Regarding Antennas and Relays

[Kelderek]

I have a vessel in orbit of Jool that has a 100G relay antenna and no other communication parts other than a probe core.  When I am controlling this vessel I noticed something peculiar: the ship seems to prefer a direct connection to the KSC even if it is a lower signal strength than using other relays that I have in my system.  I used the custom setting to disable the extra ground stations, so when the KSC is on the opposite side of Kerbin from where Jool is, then it is forced to use the relay network I have, but when there is a direct LOS to the KSC then it always uses that path.  The direct path is a lower signal strength than what my relay network provides so I am wondering why it wouldn't always choose the path that provides the strongest signal.  I did a simple test with a Communotron 88-88 and it behaved the same way as the relay one did from Jool orbit, it also preferred a direct connection to KSC even though it was a lower signal strength.

It seems like this would defeat the effort of building a deep space relay network if a vessel doesn't always take the highest signal strength path - it would also affect how much science you could get.  Am I missing something here?

[FullMetalMachinist]

1 hour ago, Kelderek said:

Am I missing something here?

Very interesting. No, I don't think you're missing anything. It seems like it would make sense to always choose the strongest path, but it certainly isn't in your case. 

@RoverDude, any thoughts? 

[bewing]

Weeeeellll, there are still some little logic problems left in the CommNet path selection. The problem is loops and dead-ends. Imagine you have 6 relays around Kerbin, with 100% signal strength to each other, and 75% signal to Kerbin. So your ship at Jool connects to one of these relays in the ring. And that relay bounces the message to some other relay that it has the strongest connection with. Which bounces it to another .... forever. So you can't just send it where the strongest signal is without more logic than that -- since the message eventually has to end up getting to Kerbin.

[Kelderek]

Well my best guess is that this may be intentional just for the sake of simplification.  Calculating signal paths and strengths could potentially bump up CPU usage.  An algorithm must be in place to check the current path for obstructions and then another to find an alternate path that is clear - somewhere in the mix of all of that we would find our answer.  I don't even know if the alternate path I showed in my screenshot is the best one for signal strength of all the possible paths across my relay network, I only know it was better than the direct one when there is LOS to the KSC.

Still, it would be nice to know why it works this way and if it was indeed intentional or if I managed to uncover something odd or out of place.  Perhaps if @RoverDude is able to weigh in on how signal paths are selected it may lead to a better understanding of how best to lay out a network in our solar systems.  

[monstah]

18 hours ago, bewing said:

Weeeeellll, there are still some little logic problems left in the CommNet path selection. The problem is loops and dead-ends. Imagine you have 6 relays around Kerbin, with 100% signal strength to each other, and 75% signal to Kerbin. So your ship at Jool connects to one of these relays in the ring. And that relay bounces the message to some other relay that it has the strongest connection with. Which bounces it to another .... forever. So you can't just send it where the strongest signal is without more logic than that -- since the message eventually has to end up getting to Kerbin.

A*?