>From C, sigspy shows >A = Quality:246/92 Signal level:-27 dBm Noise level:-37 dBm >B = Quality:238/92 Signal level:-30 dBm Noise level:-37 dBm > >Using WIANA, the Neighbour Status table shows >C to A @ -68 >C to B @ -78 > >Using WIANA, the Neighbor Status picture shows >C to A @ 103 >C to B @ 97 > >Now by my reckoning, C to A is the better signal (-68), yet the node >selected B. >Why is that? > AODV works on the shortest distance, and they are both the same number of hops. There are also algorithms to test signal strength against threshold values (minsig), and to test the link to see that data is sent across it successfully. These apply in this instance as both gateways are one hop away, but both appear to be well above these values. So, with two routes of equal value, it ends up that the node chooses whichever gateway responds first. To modify this behaviour, the Preferred Gateway and Lock to Gateway settings exist. It is not whether a node has the best link (as the gateway might be several hops away, the client meshbox won't know which is the best link to follow), but whether the links are good enough. If you have 11Mbps traffic, then increased signal strength isn't going to help you a whole lot.
| i promise not to keep asking so | simple questions but i need these answers so i can start my research in my | library of networking books
Routing on the internet is more stable because in general the internet is a star network (Like cable TV) and where there are alternatives they are resolved by routing protocols which are deliberately slugged to be slow to change, and thus minimise the routes switching around.
In your reading look for "routing flap"
Also on the internet the paths are almost always asymetric. This is because ISPs want to do, and BGP the inter-isp routing protocol facilitates, something colloquially known as "hot potato routing." Hot Potato routing means getting the packet onto the target ISP's network as fast as possible (the network is identified by an ASN -- Autonomus System Number. This is partly why one gets extremely variable VoIP performance on long-haul, unless the person you're phoning uses the same (Tier 2, Tier 1) ISP
There is a good discussion of routing in Computer Networks by Andrew S Tannenbaum from Prentice-Hall. And yes, despite the advert and the confrontational tone Jeremy is right, for a fixed (wireless) network conventional routing is probably more appropriate, but for the original community wireless goals AODV has the big benefit of not needing someone vaguely IP literate to get things going, all you need do is set your ESSID and put a stick in the air