Thing's data pre-processing functional component provides several functions at Things and platform level which are summarized below:

1) It provides the capability for basic pre-processing on raw data streams including filtering, selecting or aggregation on real-time data using light weight Complex Event Processing running on the Thing;
2) It provides the capability to run pre-processing on the object storage using storlets (http://iot-cosmos.eu/node/2151). Storlets can be both generic and domain specific as well. For domain specific storlets, an application developer can write his own storlets which can be run using restful web interface;
3) It provides the functionality to run different pre-processing techniques including aggregation, interpolation, and data cleaning and feature scaling using Apache Spark on the platform level.

More information can be found in Deliverable D6.1.2
http://iot-cosmos.eu/content/d612-reliable-and-smart-network-things-desi...