Some commentators feel that the Internet of Things is a network of items that exchange very basic data without much intelligence being involved. They point to a lack of analysis that could interpret this data, and do something meaningful with it. However, the Internet of Things endpoints are becoming more intelligent. This is having a marked effect on data architecture, which can be simplified if data doesn’t need to be transferred in order to be analysed. When connected things analyse themselves and their own environment, they also become more responsive.
Smart cities using data collecting endpointsIn urban environments, transport objects are increasingly able to carry out elements of analytics, at least for issues that are relevant to them, without reference to a central processing hub. But to what extent can these objects learn, then develop rules, and carry out appropriately intelligent actions?
Well, to some extent, is the answer. But machine learning uses a lot of computational power, so it makes sense for the end point IoT objects to provide some processing, then pass the rest to a back end. This can run analytics on the data combined from many end points, then feed actions, rules and updated information back to the endpoints as appropriate.
Industry leading the learningHeavy industry has been a leader in connecting sensors on physical equipment. Take an oil refinery - it’s full of equipment that measures flow rates, temperatures, vessel thickness, and large numbers of other variables that flag the need for maintenance, or for control operator intervention.
In a refinery with a complex infrastructure, no single person may have a complete overview of the way the system works. This is a classic case where machine learning, the basis of artificial intelligence, can provide a level of oversight, and error prediction, that no human can manage. However, the learning is refinery-wide and not necessarily associated with individual endpoints such as single equipment types.
But once the data is being collected, and compared with real life events, machine learning can maintain models of likely outcomes, and produce forecasts. The more limited data being collected on the individual endpoints,can be processed fast enough to provide very tight control, since it doesn’t need to be transmitted to a network, and processed centrally.
Running analytics on data being collected by connected endpoints could revolutionise every kind of urban transport - buses, cars, trains and drones. Many of our IoT transport systems currently sense and report - they don’t predict and forecast, which would make urban and motorway traffic management, much more intelligent.
One of the other environments that currently collects reams of data but is short of meaningful analytic intelligence, is health services. The endpoint is frequently still manual - the chart at the end of the bed. Gradually, this is changing, but when the IoT begins to pass individual patient data to hospital and area-wide intelligent systems, our healthcare systems will become much more efficient.
Maybe you'll find this white paper interesting: