Improving ACK in sketch PiGateway.ino vs Gateway.ino

[joelucid]

Would you consider 96% of success in getting ACK a good value?

I track retransmit rate for my sensors. After running for a couple of months with excellent RSSI the three AC-powered Motes I deploy (these are the only ones that use acks) have a retransmit rate of 1.5%, 1.6% and 1.8%. This is not Tom's implementation, but I also tune transmit power.

If your GW is a raspberry pi before version 2 and you're using 433Mhz that could explain transmit problems. I couldn't get two version 1 Pi's to work well at all as GW's. Version 2 works better but there are still interference issues. Specifically I have a "2 phase" RSSI problem: From the same Mote the GW will see packets sent at the same power at two very different RSSI levels. As an example one node comes in at -61 or -42, but really not much in between. This problem only exists when the GW moteino is hooked up directly to the serial connection of the Pi (not USB, which is too high latency for my purposes).

Probably not your issue - but I like to throw it out from time to time to see if anybody has a solution 

Joe

[Sergegsx]

I have the GW connected to a Raspberry Pi 1, but via a USB converter. Does this mean something in your experience?
Why do you mention RPi1 and 433 ? I do have that combination but as I say using usb converter.

what do you mean by retransmite rate of 1.6%?

my 96% means that after the default 3 retries done by send command. 96% of the messages get a correct ACK back.

thanks

[joelucid]

I have the GW connected to a Raspberry Pi 1, but via a USB converter.

I have never tried with FTDI adapter but always connect directly to the Pi's UART so this still could be problematic with the pi1.

Why do you mention RPi1 and 433 ? I do have that combination but as I say using usb converter.

I mention it because some people here have had these issues with the Pi 1 and as I remember they all had 433mhz. Others (including Felix) never saw problems with the Pi1 and at least Felix probably doesn't use 433mhz.

what do you mean by retransmite rate of 1.6%?

my 96% means that after the default 3 retries done by send command. 96% of the messages get a correct ACK back.

Oh that's much worse than mine then. Mine is the number of failed sends (because of missing ack) divided by the number of updates sent (several attempts to send one update counting as 1 update). So if one update fails completely after 3 attempts that would count as 3 failed sends and 1 update or a retransmit rate of 300% for that one update. Missing ack rate would really be a better name for the measure.

If your 96% successful transmits all succeeded on the first attempt (not likely) that would translate into a retransmit rate of 12% in my world. I wouldn't be surprised if it were >20% in your case given that 4% of your updates fail 3x in series. That is definitely a very bad ratio of unsuccessfully sent packets.

If you want to test if its the Pi simply disconnect your GW from the Pi and check if things get better. I think I connected it to a macbook instead back when I debugged these issues.