Many recent research works deal with wireless sensor networks and their various problems. One of them is the management of WSN data. Generally, the two main types of handled data are either those collected using a sensor that measures a physical quantity, or those generated and managed by the operating system. The purpose of this thesis is to propose solutions to this problem. A micro-file system has then been designed from a hybrid real-time kernel which is both a native event-driven and multithreading operating system. This kernel implements a concept that offers an abstraction level for the management and the scheduling of process and events. This concept was adapted, in the micro-file system, in order to achieve data access. The other main characteristic of this micro-file system, in comparison with existing systems, is to merge, within the same system, functionalities about memory management and data interrogation. These two elements, namely the micro-file system and the real-time kernel, associated with a multi-component wireless sensor provide an adaptive platform dedicated to environmental data collection applications.
