Sensor data fusion in the context of electric vehicles charging stations using a Network-on-Chip

This paper presents a platform for evaluating sensor data fusion algorithms based on a Network-on-Chip (NoC). Initially the NoC is simulated by MPSoCSim, which contains an ARM and several MicroBlaze processors. The approach is furthermore evaluated on a real FPGA system in an automotive context. Charging stations for electric vehicles consist of expensive sensors, as the energy measurement system is an important part used for billing purposes, for safety of the charging process and for security. The sensors, that are usually used, are special made solutions, which are thermally stable and highly accurate. The NoC infrastructure presented in this paper results in a new method for sensor data fusion, which decreases costs by using inexpensive sensor units. These sensor units are sensitive to temperature changes. The target of the sensor data fusion approach is to replace expensive sensors with inexpensive ones without loosing accuracy. The presented simulation platform allows an easy exchange of sensor models and sensor data fusion algorithms.