The Internet of Things (IoT) promises to help us slash waste, transcend time and space and give us a greater awareness of the world around us. In fact, it seems that IoT is poised to dramatically improve efficiencies in virtually all major industries. However, with this growth comes new challenges for which we need to develop localised storage, cloud and data analytics solutions to harvest the economic benefits.
There are billions of connected sensors and devices, many of them running in private clouds and networks initially for security and IP reasons.
Take the electric utility industry for example: power plants, the grid, etc. are built predominantly to deliver electricity during a small peak period during a regular workday. They otherwise remain significantly underutilised during the remainder of the day. What if we could connect all of these power-consuming machines and regulate them centrally? What if, during that peak hour, these connected systems would automatically increase the temperature of air conditioning by one degree, turn off dim lighting in unused rooms, or stop washing machines for even 15 minutes, and most importantly draw energy from electric car or home batteries to help bring down demand spikes and, in turn, help improve the utilization rate of the whole infrastructure?
The astounding growth of data
We spend hundreds of billions on cars, roads and maintenance, but most of us end up using those assets for a fraction of the day. Smart sensors that actively manage traffic coupled with cell phones, GPS systems and that – in the future – engage with autonomous cars could dramatically improve the utilisation of those assets while bringing down the cost for each person using them.
This astounding growth of data, devices and applications – driven by the productivity and economic advantages that IoT will provide – will require a massive investment in hardware and software infrastructure, and may also force us to rethink how we manage the digital universe by moving big data analytics to the edge of the cloud.
Consider this scenario: Suppose you want to mine data from security cameras to get actionable data from facial or license plate recognition applications. Based on the rates posted by Colocation America – streaming data from 20 security cameras with recording at a generic five megabits per second (Mbps) would take 100Mbps bandwidth and rack up approximately $36,000 in bandwidth fees over five years. That gives us a good idea how cost prohibitive it is to transport data.
Let’s also go back to our example of managing the power grid. Even tiny amounts of sensor data add up quickly. A single smart meter report can generate 50 to 100 kilobits. If you pinged each household meter every minute, you would generate 110 petabytes a year – and that’s just the snapshot of consumption in homes. A single commercial building might generate 100 gigabytes of data a year just from its air conditioning and heater systems alone.
Autonomous vehicles, or self-driving cars, are another good example of the need for both local and cloud-based storage. They will potentially have to perform several multiplication operations per watt, and they will have to perform these calculations in real time so that the car can differentiate a paper bag from an obstacle like a rock. Self-driving cars will reach out to the cloud for traffic information, but navigation and much of the machine learning that improves its performance will need to happen locally under the hood and in real time in order to support the vehicle’s mission-critical systems.
Life on the Edge
So what will the IoT ecosystem look like? You’ll one day soon have billions if not trillions of sensors gathering data and performing none to a moderate level of processing: IDC predicts that 152,000 devices will link onto the Internet every minute in 2025. These will then link to IoT gateways that will manage internet connectivity, security and, to some degree, data analysis and storage. Raw data will stay local while anomalies and summaries will go to data centres.
Edge data centres, which replicate the bulk of the Internet in local communities so consumers aren’t retrieving Web pages from across the country or even across the world, will play a similar role on a regional level, reducing data bottlenecks by cutting the need for data to travel long distances.
The Size of the Problem
It appears likely that the vast majority of that data will be “transient” or status data, i.e. routine recordings of temperature or pressure from sensors, or video of pedestrian traffic from street cameras that we don’t need to save, but we need to create actionable intelligence from it. A huge portion of this data, will be mind-numbingly repetitive. But this granular information is raw material of Big Data and, in it, may lie potential for exciting new possibilities. Mining the up-down transportation details of elevator worldwide helps companies see if there are patterns that can help architects or energy planners. Others want to leverage data from motion sensors plugged into LED lights to lower traffic jams and congestion.
“Transient” data is just data no one has figured out how to use yet.
Competing in the Digital Gold Rush
The fact is that IoT appears to be newest “gold” and the “rush” is on to capitalise on it. While many companies have already developed a product strategy for IoT, others may have to completely rethink how to compete in this market.
For example, a company producing farm equipment like a tractor might innovate by building a “smart connected tractor”, but might miss the point that it ends up competing with companies who are able to pull a holistic “farm automation solution” together that connects multiple systems such as tillers, irrigation systems and seed optimisation systems together with advanced cloud based databases such as weather forecast apps connected to irrigation and seed databases, creating a farm management system.
Companies might find new ways to monetise data collected and analysed in this fashion. Think about it. What if a company not only sells equipment at a discount, but bundles it with some predictive analytics software subscription and charges based on a final target crop yield?
Companies may soon realise that the one who offers a “system of systems” solution with a very strong ecosystem tied to it will define and shape the competitive landscape.
This article expresses the views of the author and not necessarily that of his employer.
René Hartner, VP of Corporate Business Development at SanDisk, a Western Digital Brand