Yo,
I did solve your issue by creating a perfect predictor for random numbers generated by state-of-the-art RNGs.
Now, I'd happily share it with you, but I'm kinda busy using my new tool to hack some banks and writing a Nobel prize paper.
Seriously...
Your nodes have random positions and they are moving.
You could run the simulation twice (one to record the node position, the second to use it), but it's useless, nodes are moving.
Think out of the box... place two extra nodes in two fixed positions (at the opposite corners of your area) and ping between those two.
Have fun,
T.