Sensor analytics is the statistical analysis of data that is created by wired or wireless sensors.
A primary goal of sensor analytics is to detect anomalies. The insight that is gained by examining deviations from an established point of reference can have many uses, including predicting and proactively preventing equipment failure in a manufacturing plant, alerting a nurse in an electronic intensive care unit (eICU) when a patient's blood pressure drops, or allowing a data center administrator to make data-driven decisions about heating, ventilating and air conditioning (HVAC).
Because sensors are often always on, it can be challenging to collect, store and interpret the tremendous amount of data they create. A sensor analytics system can help by integrating event-monitoring, storage and analytics software in a cohesive package that will provide a holistic view of sensor data. Such a system has three parts: the sensors that monitor events in real-time, a scalable data store and an analytics engine. Instead of analyzing all data as it is being created, many engines perform time-series or event-driven analytics, using algorithms to sample data and sophisticated data modeling techniques to predict outcomes. These approaches may change, however, as advancements in big data analytics, object storage and event stream processing technologies make real-time analysis easier and less expensive to carry out.
Most sensor analytics systems analyze data at the source as well as in the cloud. Intermediate data analysis may also be carried out at a sensor hub that accepts inputs from multiple sensors, including accelerometers, gyroscopes, magnetometers and pressure sensors. The purpose of intermediate data analysis is to filter data locally and reduce the amount of data that needs to be transported to the cloud. This is often done for efficiency reasons, but it may also be carried out for security and compliance reasons.
The power of sensor analytics comes from not only quantifying data at a particular point in time, but by putting the data in context over time and examining how it correlates with other, related data. It is expected that as the Internet of Things (IoT) becomes a mainstream concern for many industries and wireless sensor networks become ubiquitous, the need for data scientists and other professionals who can work with the data that sensors create will grow -- as will the demand for data artists and software that helps analysts present data in a way that's useful and easily understood.