As someone who is in the business of consulting manufacturing companies, over the past 10 years I’ve noticed an increasing number of clients who are placing a greater reliance on operations data to drive their business on analytics. While their primary business is making products—which are becoming increasingly complex—they are becoming skilled in the use of analytics to improve their products’ safety and reliability, cost-effectiveness, performance, and end customer experience. They are striving to become analytics-based competitors.
Along comes the Internet of Things (IoT). As products become increasingly sophisticated as computing devices in their own right, so does the data they produce—the connected car, the connected wind turbine, the connected train, the connected oil rig—each producing accelerating amounts of data.
Where to begin
To talk about the Analytics of Things (AoT), let’s begin with a very simple diagram of the Internet of Things:
The Operations subsystem encompasses the acquisition of data by sensors in connected devices, the transmission of that data to a decision-making engine, and the application of previously deployed or embedded software/algorithms on that data to enable an action—which may be automated or semi-automated with a human operator.
All this activity takes place at or near the “edge”; that is, where the “doing” takes place in IoT. You can think of a car and its driver as the Operation of Things subsystem in the Connected Car flavor of IoT. The car has a vast amount of intelligence built into its Electronic Control Unit, and various other subsystems, which—augmented by a skilled driver—acts in perfect coordination.
If you believe thinking should come before acting, then the notion that Analytics of Things is where the “thinking” happens will make sense. The Analytics of Things subsystem of IoT is an extension of the company’s analytic ecosystem that includes:
- Acquisition and management of sensor and other third party data
- Digital device models that mirror, predict, and prognosticate the behavior of things
- Ability to provide intelligent feedback and simple analytics (or rules) to edge devices for improving actions
In the Operations subsystem, devices are operated by local embedded software, possessing an ability and intelligence to act somewhat autonomously. However, the Analytics of Things ecosystem includes business context data that is unavailable at the Edge because it is an extension of the company's existing business analytics.
By using all available and relevant data, the algorithms deployed to execute actions at the Edge can be improved and updated, with the goal of improving device performance in the field. When companies bring all data into the analytic ecosystem, it can be leveraged to provide keen insights that would otherwise be impossible to make in isolation at the Edge.
Think of a vehicle health monitoring system, which contains predictive models that integrate vehicle build and quality information, maintenance records, and location-specific data. Together, it can be used to predict what may occur in the vehicle, and even how to respond through driver-assisted instruction or guided action by the car itself.
The specific implementation of, and inter-connection between, these two systems may vary greatly in design among different industries and deployments. Architecture choices depend on factors such as analytic capability of devices at the Edge, extent of connectivity, need for real-time response, security and safety requirements, as well as other considerations.
This blog is the first in a series on the Analytics of Things. Please look for upcoming topics to include, “Are You Ready for Analytics of Things?” and “Defining an Analytics of Things Capabilities Roadmap”.
