Ideal signal strength/quality and round-trip time for cellular connection in eastern Africa

Connectivity Cellular
1

Some of our clean water well remote monitors deployed to eastern Africa appear to be experiencing high latency while others are not.

Our remote monitors enqueue messages for hourly transmission and deliver messages to our data platform one by one (via Particle.publish()), waiting for an ACK in response (via Cloud API function call). The ACKs are not reaching the remote monitors before they time out and wait for the next hourly transmission attempt.

The monitoring of the water well and maintenance of the message queue are done on a microcontroller, and the Particle B5 SoM simply relays messages and ACKs between the microcontroller and our data platform, through the Particle Cloud, of course. The Particle device is powered off most of the time for power conservation, and establishes the cellular connection hourly.

Delivery of ACKs to our remote monitor in Kenya is going much better than for the one in Malawi. We're still early in our piloting of the this remote monitor, so don't yet have many in the field (not a lot of data points).

At a quick glance, device vitals for the one in Malawi indicate good signal strength/quality. Is the signal quality not as consistent as we're interpreting it to be, or is the round-trip time high enough to indicate problematic conditions? (Can't see how to upload the .csv file with history of device vitals. I plan to ask Claude's review of the data.)

Latest device vitals:

Jul 28th, 2026, 08:49AM
100% cellular signal strength
75% cellular signal quality
Operator: Airtel/Zain/Celtel ltd.
Access technology: 3G
Cell Global Identity: 650-10-9009-350494
3088ms round-trip time
83.5kB of 161.7kB of RAM used
0 cloud disconnect events
0 rate-limited publishes

Some of my pending steps are to dig into the timeouts in our microcontroller and Particle device firmware. I just deployed a fix to our data platform, increasing the timeouts there to be greater than the timeout for calling a function on the Particle device, so that our observability isn't artificially limited.

Thank you for any insights you can provide!

2

The function call made back to the device should occur pretty quickly, especially since the device just sent data in the other direction. At the lowest level, it's a CoAP transaction and will have three retries over a 20 second period before timing out at the call function API request level.

One non-network-related failure that can also occur is if you block the execution of loop for more than 20 seconds. Because function calls are dispatched between calls to loop, even in threaded mode, if you block loop for longer than the CoAP timeout they will appear to fail from the cloud side. I don't think that's happening to you, but it's worth remembering as a gotcha.