## RFI discussion

Message boards : Nebula : RFI discussion
Message board moderation

AuthorMessage
David Anderson
Volunteer moderator
Project developer
Project scientist

Joined: 13 Feb 99
Posts: 114
Credit: 488,656
RAC: 51
Message 1835145 - Posted: 10 Dec 2016, 8:10:25 UTC

Eric, Dan, Jeff and I met this morning to discuss RFI. The main topic was drifting RFI, which (I learned) is typically due to consumer electronics with cheap oscillators that vary over time, e.g. because of changing temperature. Our current Zone RFI algorithm detects RFI with very stable frequency, and it misses drifting RFI.

The Siren RFI code (I linked to a paper about it earlier) has an algorithm that seems to work well for drifting RFI. For each signal, it looks at 10 sectors centered at that signal in time/freq space, and looks for sectors with an excess of signals.

We decided to use something like this. The algorithm as it stands is O(N^2), and we thought of various ways to improve it, so we'll re-implement it using R-trees.

The other thing we discussed is time-bounded zone RFI. Our current algorithm looks for frequency bands with an excess of signals over the entire 17-year duration of SETI@home. It wouldn't detect, for example,
an RFI source that's present only for a few days or weeks. Siren also contains code that we might be able to use to detect such RFI.

The bottom line is that there will be a pause of a few weeks while we figure out what we're going to do and implement it. I'll probably be coding over the holidays.
ID: 1835145 ·
kittyman
Volunteer tester

Joined: 9 Jul 00
Posts: 50173
Credit: 960,009,656
RAC: 258,174
Message 1835147 - Posted: 10 Dec 2016, 8:28:23 UTC - in response to Message 1835145.

Eric, Dan, Jeff and I met this morning to discuss RFI. The main topic was drifting RFI, which (I learned) is typically due to consumer electronics with cheap oscillators that vary over time, e.g. because of changing temperature. Our current Zone RFI algorithm detects RFI with very stable frequency, and it misses drifting RFI.

The Siren RFI code (I linked to a paper about it earlier) has an algorithm that seems to work well for drifting RFI. For each signal, it looks at 10 sectors centered at that signal in time/freq space, and looks for sectors with an excess of signals.

We decided to use something like this. The algorithm as it stands is O(N^2), and we thought of various ways to improve it, so we'll re-implement it using R-trees.

The other thing we discussed is time-bounded zone RFI. Our current algorithm looks for frequency bands with an excess of signals over the entire 17-year duration of SETI@home. It wouldn't detect, for example,
an RFI source that's present only for a few days or weeks. Siren also contains code that we might be able to use to detect such RFI.

The bottom line is that there will be a pause of a few weeks while we figure out what we're going to do and implement it. I'll probably be coding over the holidays.

And I don't doubt you shall....................
Not kicking at you Dr. Anderson.
Some of us don't stop for holidays.
The kitties work 24/7/365, and we rely on you to herd the kitties in the right direction.
Some of know that herding kitties ain't easy.
I shall spare you the clip, because I am sure you have already seen it.
if not, PM me.....LOL.

I have abused and cursed you in the past for certain 'offenses'. but I support you nonetheless in my total support of the science.

I am sure as a man of science, that you know who I am and have forgiven me for my doubt at times.

Carry on then, Sir.

Meow.
"The secret o' life is enjoying the passage of time." 1977, James Taylor
"With cats." 2018, kittyman

ID: 1835147 ·

Message boards : Nebula : RFI discussion