SENSOR DATA
IN BUSINESS
NIKO VUOKKO – SHARPER SHAPE
BIG DATA – IS IT JUST LARGE DATA?
• Data is gathered from multiple sources of varying quality
• Operational use of data is often awkward
• ”Data to knowledge” is a necessary intermediate step, but it’s
nothing more than intermediate
WHO GENERATES DATA?
Human-generated data:
• 5K tweets / s
• 25K events / s from a mobile game (about 200 GB / day)
• 40K Google searches / s
Machine-generated data:
• 5M bids / s at US options market
• 120 MB / s of diagnostics from a single gas turbine
• 1 PB / s at collision time from CERN LHC accelerator
SENSOR DATA
• Generated by machines
• May monitor a controlled subject: the myriad of sensors in a nuclear plant
• May observe the natural environment: ESA Copernicus satellites
• The Big Data upheaval has raised expectations about possibilities
OBSERVING THE ENVIRONMENT WITH SENSORS
• In data solutions often ”the real world comes and botches everything”
• Machine-generated data can usually avoid this problem
• Observing the environment with sensors, however, is an exception
WHAT ENVIRONMENT COULD WE OBSERVE?
• Space
• Natural phenomenons
• Human activity in large scale
• Infrastructure
MONITORING INFRASTRUCTURE
• We place high expectations for our public infrastructure
• Disruptions and breakdowns are very expensive
• The amount of infrastructure is colossal
• Typical infrastructure permits only external monitoring
WHY IS MONITORING INFRASTRUCTURE SO
EXPENSIVE?
• Monitoring needs to be extensive, continuous and diverse
• Many important observations are vague and cryptic even for a human
• Mistakes are very costly
HOW COULD WE SAVE ON THIS?
• Separate the data collection and evaluation phases
• Automate both!
• Machines can reliably run large and complex evaluations
• The data will have many unexpected use cases once it exists
DATA COLLECTION WITH UAVS
• UAVs minimize the data collection costs
• Access to hard-to-reach places
• Technology advances rapidly --> prices will drop further
• Europe has a single company with legal permits for BVLOS flights
AUTOMATIC EVALUATION OF THE DATA
• Machine is by far faster and more reliable evaluator than a human
• Proper use and complexity of data has been the historical problem
• Digitization of information allows for wholly new solutions
THE FUTURE OF SENSOR DATA
• Sensor size and cost will drop while quality and capabilities grow
• The last obstacle: Efficient distributed data gathering is only now
becoming feasible
• Massive potential of totally uprooting old business models