Assuming I wish to build a mesh of a co-ordinator and several hundred devices ( all configured as routers ) how can I calculate and test the throughput of the system? How can I estimate the maximum number of meshable devices?
About 75% of the 'routers' are active machines reporting a few simple perameters back to the co-ordinator but also acting as potential routes home for the rest of the system. At peak load there may be a requirement to send data from a machine every 8 - 10 seconds or so seconds at most. This lasts for about 2 minutes ( and then there is a quiet period from that machine )
The balance of the routers are controls sending commands to a single router at similar rates. the 'controller' routers also send a small batch of status info destined for the co-ordinator.
While I can withstand a significant amount of latency in all of this data getting from the mesh back to the co-ordinator I need to ensure that the control to paired router commands is as responsive as possible.
Generally the controller is paired to its machine target in a hopless link although that can't be guaranteed it would be at most 1 hop and should be limited to no more than 2 hops. in the worst case i would expect data from the furthest end of the mesh to have perhaps 3 or 4 hops to get to the co-ordinator and hopefully at most 6 or 7.
Adding a dummy 'relay routers' to the mesh, just an 'xbee on a pole' would be a way to offer the mesh a multipath return to the co-ordinator if required.
The nearest analogy i can come up with is there are several lights scattered about ( machines ) and about 25% as many 'light switches'. A light switch can be asked to turn on 1, 2 or perhaps 4 lights. A simple packet of data is sent from the light switch to the co-ordinator to pass a pc attached to it that the light switch has been operated. The associated light then changes state and tells the co-ordinator that it has done so. every point in the 'lighting circuit' can tell the co-ordinator that the power is off or the bulb is blown, the light switches can do a similar thing.
The light switches need to occasionally ( every 3 to 5 minutes ) pass a packet of data ( < the maximum packet size .... 145 bytes rings a bell ) back to the co-ordinator.
I'd like to come up with some theoretical maximum mesh sizes given my paket sizes so that I can test it practically. ( unless of course the answer is that the mesh will handle the data throughput with ease