I'm looking for a cheap and cheerful, but ultimately working utility or library for tracking a cubesat and predicting link windows. I've been looking at using Orekit. I'm a student, so a lot of its capability is just over my knowledge horizon, but its authors' assertion that it has been used and benchmarked by CNES looks like a good bet.

I've been asked to look at PREDICT (http://www.qsl.net/kd2bd/predict.html), which seems to have a lot of use by amateurs and a little use by NASA and other organisations. It uses the same SGP4 propagation model as Orekit, but has no apparent facility for updating reference data. Its last release was in 2006.

Can anyone offer their experience of PREDICT and tell me whether this is going to be up to scratch for a satellite we definitely want to communicate with on each pass?

Many thanks for the attention.

  • $\begingroup$ It really depends on what your requirements are -- how accurately must you point your antenna, for instance? Although in all likelihood, the answer to your question about PREDICT is "probably". $\endgroup$
    – Chris
    Commented Jul 17, 2017 at 12:55
  • $\begingroup$ Thanks for the reply. Using a quick calculator, the full half-power beamwidth will be about 2.5 degrees. What do you think? $\endgroup$ Commented Jul 17, 2017 at 15:18
  • 1
    $\begingroup$ That would be worth adding to your question. Even better would be if you derive a worst-case position error from that number, using the expected altitude of your spacecraft. $\endgroup$
    – Chris
    Commented Jul 17, 2017 at 15:43
  • $\begingroup$ Is this project a student project, or are you building and launching a CubeSat? The various commercial tools (Systems Toolkit, FreeFlyer, etc) usually give free licenses to student projects, and those tools have a lot of functionality and are well validated. (and you don't have to build it from scratch, like with Orekit) $\endgroup$ Commented Sep 21, 2017 at 15:59
  • $\begingroup$ Just saw your question on AI. (Don't be discouraged--it can be quite difficult to get started on new stacks. We like game-AI related questions, and the question already had a net upvote, which probably means at least two actual upvotes.) Hope to see you back on AI soon. -DZ $\endgroup$
    – DukeZhou
    Commented Dec 21, 2017 at 19:21

2 Answers 2


I used PREDICT operationally for the first few Planet Labs tech demo missions (Dove 1 / 2 / 3). It's fine. The quality of your ephemeris data is going to be the limiting factor, not the software you use to propagate it.


PREDICT uses legacy SGP4 code. SGP4 code is usually referred to a 1km/day rule of error increase. (However that actually strictly depends on the orbit type and satellite.)

The second thing to note is SGP4 was updated by Vallado in 2006 with some error fixes and PREDICT does not see to have implemented those fixes.

Therefore, using an updated SGP4 propagator like STK or Vallado's code would be better in my opinion. But to be fair, as far as I remember, none of the fixes of Vallado was about usual cubesat orbits' ranges.

If you are looking for a much better method than SGP4 but also if you do not have another source of tracking except TLEs, orbit determination is your answer.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.