I’m really hoping someone can respond because similar questions have been asked on these forums going back years with no replies
I have a network project i’m working on, we absolutely need a more solid and reliable wireless system. We have AP’s that actually move around in the network so we cannot predict where they will be, what order they will be in, and what other AP’s they will be able to connect to
Our only real issue at the moment is that HWMP+ is absolutely insisting on using the most direct path to the destination. Meaning if the destination is C and a packet comes from A it will always go A->C even if the link between the 2 is absolutely terrible and borderline useless. Despite there being a much, much, much better link if it went A->B->C. P throughput from A to C could be ~30mbit while A to B and B to C are both 500mbit, but the metric values seem to ALWAYS be higher on the A->B->C link and hence its never used unless C moves completely out of range of A and never connects
Is there a way - even if I script it myself - to influence the metric values in the forwarding database of a mesh?
All the documentation I read says HWMP+ adjusts this dynamically (and yes I have re-optimize paths ticked) but with NO other options at all, no information as to how its calculated. Are there set thresholds? i.e. P throughput at 10mbit = X 20mbit = Y 30mbit = Z etc. Is it calculated against other neighbors?
Bottom line is I need a better meshing solution than we currently have. ONLY way so far to make it work is to force a client onto a connect list that bars them from connecting, but this is not at all viable because the position of all AP’s changes over time
I have also wondered why MikroTik, being a routing and radio company, does not have a routing protocol suitable for radio links.
I would expect some routing protocol that can work with radio-oriented metrics like CCQ, SNR and rate.
I asked it to support but it appears to be the reality with little indication of change.
(in my case I was looking for an IP routing protocol that functions well in a routed network with fixed PtP radio links, a somewhat
different situation from yours but still not well supported, same issues: routing protocols that favour “shortest path” (least hops)
which is often not optimal in radio networks.
We are considering writing scripts or external programs that read radio parameters and adjust routing protocol metrics.
I designed a protocol for that purpose, RSPF, around 30 years ago. Over the following 15 years, a few partial Linux implementations were done. This predated 802.11, and I never got around to writing the v3 that would support that (it was designed around AX.25).
I presume you are K1IO? Our application for this is amateur radio oriented as well, as you may have guessed from my username
I am a bit wary about a protocol with the acronym “SPF” in its name, as it is so obvious that the shortest path is rarely optimal in these networks.
However, that may just be something that got stuck in the name and not really reflects the behavior.
What I am after is a protocol that is clever enough that a path of 3-4 hops may actually perform better than a 1-hop (direct link) path.
(depending on the actual CCQ, SNR, effective rate and maybe even the average loading of the links)
