Noise vs. Signal

We’ve been mucking with some algorithms for determining the baseline noise of the CCD in the phone — it is sensitive to heat. However, it is still “detecting” pixels significantly above the threshold in a very large number of instances — e.g., 60988 events out of 62250 samples — which seems way too high. A re-analysis of all these potential events is needed to determine a better filter. Also, a real cosmic ray event would leave streaks so a line detection filter down the pipeline will need to be included.

If you have ideas, please speak up. If you have a phone I can let you load an Alpha version and collect data.


Problematic Camera Images

We’ve run into a snag processing camera images. The theory behind the cosmic ray detector is that the CCD in the phone camera is sensitive to radiation other than visible light. If I block the camera lens and then begin to take pictures I should get totally black images, except in the case where a cosmic ray event has impacted the camera sensor and caused “sparkling” or set of pixels to become illuminated.

On the G1 this works well in that I can regularly get total black images. However, on the Nexus One I can never get a totally black image. Does this mean the the CCD is very “noisy” or does it mean that it is very sensitive? Am I being bombarded by cosmic ray events, background radiation (Yes!), or the CCD is just flaky?

For you math wiz’s out there, you know, those of you with perfect 800’s on your SAT’s: How do you write a digital filter to find real events while blocking out benign electrical noise? Since a Cosmic Ray event is fairly low level, how do you make the distinction? What is its magnitude?

And then for the app developer: How do you write the filter — or put it in a processing pipeline — so that it is fast enough to not be a bottleneck to obtaining samples — i.e., it has to be faster than the time it takes to collect a sample.

If you have ideas please feel free to attend or jump in to the discussion. Thanks.

Our new project is DECO

Our new project is DECO, the Distributed Electronic Cosmic-ray Observatory which was originally defined here:


There are almost 2 billion cell phones in the world1. Many, if not most, now come equipped with camera, GPS, embedded computer, and Wi-fi, making each one a powerful distributed sensing element. Unfortunately, most of this massive sensing and computing power currently goes to waste. But it doesn’t have to! We propose to harness the combined power of the world’s cell-phones to create the world’s largest cosmic ray telescope.

Cosmic rays are high energy particles that collide with the upper atmosphere and cause a “shower” of particles (electrons, muons, etc…) to rain down to the Earth’s surface. It is possible to detect these secondary particles using the CMOS or CCD sensor in the cell phone’s camera. The highest energy cosmic rays are 108 times more energetic than the highest energy particles created in particle accelerators such as the LHC. Interestingly, little is known about these highest energy cosmic rays, mostly because they are exceedingly rare events (we now detect only a handful each year). However, by creating a large, distributed cosmic ray telescope, we may be able to record enough of these events to finally understand the physics behind these highest energy particles.

Modeled after the distributed computing efforts of SETI@Home and Folding@Home, the Distributed Observatory will become the first distributed, data-taking, physics experiment. More importantly, the experiment will help answer one of the most pressing unsolved mysteries in modern physics: from where do the highest energy cosmic rays originate?

Some of the code for this already exists and runs on old G1’s. Our goal is to evolve the existing code to “sense” better on newer hardware and to provide the core collection infrastructure.