SXSW Interactive 2015
Real World: Life of Data from Sensor to Analysis
It seems simple - slap a sensor on an Arduino or Raspberry Pi, write some code, collect and analyze the data and then figure out how to sell all or part of that system as a product. Easy to prototype at home, ask for $ on Kickstarter, and bam, you are in business!
Not. So. Fast.
You've just built a complex system of systems. Your sensor interacts with the real world and a microcontroller + communications. That hack, er, endpoint then interacts with a set of networks - access points, gateways, routers, switches - to deliver data to a cloud data center. Lots of endpoints start to impact multiple networks. Your cloud has to ingest/ingress all of the incoming data the billions of endpoints you hope to sell. And you still have to make sense of the sensor data at many points across those systems.
Eventually you'll want to build a second generation endpoint, probably with an upgraded sensor or more sensors. Time to rewrite lots of code?
Let's talk with some experts about planning ahead.
Share this idea
Additional Supporting Materials
- What is does data normalization mean across lots of identical sensors, different types of sensors measuring the same thing, or even multiple generations of a single sensor technology?
- Do I need to run my analytics in a cloud? Where does it make sense to analyze my sensor data at other places in my end-to-end solution?
- What does "data hygiene" mean and how does it apply to IoT - sensor data, big data, and all that stuff?
- How do I know if my sensors are measuring what I want to them to measure?
- What is the difference between feedback control loops and software defined intelligent systems? Between reaction and insightful changes to operations?
Gina Longoria, Industry Analyst, Moor Insights and Strategy