In my work on radio reflections, I have been working with an ultra-low power transmitter and attempting to coax out reflections against the atmosphere or other local objects. I’ve had some limited success and through a lot of work I believe I’ve figured out some of the reasons I don’t have the kind of success I want.
The most paramount of issues is the amount of power I am using. If you ask me I think SNR is just as important to a radar as any other communication system and when using such a low amount of power, the order of 10mW, I just can’t push very far.
A secondary issue I’m also dealing with is learning about how phase is affected by reflections and my equipment. I’ve been seeing phase measurements that don’t meet my expectations and I can’t find a good explanation about. I want to talk about some of that during this session.
The mag ** 0.001 output is the standard reflectivity where mag = signal.correlate(rx_data, sent_signal, mode='valid') but of course in this case mag is 3D because it contains each (correlation) scan as a row.
You can see a large spike of energy at the beginning. Both the receive and transmit antennas are directional and pointed upward but they are near each other. I’m sure this is the energy bleeding horizontally or come from objects within one sample distance and it produces not only a large spike but the point of synchronization.
This spike of energy lasts for a many samples where it finally dies off very quickly. You can see some ripple extending around 0.5km to about 4km and I really feel like this is due to reflections.
The phase is colored blue for -PI and red for +PI. You can see the phase holds very steady at the beginning and then it degrades out into what appears to be a near constant phase per scan. I have no explanation as to why.
The np.abs(mag[1:, :] - mag[:-1, :]) ** 0.001 is my take on trying to coax out more range or sensitivity. This takes each subsequent row and subtracts it from the current then takes the absolute value of that and finally undoes any logarithmic scaling to try to flatten out the value space.
You can see a very light band around 14km which is very interesting. Is there something there?
Zoomed in it is easy to see that the band around 14km does exist. It is more centered on 13km and then now there is a band from 4km to 9km. I don’t think these are symptoms from the autocorrelation of the chirp which means it didn’t produce these bands as it slide across signal window.
Conclusion
This test ended in a failure. The data stayed consistent even though the weather conditions went from light showers to overcast. There was some change though and it might not be a complete failure.