Simulation-based development of embedded sensor fusion applications

This paper proposes a methodology for the design, testing, and fine-tuning of sensor fusion applications for distributed embedded sensor networks. The presented approach involves the following steps: (i) gathering data about the process environment, (ii) modeling of the process and the sensors to be used, (iii) selection of sensor fusion functions to process the sensor measurements, (iv) simulated closed-loop hardware-in-the-loop testing, (v) validation by open-loop testing in the real process, and, finally, (vi) application of the developed embedded sensor fusion application in the system. Particular steps may be repeated in a loop, if the result of a given step is not satisfactory. The main contribution of the approach is a separation of the modeling of process environment, sensors, and sensor processing. This allows for a strategic refinement of the model and supports the reuse and change of parts, e. g., in case of a sensor upgrade. The application of the approach is shown by a MATLAB/Simulink model that incorporates block diagrams describing process environment, sensor behavior, and sensor fusion algorithms