Sunday, January 29, 2006

How Listening to Woodpeckers Meant Upgrading Hastings Internet

(Photos below) For the last 4 years, our internet access has depended on a Hughes satellite connection that at best could do about 1,000Kb/s down and maybe 56Kb/s up. Four years ago, compared to a phone connection, this seemed fast. But with the advances in demand for moving data, this had become painfully slow.

Yuan Yao, a graduate student at UCLA,with Professor Charles Taylor and the Center for Embedded Networked Sensors (CENS), began work in late 2004 at Hastings to see if she could use voice recognition software to identify individual acorn woodpeckers in the social groups at Hastings. Walter Koenig and his colleagues have been studying the complex social relationships of these birds for decades, but to see which birds were living with who on any given day, a field assistant had to set up a blind, sit quietly with a telescope and identify each bird using the unique color bands on their legs.

Over the next two years, Yuan diligently made digital recordings of all the woodpeckers in the Plaque Tree group at Hastings. Because all the birds had unique leg bands, this is one of the few places in the world where recordings from each bird could be attributed to known individuals. Later, she recorded individual in another social group that lives nearby. Each individual bird was recorded many times and software was developed that could consistently tell them apart. Luckily, it turns out that the voice recognition persists for at least two years; bird’s unique voice patterns seem stable. Yuan and others built several small listening devices with this voice recognition software. Each has a unique identifier that allows them to be accessed wirelessly on the internet. Working as a small network, they can record the identity and location of each woodpecker based on their wakka calls. However, these devices needed much faster connections to the internet than Hastings had at the time.

So, funds from UCLA were set aside to upgrade the Hastings internet access. It took two years in various attempts to get this done. First, we explored a ground line (T1) with SBC. Although they assured us that it could be done in 10 working days, in about 129 working days, the engineering folks at SBC told the sales folks at SBC that Hastings would have to pay for about 17 miles of upgraded underground phone lines. We asked our current vendor to upgrade, but they could not deal with the legal department at UCLA.

Finally, we found a company named Ground Control, from San Luis Obispo. Ground Controlwas able to work out a contract with UCLAs purchasing department. They tried to get us an account on the Tachyon system, but Tachyon stopped providing new connections. So, we went with another Hughes product, the DW7700 modem using a newer technology called VSAT (Very Small Aperture Technology). Basically, there is a much reduced latency in the signal to and from the satellite, and we can transmit at 256Kb/sec. and receive at nearly 2,000Kb/sec. We are told this is good enough for video conferencing, and we will try.

But what this is all really is good for is Yuans bird ID devices. Soon, she will be deploying groups of these listening devices that will start tracking the acorn woodpeckers and making the data on location of individuals available on line anywhere.

On Friday January 27th, Ground Control installed the new 1m. dish. But for weeks before this, Jaime del Valle and I had been scrambling to get Hastings ready for the big upgrade.

First, we had to move all the wireless internet access, used by the visiting classes and scientists, to a completely separate system that cannot see our other computers. Providing a wireless access point is a security risk unless it is isolated by a firewall from the other parts of the local area network. To do this, we had to move the current Hastings connection to the Berkeley Seismological Lab and the NSF-funded Earthscope project to a secure, hard-wired part of our network. This meant we trenched a new wire from the offices to the Hastings Cabin. Heck, we only hit one water line! Easy to fix.

Second, to connect the rest of the Hastings wireless access points, we had to install power to the large weather station on the hill above the offices. This meant trenching power from the water tanks to the hilltop. And sure enough, we hit a wire that tells the well when to fill the tanks. We fixed this, but in an effort to close the trench with our little Ford tractor, I managed to drive it into a newly discovered spring on the hillside above the office. With great skill, Jaime was able to back the tractor out the next day, using the front bucket to push and lift itself out of the mud. Unfortunately, I did not get a photo. Too bad; it was an impressive mess. All that remains now at the spring is a large pond of liquid mud with no trace of the tractor tracks.

In the next few weeks, we will move the previous Cabin wireless access point to the hill above the offices, transmitting to the School House where the wireless signal is once again re-transmitted to that neighborhood. When working, we will have wireless coverage for all housing and the classroom in the headquarters area, and for any wildland study that might need such access in the central part of Hastings.

Over the weekend of January 28-29th, Andrew Stromberg visited from the UC Davis Computer Engineering School. Andrew installed a new firewall, a router and reconfigured the wireless access points. He restored the local internet connection on the many reserve computers in just two days. Many thanks to Andrew; setting up the system was not simple.






New dish going in; note older, smaller Hughes dish on roof in background.

















Aiming new dish
























Jaime and new trench from office to Hastings Cabin.





























Trench to Hastings Cabin and former wireless access point.