Hoping someone can help reduce my implementation time
I'm implementing a network that is infact a roaming mesh whereby the AP's themselves physically move around. This presents a number of challenges in trying to maintain a minimum throughput and dynamically recalculate the paths through the mesh on-the-fly to ensure the best throughput to each node/AP
Am currently using WDS and a flat layer 2 network. There are 2 important pieces to the network, a control center and an end-point. The control point stays stationary but is connected to the wireless mesh, all nodes and the end-point move around. More nodes are added as more coverage is required. The problem is that all devices hang onto their existing connections for far too long. If the control center and end-point are connected to each other with a very weak signal, and another AP is placed in between to help boost coverage, they both connect to the extra node but they don't use it to pass traffic, they would rather talk to each other. But the signal is bad so the bandwidth is very low. I'm relying on connect lists and scripts to try and force a disconnect and reconnect so they will recalculate a better path to send traffic through, but i'm having big issues with having it drop either too early or too late
I need a way for the network to pick a better path much more quickly. I'm wondering if I should stay with WDS or instead move to having all devices as AP's. And maybe change to a routed network design?
The big problem here is more nodes need to dynamically be added in the future. So I can't use static mesh (must be dynamic) and I can't predict the L2 topology
It might be CC->Node1->Node4->End
Or it might be CC->Node3->Node6->End
Or even CC->End
So adding 1 subnet per device on every device is not scalable and managable
Can anyone provide any additional help?