Peter
I understood from your original question you had a reason to decide which nodes should be able to reach the server at any given time. That's up to you to define and implement it.
In some settings that could just be based on distance (range). Other problems might schedule this, or based on messages received. It all depends on what you are modeling.
Peter