Birdie injection plan

Message boards : Nebula : Birdie injection plan
Message board moderation

To post messages, you must log in.

Profile David Anderson
Volunteer moderator
Project administrator
Project developer

Send message
Joined: 13 Feb 99
Posts: 95
Credit: 390,229
RAC: 0
Message 1857024 - Posted: 22 Mar 2017, 20:14:25 UTC

In radio astronomy, a "birdie" is an artificial signal added to the input data, for the purpose of verifying that the data processing system works correctly.

For example, in SETI@home we have periodically added a fixed-frequency sine wave to the analog radio telescope signal, then verified (by looking at spike signals computed by the client) that we're detecting it, and that the frequency is correct. This checks several parts of the system: signal digitization at the telescope, splitting, and most of the client logic.

This type of birdie, however, isn't useful for checking RFI removal, scoring, and the rest of Nebula. For that we need a much more sophisticated type of birdie: one that mimics an actual ET signal coming from a particular point in the sky. Such a signal would be detected only while the telescope is pointed close to that location.

It would be difficult to add such birdies into the raw time-domain data. Instead, we're planning to inject them at the start of the Nebula pipeline by adding signal records (spikes, Gaussians, etc.). Eric Korpela and I have begun working on this. For starters we're adding birdies that are always-on narrow-band signals. The basic idea is:

  • Create ~1000 birdies with randomly chosen frequency, sky position, power, and intrinsic bandwidth. Half the birdies will be "barycentric", meaning that the sender is compensating for its acceleration in the direction of Earth.
  • Scan the history of telescope pointings; whenever it's pointing close to a given birdie, randomly generate spikes of appropriate power, FFT length, position, etc. For non-barycentric birdies, add a frequency offset for the sender's Doppler drift (assume Earth-like parameters).
  • Add these spikes and Gaussians to the input to the Nebula pipeline.
  • After running the pipeline, check the resulting multiplets and pixel scores, verify that the birdies are there (e.g. they weren't flagged as RFI) and that they appear close to the top of the score-ranked list.

Doing this efficiently presented challenges. I won't bore you with the details, but suffice it to say that R-trees came into play once again. I hope to have things working in a couple of weeks.

ID: 1857024 · Report as offensive
Profile Chris SCrowdfunding Project Donor
Volunteer tester

Send message
Joined: 19 Nov 00
Posts: 39794
Credit: 31,622,380
RAC: 25,899
United Kingdom
Message 1858417 - Posted: 30 Mar 2017, 10:10:21 UTC

Sounds a good plan, look forward to hearing more later.
ID: 1858417 · Report as offensive

Message boards : Nebula : Birdie injection plan

©2017 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.