This PhD thesis concerns the dynamic reconfiguration and simulation of heterogeneous Wireless Sensor Networks. These networks consist of a multitude of electronic units called ?nodes ?, which communicate through a radio interface. The radio interface is the most power-consuming on the node. This is why the minimisation of the radio-time would lead to improved energy efficiency. We have implemented a software solution which consists in sending an update to a node which is coded in a high-level language (MinTax). This file is compiled by the node and machine code is generated for the target hardware architecture. Owing to the abstract nature of MinTax, multiple hardware architectures. as well as operating systems are supported. As a second part of this PhD, work has been focused on a network simulator called IDEATLM.IDEA1TLM allows us to predict which circuits and configurations are the most appropriate for a given task. This solution has been improved to allow a faster simulation of electronic systems which are different from a hardware standpoint, yet part of the same network, as well as to model the actual software running on them.
