DOA Estimation Based on Gridless Fuzzy Active Learning Under Unknown Mutual Coupling and Nonuniform Noise: Experimental Verification

In this article, we proposed an active learning method for the direction-of-arrival (DOA) estimation based on uniform linear array (ULA) antennas with very low computational complexity. In particular, an adaptive fuzzy sampling strategy was used to make a model and refine it according to the snapshot data collected by multiple antennas to estimate the DOA angles. The proposed gridless fuzzy pipeline sequential design zero-forcing (PLSD-ZF) sampling can reduce complexity by dividing the angle space into subsets and using the measurements of the appropriate subsets to obtain more accurate DOA estimates. Our simulation and experiment results demonstrate that the proposed fuzzy PLSD-ZF scheme outperforms other well-known ones in the presence of unknown mutual coupling (MC), nonuniform noise (NN), and data with a mixture distribution (MIX).