The Internet of Things (IoT) is a hot topic, but it’s about more than just connected devices. The real value in IoT comes when you begin to use the data in creative ways. The December focus for the Data Platform and Advanced Analytics Partner Community is about how real-time analytics can introduce new opportunities for partners and create business value in existing IoT systems.
Turning device data into intelligence
At the most basic level, IoT offers information from devices, sensors websites, social media, applications and more. By creating Microsoft Power BI dashboards and reports, you can see both current device status and key device data over time.
What happened? Was a device unavailable? Did a device overheat? Did performance degrade or improve?
Once you start storing IoT data and understanding what happened, you are ready to start looking at why this may have happened. Interactive dashboards can help identify the trends and patterns leading up to a specific outcome, good (performance improvement) or bad (failure).
Why did it happen? Did the device overheat before failing? Where did things change from a normal state and start to go wrong?
After you identify why something happened, you can become more responsive in correcting similar situations in the future. By using machine learning, you can learn which observations are predictors, then watch for those conditions and proactively handle the situation.
What will happen? Have conditions leading to an upcoming failure been detected? Send notification alerts or create a maintenance ticket to avoid hardware failure or downtime.
After understanding device data and process lifecycle, you can focus on adding additional intelligence with advanced analytics to improve efficiency and productivity. By looking for opportunities for change and improvement, you can build a system that makes valuable recommendations based on insight and automates control in known situations.
What should be done? Should a command be sent to a device to control and improve its condition? Should additional devices be activated up to help with the load?
Incorporating real-time analytics
Microsoft Azure Stream Analytics is a cost-effective way to quickly add real-time analytics to your IoT solution. Data in motion through Azure IoT Hub or Event Hubs can be queried with SAQL (Stream Analytics Query Language). This query language combines basic SQL syntax with real-time extensions, such as windowing (HoppingWindow, SlidingWindow, TumblingWindow) and scaling (WITH, PARTITION BY, OVER). Rules are added to provide insights, so that corresponding actions can be processed, such as sending alert notifications, controlling devices, or creating a service ticket. Incorporating predictors from a machine learning experiment into real-time analytics allows quick action to be taken before a known issue becomes worse, eventually causing downtime.
A few basics
- Set up rules to recognize specific data cases
- Control devices based on these rules
- Publish events for Event Hubs to read
- Web Jobs read the published events and act on them by communicating with other Azure App Services (i.e. Web Apps, Mobile Apps, Logic Apps)
- Pass data to storage for use with Azure Machine Learning
Consider these common real-time use cases as you think about using Azure Stream Analytics in your own solutions.
- Call center analytics
- Connected car scenarios
- CRM sales alerts for customer scenarios
- Cross sell offers
- Customer behavior prediction
- Data and identity protection services
- Financial portfolio alerts
- Fleet monitoring
- Fraud detection
- IT infrastructure and network monitoring
- Log analytics
- Medical patient monitoring
- Predictive maintenance
- Sales tracking
- Smart grid
- Streaming ETL
Stream Analytics machine learning functions
Incorporating machine learning with real-time analysis allows you to dynamically predict an outcome or threshold and handle conditions appropriately. Azure Machine Learning can publish web endpoints for operationalized models. Azure Stream Analytics can then bind custom function names to such web endpoints.
Streaming Datasets in Microsoft Power BI
New Azure Stream Analytics and Power BI features, currently in private preview, will soon allow:
- Azure Stream Analytics jobs to output to Power BI streaming datasets
- The ability to create Power BI streaming tiles based on Stream Analytics output
This feature set will allow real-time dashboards to show the latest value from the Stream Analytics output and values over a set time window.
- Hands-on with Azure Stream Analytics [Microsoft Virtual Academy training]
- Stream Analytics Query Language Reference [Article]
- Get started with Azure Stream Analytics to process data from IoT devices [Article]
- Getting started using Stream Analytics – Real-time fraud detection [Article]
Data Platform and Advanced Analytics Partner Community
- Sign up for the December 13 Data and Analytics Partner community call
- Community call schedule
- Yammer group
- Blog series
- Training and enablement