My journey towards IoT skilling was accidental. I was asked by my bosses working then for an Automotive major to make a use case with security in mind for connected factory and digital twins.
Cloud was nascent then and seen more like a VM over the internet (IaaS as it came to be known later). Needless to say, We could hardly make any progress on the use case.
The struggle we had then were plenty – mainly concerns on security, easy first wins, prototypes and also on an approach for fast scalability. No one from business liked the idea even of being controlled from internet.
My second chance with IoT came while I was contemplating my own StartUp for hardware robotics and needed a chip which would talk to the internet, run a http server, read servers and do some event processing.I actually never liked the LED On Off gigs that used to run then as a demo. I then used an ethernet shield over a LAN using Arduino UNO and it worked but I did not like that I had to use a LAN cable.
Some research got me to ESP01 then – and it was a nightmare getting that to work. (We have ESP12E now - works like a cake).
Things have changed now and for good. I especially like the Azure IoT stack offered.
(Image left: Keya IoT Twitter plant I made back in 2014). Oh! My orchid.
(Azure IoT Stack - Source Microsoft docs)
My tips & steps for starters while starting Azure IoT Skills are as follows –
1. Start with an IoT Hub as it allows both cloud to device & device to cloud communication
2. Start with a NodeJS app to send and receive simulated telemetry to/from the IoT Hub
3. Try to interface the simulated date over a Azure blob
4. Use a real sensor now, you can use DHT 11 or DHT22 or BMP180 (Real cheap) with ESP8266 (Use ESP12E board) to replicate steps 2,3
5. Go serverless & try to use Azure functions now, example if the temperature is above 30 degrees send an email
6. Finally, stick an ML algorithm within Stream analytics to predict rainfall using humidity & temperature values from the sensor. The AI app is found in the gallery. I never got a great model but in range of 60% probability. (Does the job for now - Comment up if you got a better score).
By being successful, in these six steps, you would be much more confident to skills on the SaaS aspect of IoT.
You may also consider using a Raspberry Pi instead of ESP12E/ESP Huzzah but it would cost you more plus you need to get that Pi is a Single board computer and ESP is a microcontroller. A microcontroller is much better suited to run a specific hardware task (read sensor here) in an infinite loop. But again, if you have a Pi Handy – Get it headless and connect the sensors.
Shoutout@ I have launched an Iot Series over Udemy and if you have liked this approach and need extra support in a short time then check out below courses from me. You wil have my 121 support via Udemy in this course via QA to get you a successful PoC for IoT.