How I tracked FBI aerial surveillance

ArsTechnica – by John Wiseman

On May 5 the Washington Post published Surveillance planes spotted in the sky for days after West Baltimore rioting.

They gave the registered owner of one of the aircraft: NG Research. That was enough to start digging.  

Googling the registration numbers of the planes from the Washington Post article and looking in thepublic FAA records, I found several forum posts over the years where people said they thought the planes were suspicious and were FBI. These forums were usually conspiracy/paranoia/gun rights types of sites, but maybe they were right this time.

“People on the Internet” claimed that FBI planes squawk 4414 or 4415 on their transponders, and use callsigns beginning with “JENNA” or “JENA”. It seemed far-fetched.

For fun I planespot/radarspot/whatever. I have a little software defined radio dongle that I use to pick up aircraft transponder pings. I can pick up aircraft from all over the Los Angeles basin with it. I log up to one ping per second per aircraft in a database. You can see more information about the setup and information I can receive at this page: Tracking Aircraft Over Los Angeles.

Since I had two months worth of transponder pings at this point, including transponder squawk codes and callsigns, I checked to see if there was anything to it.

Of the 15,000 or so aircraft I had tracked, I found 8 that had used 4414/4415, and one with a JENNA callsign—that also had squawked 4414. And they were all registered to generically named companies. I felt like there might be something to this.

I wrote up some of what I’d found in a comment on Hacker News (scooping the AP by 25 days). N404KR was a plane that I saw a lot over Los Angeles. I didn’t have position info for it, but the fact that I saw it in the air for hours at a time was unusual. Most planes are going from point A to B, and are only overhead and in range for a few minutes.

It all seemed very circumstantial, with suspicions based on fringe internet forum posts, and the names, addresses and low internet visibility of a bunch of companies in the FAA records. But then on May 7 the FBI confirmed that the planes that the Washington Post had written about were theirs, verifying the link between them and NG Research. I felt like this put everything on firmer ground.

I looked through the FAA records and found about a dozen suspicious companies based on their names and addresses. According to the public records, one of them even shares a PO Box with the U.S. Department of Justice—that’s pretty sloppy work on the part of whoever’s responsible for generating these front companies, or maybe they’re not even trying to hide it.

Those plus the companies linked to the suspicious aircraft I’d seen over Los Angeles gave me 17 suspected front companies, which had almost 100 aircraft according to the FAA records.

The ACLU had reportedly filed an FOIA request concerning the planes over Baltimore, so I asked them (via Twitter) if they’d be interested in evidence of potential persistent surveillance over Los Angeles, too. They said yes, and I began thinking about how to present the information I had.

Then I discovered that @MinneapoliSam was on the same track!

I created a spreadsheet with all the information I’d found so far.

Something I hadn’t noticed back on May 5 in the Washington Post article was that one of the planes they mentioned, that had been confirmed by the FBI as belonging to them, had squawked 4414!

When Karpathy’s char-rnn recurrent neural network language model code hit Hacker News, it activated the part of the brain that had been thinking about the obvious patterns in the front company names. I only had 17 companies in my data set, but I trained a model to generate new names. Because tracking government surveillance should be fun.

Then the FBI aerial surveillance story began to take off in the mainstream media.

http://arstechnica.com/tech-policy/2015/06/how-i-tracked-fbi-aerial-surveillance/

Start the Conversation

Your email address will not be published. Required fields are marked *


*