The paper presents a system-level approach for the modeling and simulation of a genuine heterogeneous system composed of individually powered Wireless Sensor Network nodes. The models are written in SystemC-AMS, an open-source C++ extension to the OSCI SystemC Standard dedicated to the description of AMS designs containing digital, analog, RF hardware as well as physical, optical or chemical IPs. The paper is composed of two parts. The first part details the study case, a system of WSN nodes that can monitor a physical seismic perturbation, transmit information on this perturbation to other nodes by means of 2.4 GHz RF communication links, and finally compute the epicenter of the perturbation by asking the 32-bits processor embedded in a node to solve the system of nonlinear equations relative to the triangulation algorithm. Each node is powered by an autonomous kinetic battery model. The second part presents the corresponding implementation in SystemC and SystemC-AMS, and gives an insight on how all the disciplines are elegantly intertwined, with an optimalmodel of computation associated to each hardware component of the simulated system. This part proves that the joint use of the Timed-DataFlow (TDF)Model of Computation for AMS parts, RF baseband equivalent for RF parts, and Communicating Synchronous Finite State Machines (CSFSM) for digital parts significantly reduces the simulation time while keeping excellent accuracy and code readability. After some results, the paper concludes on the possibilities offered by this approach in terms of validation and optimization of heterogeneous systems using an open-source simulation framework.