steinway (100 packets): 5% packet loss, round-trip min/avg/max = 4.8/40.6/290.4 ms

I should note that I had not set the distance setting (wl0_distance=32168) on this particular AP. In the future I will collect statistics with and without the distance setting applied. I will also try various transmit power levels as Brian suggested.

-- TimUtschig