Making sense of MQTT sensor streams!

very rough and ready

//what's going on now

• the number of steps involved in typical IoT path is ridic.
• AWS Down? lol- now you cant use your light switch.
• If an IoT firm goes bust, you have a paperweight.
• That range of user frustration from such events ranges, based on what function that paperweight previously held - be it your find-my-thing, or, the WiFi friendly deadbolt on the front door
• some local autonomy would be cool, and unquestionably, cheaper. maybe, even, one could suggest, it could be a bit more secure?

//what would be cooler, though

• local hosted IoT sensor MQTT gateway / aggregator w/ the UNO Q

//why?

got a lot of sensors, and a lot of random wireless dev boards about, often grafted on to the back of an interesting sensor..

I like the Espressif ESP-IDF and ESP32 for raw power thrown at [whatever the task is], but I like the battery life of Nordic nRF parts and their developer offerings are very nice to work with,

ESP32S3 for WiFi MQTT

nRF51822's - because I have a bunch of of the same target board and because BLE because it has "low energy" in the dang'ol'name'dagnabitman

//which means

• I'm almost certainly ADHD, I am going to build one of something - two, tops, before going off to get interested in something else.

So, I've got a whole bunch of sensor / control endpoints on different interfaces (WiFi & BLE), running on whatever is on sale that month at AliExpress.

But, so far, this is all doing some heavy work with the ESP32 parts - things like NDIR CO2, or LiDAR, are power hungry enough, before bolting on a full WiFi block.

There isn't much sense in trying to power optimize a LiDAR, given how it works, but, for others, like CO2, or light/temp/RH, there are massive gains to be made in approaching a sensor node with low power operation in mind.

And then there are some really nice ways to visualize our environment, such as colour E-Ink

It would be a worthwhile project to break all the little chunks of processing up, and have the sensors doing sensor things - not rendering graphs for a physically coupled display.

I'd rather not have to prop up a typical maker SBC (cost, support components, power if nothing else)

• wouldn't it be nice to have an Arduino board with a Linux RT capable SoC built in?
• Something compatible with all the random bits I have in the toolbox, that not only would let me prototype new sensor nodes, but
• also allow me to do cool things like data processing and pattern analysis, all on board? Sensor readings with a cron job sounds nice...

//enter this Linux RT capable, Arduino Q board

//how to

no idea beyond this point, throw a board at me and I'll figure it out.

//////