With data flowing from ubiquitous sensors, the new field known as sensor data analytics, or SDA, is beginning to show its ultimate promise.
“Big data” analytics continues to gather massive amounts of attention and investment across the business spectrum.
Much of the analytics performed today is focused on understanding human behavior, such as uncovering insights and drawing conclusions from a consumer’s purchasing history, website browsing patterns, reading list, and movie preferences.
These analytics are designed to better understand consumers’ needs and to make companies’ sales and marketing more targeted, cost-effective and profitable. In these applications, the underlying data is about humans, is generated by humans, and analyzed by humans in the hopes of affecting human behavior.
With the growth and acceleration of the Internet of Things (IoT), connected devices will overtake humans as the most prevalent sources of “big data.”
While humans will still steer the analysis, pose the questions, and benefit from the conclusions, and while some of the data generated will always be about humans, the data gathering and some of the analysis will be increasingly performed by sensors and microcontrollers.
This ubiquity of sensors and the flow of data from them are leading to a new field known as sensor data analytics, or SDA, one that is only beginning to show its ultimate promise.
Sensor data analytics applied
One example advances in healthcare combines smart sensors, passive monitoring and data analytics and applies them to measure, monitor and interpret daily activity levels for patients. This is especially useful for elderly patients who are “aging in place”, those who want to live an independent life without full-time caregivers, but may need attention on an intermittent basis as physiological and behavior patterns change, sometimes suddenly.
Based on sensor data being gathered continuously, the analytics “learn” normal patient physiological and activity patterns, and can quickly alert family members and caregivers of departures from the norm (a drop in blood pressure, for example), in the hopes of heading off urgent situations (a fall brought on by dizziness due to that drop in blood pressure).
In other SDA applications in the IoT, many sensors will not be in motion — they will be embedded in devices at fixed locations and stationed within the communications ecosystem. Some sensors will have analytic capabilities on board, allowing them to not only gather and store information, but to “speak up” and/or initiate actions when needed. In the home, an intelligent stovetop could sense when water is boiling in a pot but the stove is unattended, and turn down the temperature to a simmer to avoid boil over or evaporating all the water before the cook can return.
For more demanding analytic tasks, the sensor data analytics could be passed to analytic engines operating in the cloud, making the connection to the IoT in that manner.
Two main categories
These sensor capabilities and information architecture open up many interesting possibilities, in two broad categories:
In the first category we can use sensor data to rethink conventional products and applications and redesign them to better serve our needs. Imagine a smart bed for everyday use at home (some specialized hospital beds have already incorporated some “smart” features, but at a high current cost). The consumer version of the smart bed will be designed with embedded sensors to monitor pulse rate and respiration and make that data available to you and to your primary care physician. This will allow the data to be analyzed and used to improve sleep quality, leading to more restful sleep and to happier and more productive days.
The second category opened by sensor data analytics involves the monetization of the information itself. Taking the hospital smart bed to the next level, the sensors in the bed can detect whether or not the patient has been moved or turned within a certain timeframe to help avoid skin ulcers or “bed sores”, a leading cause of hospital readmissions. Using this information proactively can not only improve care quality, but also have a major financial impact on the hospital’s billing and cost structure.
In the manufacturing or facilities world, a stationary motor presents a vibration signature as it rotates, and that vibration pattern can be recorded by an accelerometer. A healthy motor generates a frequency signature that is relatively constant.
As the motor wears out, a defective ball bearing or a slip in the gears will cause the motor to vibrate more vigorously and generate a broader vibration frequency spectrum. By detecting frequency content changes, sensor data can be analyzed to derive motor health over time. Armed with this information, factory managers could schedule preventive maintenance more efficiently, saving time and money.
Or, to use just one supply chain example, imagine the intelligent beer keg. Using sensor data to determine how much beer is left in the keg, bar owners and their distributors could derive real-time beer consumption information, allowing them to order beer only when it’s needed, avoiding lost sales opportunities and lowering inventory costs.
Next Page: Good news, bad news