Understanding SpectrumAlert

[s3m1s0n1c]

Thanks for your work on this awesome project.

I would like some help understanding how it works..

I have run the training for 24 hrs and now monitoring but it seems every freq it scans is an Anomaly and earlier on I had the same for signal strength..

I have upped the contamination_rate = 0.1 but same thing..

Do I just keep upping it until I find a neutral spot?

Just confirming.. It shows the signal_strength for each freq regardless of a signal? there's no squelch to just detect signals present?

Thanks in advance
Sonic

[Slayingripper]

Hello,

I need to preface this by saying that I'm not an RF engineer, so I've based this on various models I've seen circulating.

The issue is that the Isolation Forest model is using a fixed contamination rate, but in RF environments, the "normal" baseline shifts constantly.

I've made this more dynamic now, and it seems to be working, but I will have to confirm.

I shall push out a fix later for this after I've confirmed this is the actual issue.

[s3m1s0n1c]

No Worries, Yes that makes sense..

Thanks @Slayingripper

Will give it a run when I get some time..

Cool little project to mess with

Thanks
Sonic

[Slayingripper]

No Worries, Yes that makes sense..

Thanks @Slayingripper

Will give it a run when I get some time..

Cool little project to mess with

Thanks Sonic

I've created a new dev branch to fix a few issues and make the whole thing a lot easier to use. It seems using the "lite" version, it works okayish, so I'm investigating why the more intensive version is not as precise.

[s3m1s0n1c]

Ok cool :) Let me know when its ready to test..