OK, then I can say with reasonable certainty that the signals you saw were not real, or at least were not on the frequencies you think you saw them at, rather they were images of signals on other frequencies.
The below is not a slam on low end RTL-SDRs, it is just technical fact. The RTL-SDR allows performance at a price point that was simply unimaginable 15 years ago. Today, for a few dollars (well under $100 USD) you can get instantaneous bandwidth and tuning range that cost thousands of dollars less than 2 decades ago. But they do have some very real technical limitations, and their performance, by todays standards, are modest, at best.
Low cost RTL-SDRs like that, due to their low bit depth and corresponding modest-to-low dynamic range, are very susceptible to imaging and overload issues. New users are especially plagued by this, as they tend to keep the gain turned up and not recognize when they are seeing an image, vs a "real" signal on a "real" frequency. This can get extremely bad when combined with an external, and improperly used, LNA. Yes, it can be frustrating.
I would bet that is what is happening in your instance. You are either seeing an image frequency, or the SDR is just flat out not doing what you think it is doing. Either way the end results are the same, that signal is not on the frequency you think it is on.
A key factor to keep in mind is that you saw those signals at well above 2.5 GHz, to over 4 GHz. But the specs on your Nooelec V5 RTL-SDR only goes up to 1.75 GHz. Jsut because the software allows you to try and tune the receiver to those higher frequencies, to make the numbers indicate that is what frequency you are on, does not mean you are actually looking at those frequencies.
Your description of "the ones at 4Ghz move fastest and the one lower move slower" is classic for images. The ones at higher frequencies appear to move faster because they are actually at multiples of the lower frequency signals. And when you multiply an error that error gets bigger. So an apparent signal at twice the frequency it is really at will move twice as fast, and one 4 times as high will move 4 times as fast. In my last example there, a change of 100 kHz moves the signal 4 times as far, or 400 kHz. Images are not harmonics, but they may have harmonic relationships.
T!