In this video:
- Bernard Kufluk, Watson IoT Platform Product Manager, IBM
He steps you through 4 stages of IoT app development:
- Device registration & connection
- Data governance
- IoT app dashboards
- Data analytics
IoT device registration and connection
(0:37) Working within Watson IoT Platform, he shows you how to register, connect, and configure a Raspberry Pi as a gateway. Then, he shows you how the gateway device automatically registers the Arduino motor controller and its attached sensors.
(7:10) Finally, he demonstrates how to download and update the firmware on the Arduino motor controller so it understands the additional sensor readings. He shows the devices view and the actions view to explore the device metadata and the logs to see the completed actions.
(11:09) Working within Watson IoT Platform, he shows how to work with the data, both raw data and calculated data, to make it more consistent for IoT apps to use. For processing data, you can connect apps directly to MQTT API or connect your IoT Platform instance to MessageHub, a Kafka based message service.
For storing the IoT data, he shows how you can send it to a Cloudant database, configuring the Historical Data Storage in Watson IoT Platform.
Finally, he shows how to use Watson Data Platform to find and add appropriate data sets to a data science project, and use the sandbox to gather the right data together to work on. You create a pipeline, extract it, writing it to a CSV file that you can do data science or self-service analytics on it. The extracted data set is added to the project.
Dashboards and the developer experience
(23:14) Back in Watson IoT Platform, he demonstrates the built-in dashboards and the dashboards you can create and customize for your IoT application. You can add various cards based on different chart types or gauges to look at your data coming from your IoT devices. These dashboards are useful for system administrators, but you also can use them to create portals with real-time data for the end-users of your IoT app.
(30:20) Next, he shows the rules section of Watson IoT Platform, that defines triggers and actions for working with the IoT data. You can control when alerts are sent. He also showed how the cards in the dashboards updated to show alerts being sent from rules being triggered.
Finally, he shows the security policies that can be defined, including how devices can authenticate with the Watson IoT Platform. He shared the risk and security overview dashboard, which shows which devices or apps are complying with the security policies.
Data analytics and data exploration
(33:51) In IBM Cloud, he shows how you can store the IoT data coming from Watson IoT Platform into IBM Cloud Object Storage. Then, he shows how to use the IBM Analytics Engine, which is backed by Apache Spark and Kubernetes, to provide the analytics processing power.
Working within the IBM Data Science Experience, he shows how to directly stream real-time data from Watson IoT Platform to Object Storage and do some basic analytics when receiving the data. He shows how to build a supervised machine learning model, and then deploy that model to Watson Machine Learning service running in IBM Cloud.
Lastly, he shows how the Data Science Experience supports all open frameworks, like Keras and TensorFlow to achieve a similar machine learning model. Then, he shows how to send the anomaly score back to Watson IoT Platform using MQTT so that other devices can pick up that score and take an action based on that data analysis.
Resources for you
- Watson IoT Platform Developer Community
- Watson Data Platform Developer Community
- IBM Data Science Experience Videos
- Get started with IBM Watson Machine Learning
- Try these products out for yourself using these developerWorks tutorials: