Statistical Approach for Automotive Radar Self-Diagnostics

In this paper, the problem of on-the-fly estimation of the radar state (self-diagnostics) is considered. We propose to use repetitive objects of the road infrastructure, such as lampposts, for continuous diagnostics of the radar state. The selected approach allows accounting for the external factors, such as water layer or dirt on the bumper, which can significantly affect radar performance, but cannot be retrieved with the internal calibration. The statistical model for RCS of repetitive targets is considered, and the estimator of the actual radar gain from the received data is derived. It is demonstrated that observing a few tens of targets is sufficient to provide a reasonable estimation of the radar performance within the operational mode.