I've been looking into the existing radio communications protocol options that have a bit more throughput than the typical utility-metering applications, and with the added relaxations that it doesn't need to be low power, and it would be used exclusively in outdoors/remote areas with no existing infrastructure.
The task is to link an agricultural robot to its operator, with the operator being some nontrivial distance away (typically 1km, but we want to support 3-5km, of course assuming line-of-sight visibility). The fields where the robot would operate will be remote and far from any infrastructure, thus we expect that there will most likely be either very patchy, or no cellular coverage (otherwise we'd just use that). It is not required for the robot to connect to the internet or any cloud - what's important is to be able to:
- receive commands from the human operator (1kbps would suffice here);
- send telemetry to the operator (3-10kbps would be fine);
- send occasional image for human verification (i.e. suspected obstacle, ML decision data, crash dump) - in order for this to work with reasonable reaction speeds, I'd be happy with 100kbps or more;
- typically an operator will handle several robots, but let's say no more than 8, with 2 or 3 being typical.
What I've looked into
- LoRa: the obvious choice, but it's too slow, unless I'm missing something;
- Sigfox: way too slow;
- DASH7: seems the best fit, but a bit too low-power and the 5km range is not thoroughly corroborated besides the Wikipedia table, so it's likely an overly optimistic thing; likely a 500-1000m real-world range;
- XBee, specifically XBee SX 900 / 868: seems a bit slow, but otherwise very good. I wouldn't be able to use its mesh capabilities, as the setup assumes no infrastructure being built by us;
- Weightless-W: I'm not sure if this is not a regional thing, and whether it's actively supported nowadays;
- Espressif's Long Range mode: a bit short-ranged, but the bitrate is superb. We have no problem enforcing both nodes to have an ESP chip at the front end.
Also, I'm not limited to any specific hardware, and most importantly, I'm not really limited by power constraints. It's more likely that we bump into RF regulatory constraints rather than battery power considerations. Both sides will have powerful PCs (e.g. Jetson on the robot side, a smartphone + some powered add-on dongle, or a RaspberryPi on the operator side). It is likely that this radio link will be the only serious radio buzz in the surrounding area: no cellular, no WiFi, no LoRa, maybe not even walkie-talkies. Yes, our operators will be brave souls :)
Directed antennae would be an easy fix for the range limitations of DASH7 or Espressif WiFi-LR, but that adds complexity we'd like to avoid unless it's absolutely required.
The application is not military, and we'd like to stay away from military technologies, if possible.
