> Note the high ETX between 4 and 26. The link is not especially long. > The loss is symmetrical. The reason for this poor link is that 4 and > 26 have at least twice as many neighbours as the other routers.
For me this implies that a mesh node having a lot of neighbours is a bad thing. Is that true? If it is, do we need to take steps such as having the radios reduce power when the MP can see too many nodes or by choosing a directional antenna?
I worry about this issue in the context of a Village Telco in a township which, if successful, could have quite a dense and closely populated mesh.
<steve.s...@shuttleworthfoundation.org> wrote: > For me this implies that a mesh node having a lot of neighbours is a bad > thing. Is that true?
Yup - any neighbour that you're not communicating with is interference.
> If it is, do we need to take steps such as having > the radios reduce power when the MP can see too many nodes or by > choosing a directional antenna?
I haven't tried any reflexive methods of power reduction, i.e. power reduction on neighbours of a problem node. Each neighbour would have its own signal issues. The best is to make reception directional, the easiest of which is siting the MP against a wall or tree (hence the request for the etched antenna to be 8mm away from the back of the casing).
> I worry about this issue in the context of a Village Telco in a township > which, if successful, could have quite a dense and closely populated mesh.
Where the mesh gets dense, most MPs can sit on the table. This is where the LCD display would be especially handy, as it would be helpful to "see" the scattered indoor signal.
> For me this implies that a mesh node having a lot of neighbours is a bad > thing. Is that true? If it is, do we need to take steps such as having > the radios reduce power when the MP can see too many nodes or by > choosing a directional antenna?
we should keep the number of packets in the air at a minimum in order to scale well. For example we are not sending any WiFi beacons or probe responses. Also the number of routing protocol packets should be kept as low as possible. I think we have made a good decision to use three non-overlapping wireless channels and cover sectors with sector antennas in the network center.
> I worry about this issue in the context of a Village Telco in a township > which, if successful, could have quite a dense and closely populated mesh.
I think 17 simultaneous calls per channel in such a environment is a reasonable goal. If we are hitting a capacity limit because of the population density we have to add more central points / high sides with sector antennas and decrease the transmit power level. That is also what GSM does in a city center like here in Berlin. Here the GSM cells have a radius of ~800 meters.
We can interconnect those high sides with 5 GHz point-to-point links and use Asterisk trunking to interconnect them.
|
