In general, we can divide IoT into 4 categories:
1. Smart Cities - where the city uses technology to improve the quality of life of people who live and work in it,
2. Smart Homes - we use the “Smart” Devices to control the environment of our home,
3. Connected Health - How to use technology products that each of us can buy and connect the patient with the doctor who monitors him Without having to go to the hospital.
4. Industrial IoT - We use ML & Big data to generate value from sensor data. (That’s what this article is all about.)
Recently I’ve finished the course “Industrial IoT on Google Cloud Platform”. This course assumes that the person who will enrol has no idea either from IoT or from the Google Cloud Platform, at least that’s what I felt (and it’s better this way for me!).
So, it starts by explaining what IoT is, how it started, how it splits, how it works. For the first time, I saw a lesson that is not only with video speeches but has many small and easy to read texts with pictures and graphs. Groundbreaking and brilliant!
The architecture of Google Cloud IoT is divided into 4 stages: Data gathering, data ingest, data processing and data analysis.
So, the course takes you step by step in all these stages. You learn to:
1. Activate the appropriate API (Cloud IoT Core, Cloud Pub/Sub, Container Registry are the standard),
2. You use Cloud IoT Core for device security,
3. Create Publisher and Add new Subscriber, in Cloud Pub/Sub to receive messages from devices and to send (publish) to Subscribers (subscribers) to read them,
4. You use cloud Dataflow to create data pipelines from the device to the destination, which can be BigQuery, Cloud Storage, or Big Table.
5. You create your own pipeline with the ready-made templates that Google gives you. Using Cloud Functions to make your own custom pipeline.
Immediately after the theory, you go to the labs. Here you must pay great attention! Use the Student account and not one of your own Gmail. That’s why you’d better do your lesson in Incognito or whatever is called in your browser in order to avoid charges!
In the 1st Qwiklab you will create your own Pub/Sub topic, subscription to the topic, you will manually send a message to publish and you will “pull”, again manually, your message from the topic.
At the next Qwiklab you will repeat what you did in the previous one and you will proceed to the new exercises. This makes the friction, the practice even better.
At some point in the course, they show how to use Cloud Dataprep. It is from Trifacta and it is used by Google. Either way, it’s a fantastic tool. Without having to know how to write code, only by changing things in the UI you control and change your data.
I’ve worked a little bit with it and I have to tell you that I need, at least, an article only for that. It automatically detects the “shape”, the types, links, and anomalies in the database you give it. And you have direct visualization of your data.
With Data Studio You create your reports, share them with your partners, give them the right (role) to edit them or only read them. And all these by whatever data you’ve taken from your devices.
Finishing, the 18-hour course has 3 Capstone Projects where you can check what you’ve learned. They differ in a few things, especially in the device to be used:
1. Download the application Cloud IoT Inspector from the Google PlayStore in your Android Device,
2. If you don’t have an Android Device, then simulate a virtual device and you move on with this project,
3. In the last project Re-use simulation device
All projects Work with MQTT and not HTTP.
· A restful course
· It takes you a step forward, both in theory and in practice.
· In the labs, you repeat what you did until that moment, and then goes on to the new ones.
· The last 3 projects where you practice more.
· You gain more confidence in your abilities but also in the use of GCP.
My comment: IoT, as a name, became known in 1999. But for me the first time I saw the IoT (and the ML) was in the film Electric Dreams in 1984.