I'm dealing with a problem where sometimes an object I'm looking at using RF signals can appear to be somewhere it's not due to the bouncing of the waves.

Currently what I have in place is a circular buffer that keeps the past 200 timestamps correlated to when it was seen it in that zone.

Once it sees that the object has been seen in two different zones it calculates the average time it was seen in both zones and returns the zone with the greater time.

This works pretty great under ideal conditions but sometimes the zones can be in a tight clusters so it's possible it sees 3 or more zones at once.

I understand this isn't exactly an easy problem to solve and I'm not looking for an answer on exactly what to do. I'm not sure if it's even possible to get something with 100 percent correct results but maybe some advice to improve what I have.

  • 4
    $\begingroup$ welcome to scicomp! you should try to reformulate your question - at the moment, it's unclear what you're asking. are you trying to determine the physical location of an object using something like radar or tomography? what do waves bounce against? what do you mean by a "zone" and "cluster (of zones)"? exactly how do you determine the time-dependent position? do you use some sort of mathematical model for the wave propagation in your chosen domain (if so, what?) $\endgroup$ – GoHokies Aug 29 '17 at 6:04
  • $\begingroup$ Have you considered a support vector machine (SVM)? This sounds like a classification problem (deciding which zone). $\endgroup$ – combo Aug 30 '17 at 1:45
  • $\begingroup$ Is there a "time of flight" filter you could use, based on knowledge of the past states of the objects, which would eliminate the sensitivity to reflections? $\endgroup$ – SteveO Aug 31 '17 at 3:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy