Building a platform to collect crowdsensing data. Preliminary considerations

Recent years have seen growing interest in collecting and processing sensor data, in distributed mobile environments. In this context, two, somewhat contradictory, trends have emerged: (1) growing popularity of crowdsourcing-type mechanisms, for (sensor) data collection, and (2) collecting sensed data in data “silos”, which are not only unavailable to “outsiders”, but most often incompatible, thus reducing their usability for data mining. Given these limitations in data accessibility, and compatibility, enormous potential for knowledge discovery is lost. To counter this trend, we propose a generic, adaptive, system that will allow voluntary participation in arbitrary crowdsensing initiatives, with the output stored in a standard data format. The system utilizes a rule-based multiagent approach to instructing sensors when to make readings and how to, if necessary, preprocess them, before sharing the data with user-selected initiatives. The initial version of the system has been implemented, and tested in a...