SENSING AND PATTERN RECOGNITION

A self-organizing system needs one thing above all: feedback. In order to learn from previous experiences and to adapt to changes, a self-organizing entity -whether and organism, a brain or a human organization- needs to collect and process information from its surroundings.

We can enrich the collective brain of people and organizations by measuring all kinds of properties of the world, through individual sensors, smartphones and (wireless) sensor networks. We thus aim for true contextual awareness.

Interpreting and forecasting

The challenge is to develop ways of interpreting these sensor data, and to predict future states. Currently, sensors can measure many -quite sophisticated- states of the world, but intelligent pattern recognition is still lacking. Each sensor can register a small part of the world, but we need to combine information from several sensors (sensor fusion), in order to start to recognise more complex patterns and to forecast events or states based on that information.

A burglary alarm merely registers movement, but does not take into account nuances or secondary influences, such as the presence of the cat or a daughter who forgot the security code. Similarly, small portable biomedical sensors can collect a very valuable store of data, but these data are only of interest if combined with other data sources, such as diet, exercise and medication use.

Actuation

Research aims to create a dynamic filtering and escalation system within the network that triggers reaction when necessary. Actuation then can be done from within the self-organizing network itself, and does not have to be carried out by a central, external system or person.

Challenges:

> Different combinations of sensors lead to a different view. How do we determine our perspective, or allow the software to determine its own perspective?

> How can we interpret and recognize real-world events, based on collections of raw sensor data?

> Can we forecast future states, based on past patterns and current context information?

Keywords:

> sensing and actuation

> feedback

> distributed

> contextual awareness

> internet of things

> sensor fusion