stablizing latency over wireless

802.11 doesnt negotiate speed.
Each peer sends speeds they are capable of as on option. Best speed is selected by gradually sending at a higher speed until it has to retransmit at which it takes a step back. This retrying continues to always keep the optimal speed (read: highest number).
802.11 was not exactly designed for fixed wireless. Roaming and simplicity of protocol are corner stones in the design. Efficiency and “intelligence” are not.
Cost and availability got people to adopt it for other things such as fixed wireless.
It’s far from the most effective system but it made wireless available to everyone.

well, I get this feeling that 2 end can actually transmit at maximum speed but the remote (receiver) will stop the sender at certain speed or number of packet (kinda like forcing to send slower)…

I’m so scared of upgrading to 2.9 and might blow off crashing because I tried some of the RC release and it keep crashing on my hardware setup and I found 2.8.18 more stable even than 26 or 21 because of that Kernel Panic crashing thing…

however whats the best setup for frame policy and size? in the winbox there is 5 option while in terminal there is 3 option

why?

thanks

The only thing the receiving end does is to ACK the packet so that that the transmitting end knows it got there and can start transmitting the next packet. The ACK would be the only kind of flow control. Noise will also prevent transmitter from transmitting.
Are you sure you don’t create your own interference? Short’s in radio cables and to some extent radio reflection can do this. The spikes in round trip times can indicate some synchronisation problems in hardware but also retransmissions due to noise or corruption which is much more common.