I am working with a system, S1, where many remote devices report in to a central server at constant intervals, f. Each device reports asynchronously with the rest of the system.
The (complete) state of S1 can be queried via a request-response API.
Is there an 'optimal' frequency for another system, S2, to query S1 that balances resource consumption and concurrency between S1 and S2?
A naive reading of Nyquist-Shannon leads me to 0.5f. Is there a better alternative?
You are best off sampling at f. Unless you are really concerned about frequency response of your system (if you are, then you also need to be really concerned about phase response too, and asynchrous reporting means that you are not), you don't want to change sampling rates in the system. It is best to use the same sample rate, even if you are resampling.
If you resample f to 0.5 f, then you need to properly implement a new LPF on the data to avoid aliasing, and that doesn't sound like what you want to do.
If your data is all very slow compared to f, then you probably should reduce f if you want to reduce resource usage (i.e. broadcast or battery power).
