eWEEK Labs recently went into the field—literally—to perform tests of a wide-area wireless networking product. Along the way, we got a crash course in the unexpected twists and turns of real-world deployment, as well as a stark reminder that infrastructure purchases cannot be based solely on the merits of the technology.
We partnered with the Department of Intercollegiate Athletics at the University of California, Berkeley, to test LightPointe Communications FlightExpress 100—a low-cost, point-to-point FSO (Free Space Optics) wireless solution developed to provide 100M bps of full-duplex traffic over short distances (a maximum of about 200 meters).
Cals athletic department has an unusual assortment of short-term networking needs: A hectic event schedule hosted at multiple facilities spread across a large urban campus requires fast, flexible and often-temporary network build-out and tear-down.
Using the FlightExpress 100, we provided network connectivity to Levine-Fricke Field, home of Cals womens softball team.
Cal posts live statistics for its home softball games on the calbears.com Web site—by submitting the stats via a laptop configured with Cingular Wireless cellular data technology. However, Cal officials were interested in a faster connection that could provide connectivity for multiple data clients.
With the LightPointe deployment, we were able to move Cals statistics-laden traffic to a new network with plenty of headroom for additional traffic from media or staffers, in case Cal is selected by the NCAA to host any postseason games this year.
While our tests ultimately proved successful, LightPointe went out of business near the close of our testing cycle.
FSO technology requires line of sight between the devices on both ends of the connection—something we found rather difficult to achieve in the Cal scenario.
We initially looked into connecting the softball complex to the Doc Hudson Memorial Field House at Witter Rugby Field, a relatively straightforward install due to the clear line of sight between the facilities. However, Cal officials opted not to go down this road, as the field house has a comparatively slower connection to the Internet.
Instead, we chose to connect to the Strawberry Canyon Swim complex, which proved a tougher job, as a few trees and the outfield wall stood between the locations. We knew we would need to significantly elevate the installation point on the softball side of the connection.
After our initial site survey, we decided to connect one link head to a stadium light stanchion, which would provide more than enough elevation for the system to "see" the other side of the connection. However, LightPointe officials balked at this suggestion. They warned that temperature fluctuations causing the metal stanchion to expand and contract, as well as high-wind conditions, could easily knock the solution out of alignment. Instead, we settled on a small platform atop the softball facilities bathrooms.
Because the installation was temporary for the purposes of this test, we needed to make some concessions when installing the FlightExpress 100. We did not want to perform an invasive installation by drilling holes, so we made do with a combination of unsecured mounting brackets and sandbags for additional weight. Although these workarounds proved up to the task for the short duration of the test, we obviously would not recommend these methods for a permanent installation.
After both sites were mounted and installed, we performed the alignment. Unlike higher-end LightPointe models, the Flight-Express 100 does not include a telescopic sight. Instead, we needed to eyeball the deployment through a viewfinder. Inside this viewfinder were two apertures shaped like bat wings. We needed to visually align the two bat wings and then target the far link head over the small triangle at the intersection on the bat wings.
The two people performing the alignment coordinated via cell phones, as LightPointe devices must be manually tuned on each side of the connection. The link heads are unmanaged bridges, with no IP address or management interface.
According to the LEDs on the link heads, we achieved a received signal strength of 8 (out of 10) on one side of the connection but only a 5 on the other side. According to LightPointe officials, received signal strength corresponds to the resiliency of the transmission but does not affect throughput (unlike what weve come to expect with Wi-Fi transmissions).
After installation and alignment, we spent an afternoon measuring the performance of the new network. An ongoing ping sweep verified that the network suffered no dropped packets during the course of one afternoon.
However, we grew concerned as we noticed a large truck pull in to the parking lot between the softball and swimming facilities. Although the truck in question did not disrupt the signal between locations, we estimated it was a tight traffic squeeze. For a permanent installation, large trucks would need to be denied access to that part of the lot.
Using IPerf, a bandwidth measurement tool from the National Laboratory for Applied Network Researchs Distributed Applications Support team, we also took some quick measurements of throughput. We were able to attain about 94M bps for both uploads and downloads from the remote location, and we achieved more than 180M bps of full-duplex traffic when uploading and downloading simultaneously.
As we completed the install, we got one final scare when we discovered the softball fields bathroom facilities tend to get peppered with foul balls—any of which could easily knock over or at least misalign our link head. In a permanent installation, we would definitely need to address that problem.
After the install, the Cal softball team hosted its final three games of the season—a loss to UCLA and a pair of wins against Washington. During all three games, statistics were provided to www.calbears.com via the LightPointe network.
Jesse Pietarinen, the Cal programmer/analyst who coordinated the project with us, praised the LightPointe deployment, which worked flawlessly during the games. Further, Pietarinen said, no foul balls blew up the deployment.
Unfortunately, the same couldnt be said for LightPointe itself.
Technical Analyst Andrew Gar-cia can be reached at firstname.lastname@example.org.