Analytics from the sensor to the big picture

Some IoT systems are great at showing information for a specific "thing", but how do you see an overview of hundreds, or thousands of such things?

Imagine, for example, you are monitoring water quality in a commercial building. You have devices sending data about taps, waste units, filters and so on. Your IoT platform receives the data and stores it.

Having a graph or event log for a specific asset is necessary but seldom helpful. What is needed is a way of grouping that information together.

image hierarchy of informationA hierarchy of information (rather than nitty-bitty data)

To follow the above example, you may want to group together the data from assets in a given room (e.g. a production zone, a kitchen, a washroom), so that you can see that room in some summary form. You might expect to see a graph of performance for that room, with any quality issues identified.

Going further, you may want to see information about a whole floor or area of a building, again with a useful view of what you need to know. And logically a multi-site organisation would want to see what's going on for a whole site, or location.

image data poolsData pools

In an AssetWolf portal, we have a system called "data pools", which does exactly this.

Data pools are arranged in a hierarchy according to how you classify data about your assets. They allow data to cascade upwards from individual assets to fleet-wide summaries, with complex calculations possible.

Amazingly perhaps, we do this in real time, and it's fast.

Complex calculations

Data pools don't just do mundane adding-up of numbers. They can perform complex calculations — which are transparent and can be defined by you — so that statistical and scientific calculations can be run on the fly.

Whenever data arrives at the portal for an asset, a pre-determined set of calculations can be run. This uses our purpose-built Phi scripting language

The calculations are then handed up to the data pool containing that asset, and that data pool — like a "room" which contains the assets — then also executes its own set of pre-determined calculations.

(That's a bit of a simplification! There's a little more to it than that. There are low-priority fields which don't get processed immediately, and high-priority fields which do. But let's keep going...)

Once the room data pool has run its calculations, it hands its output to the next thing above it in the hierarchy (such as an area, a floor, or a location).

In this way, AssetWolf creates a high-level view of information very nearly in real time. You can determine everything about the configuration: the fields handled, the calculations performed (if any), and what the data pools represent (there's no need for them to relate to parts of a building like in the above example).