Toby McClean recently noted on LinkedIn Pulse that the value of sensor data is lost when each and every device maker or IoT system defines a new model of observation or measurement.
“The value is decreased because third-parties are not able to build reusable algorithms, analytics and visualizations,” he opined. “It is difficult to have reusability when there is no common model for sensor data; for example if every thermometer manufacturer had a different model for temperature.”
In order for it to make business sense for a third-party to build an industry or vertical specific analytics algorithm or visualization, says McClean, it must be able to work with almost all sensors of a particular type.
Patrick Gill, a Principal Research Scientist at Rambus, said one must also understand the context in which sensor data is gathered.
“For example, take an IoT thermometer. It might read the temperature outdoors, indoors, or in the wort of my latest homebrew beer. It’s thus not clear that automatic aggregation of all available temperature data will give useful results,” he explained. “Moreover, we need to understand the privacy and security implications of each piece of data individually. How can each measurement be shared or aggregated? What actuators are allowed to make the data useful without letting crooks know too much about, say, my house occupancy?”
These are all tricky issues, Gill emphasizes, and they potentially may very well have to be addressed differently for every IoT widget out there.
“I would say that looking up the format of the data (so long as the sensor has documentation) is significantly less work than figuring out the use cases of each sensor. Overall, standardizing the data format is not a big concern,” he continued. “I wouldn’t be surprised if some XML schemas turn up in more than one product; however I don’t think that having a standardized data format will remove more than a tiny portion of the work needed to make a compelling and safe IoT product.”
James Tringali, a Technical Director at Rambus, expressed similar sentiments.
“The establishment of common models tends to be driven top down from an ecosystem perspective. Software solution providers sitting atop the ecosystem, such as Google (Nest), Apple and Microsoft, have already crafted XML-like schemas to help their products sort out the sensors that matter to them. Over time, folks bringing in new thermometers, accelerometers, etc. will want to make sure their product’s API play nice with whatever software solution looks to be gaining the most market traction,” he concluded.