Vector parameter estimation in FMCW radar

Today's automotive radars are based on FMCW signaling. The estimation of range, speed, and direction of arrival (DoA) for FMCW radars can be mapped to frequency estimation problems. Current methods rely on independent scalar frequency estimation techniques in each of the range, Doppler and DoA dimensions. In this work we develop a vector parameter estimation method generalizing existing scalar estimation methods, thereby allowing joint estimation of any combination of range, Doppler, or DoA. We apply this framework to FMCW radar and derive the required model vectors. In simulation, we show that the multidimensional approach suffers no loss of estimation efficiency compared to the existing scalar method in degenerate cases, while providing the novel capability of describing targets that cannot be captured by a single frequency parameter. We provide a set of model vectors for point targets with constant velocity and a general method for finding model vectors for targets whose position is modeled by an arbitrary power series over time, as well as a method for extending a set of model vectors to include direction of arrival.