>Hi Dennis,
>
>Another approach is to make your system more "polite" at the AX.25 layer.
>This is normally done by adjusting a parameter known as P-Persistance. It
>is usually set to be (1 / number of channel users), assuming that the
>number can be varied between zero and one.
>
>Most TNC implementations offer a numeric range between zero and 255. So
>the formula for computing the P-Persistance becomes (256 / number of
>channel users). The theory here is that when your system has data to
>transmit, it generates a random number from zero to 255. It checks to see
>if the channel is already busy, and if not, then it compares the random
>number with the P-Persistance number you calculated. If the random number
>is less than the P-Persistance number you have configured, then the
>transmitter will be keyed, and the packet will be transmitted. If the
>random number is higher than the configured P-Persistance number, the
>transmission is not made.
>
>At this point another parameter, refered to as Slot Time come in to play.
>It determines how long to wait before trying to generate a new random
>number and repeating the process above.
>The theory behind setting slot time is that it should be equal to the
>amount of time it takes your station to determine that the channel is busy,
>once another station has started a transmission. Normally for 1200 bps
>packet, this is set for 100 milliseconds, and is normally a good choice,
>considering the radio/TNC combination at most stations.
>
>If you set the Slot Time to a larger value, or the P-Persistance to a
>smaller value, you become a more polite neighbor on the channel. You may
>still occupy a large percentage of the channel's capacity, however you will
>wait longer (giving others more of a chance to get their data through)
>before each transmission you make. If no one else keyed up, then the fact
>that you occupied the channel just means that at least someone is making
>use of the available bandwidth, rather than it going to waste.
>
>I've not looked for the window size setting for a TCP connection in the
>linux setup for a long time, but I think that's really the value you will
>need to reduce if the concern is how long your longest transmissions
>become. The window size governs how much data the connection will commit
>to having in transit (not acknowledged) before it allows more data to be
>queued for transmission.
I've changed the setting wich you mentioned above. As a result, my ax25
host waits a bit longer before transmitting, so that other stations also
can broadcast on the channel. But the new settings don't affect the amount
of frame's transmitted after each other... This is stil a problem. A few
minutes ago my ax25-host transmitted 8-frames(!!) in a row...., this took
about 15-20 seconds.
Maybe there are other settings wich can regulate the maximum amount of
frames transmitted at a time??
I've got another question, does anyone know how to calculate the QTH-Locator?
Greetz...
Dennis D.
The Netherlands
