Hi,
I have a node sending data to an PI gateway with the iC880A-SPI concentrator HW. The typical signal related values I observer are around rssi: 99, snr: 2.
Still I observe around 10~20% of missed packet on my gateway side. What debugging techniques do you guys use/follow to diagnose out where I’m loosing packets.
Thanks, Marko
It seems you are already counting the packets you should be receiving.
You could monitor the gateway traffic (tx/rx) and see if the gateway is busy during some of your expected window. You can also monitor the “raw” udp frames being sent to the gateway-bridge, if both are not deployed on the gateway, you might be losing packets on the back-haul side of the network.
If you are able to isolate the gateway and sensor, you might be able to rule out interference, collisions and that sort of stuff.
Thanks for the tips.
Assuming I start with excluding the concentrator and the client itself I’m assuming I should see logs telling me there was a checksum error or some other error on the RF signalling level that caused the missing packet, right? Anything specific I should be looking on the gateway logs?
I think the gateway bridge should be outputting warnings at least if CRC-error packets are being forwarded. You can definitely dump the UDP frames to a file. No magic in there, I might be confusing something but I remember it as JSON encapsulated in UDP.
Hi there,
I am also interested in troubleshooting missing uplink packets.
I’m running into an issue where one brand will lose 10-15% of the packets and the other has 30-50% of packets are lost. Best practices are saying that a 10% loss is expected from the TTN documentation but I have the end nodes and gateways about 5-10 feet away from each other.
To also troubleshoot further, I ran both gateways near each other listening in on Ch 0 through 7 but still see similar percentages of lost packets for both brands of end nodes.
Hardware Used:
Dragino LSN50v2
Dragino LG308 Gateway
Moko LW001BG
Moko Gateway
Is it normal be losing these many packets when an end node and gateway have line of sight to each other? Any other recommendations on minimizing these lost uplink messages?
Thanks.
Hi @edwin , have you also set your end devices to use channels 0-7? I was losing many uplinks until I did that as well.
Regards,
chabral.
Thanks @chabral for the quick response.
Yes, I did have to set them to Channel 0 to 7 for both brands of end nodes. After setting them to those channels, they did call home but it is odd that I am seeing that much loss from each device.
I set the interval to every 10 minutes in case I was making them call home too frequently. I did test the end nodes individually (powering only one at a time on) as well in case I was causing interference with having all the end nodes on at the same time.
Let’s try this again, this time without setting off the “Your post was flagged as spam: the community feels it is an advertisement, something that is overly promotional in nature instead of being useful or relevant to the topic as expected” sentinel.
This spacing matters, as an knowledgeable engineer (to whom I owe a lot of education) explains in THIS VIDEO around 5:00 minute mark and also why it matters (has to do with some rules-of-thumb related to particular wavelength of LoRaWAN signal). You should be good >10 feet, but that’s a minimum.
I’m on the same way.
Actually I have some endnodes on the same area near the gateay (~10m).
I’m loosing ~10% of the packets.
Which is the best technique to debug ?
Thanks in advance!
Could your node transmissions be overlapping in time?
When close to a gateway, even theoretically non-conflicting transmissions (different frequency, especially different spreading factor) can practically interfere, both because the node transmissions aren’t perfectly, cleanly, on the intended frequency only, and because the signals are strong enough that the gateway front end isn’t completely linear with the result that multiple arriving signals blur together a little.
Also keep in mind that a gateway can’t receive from any node while it is responding to one. That’s why LoRaWAN is designed to be “uplink mostly” with downlinks rare. If you do something like use confirmed uplink, or have configuration mismatches where every uplink generates a downlink with yet unresponded MAC commands, then all that time your gateway(s) spend transmitting is time they can’t receive.
Thank you for the quick response.
I have all in multichannel and i see they transmit on differente freq.
The SF should be different from one to another?
I have all my endnodes (6 endnodes). Each of them send uplinks every 10 minutes, and they are not conflicting because there is >1 min the difference between packets.