Adaptive Compressive Sensing Based Direction of Arrival Estimation Using Particle Filters

The signals in the direction-of-arrival (DOA) estimation problems lie in a relatively few dimensional manifold in their ambient space. Therefore, compressed sensing techniques enable reliable estimations. Furthermore, using sequential Monte Carlo methods enable to obtain a probability distribution for DOA estimation instead of a single point estimate. As a consequence, the probability distributions, when used in measurement matrix design, enable dimension reduction for sensor array signal processing together with high estimation performance. In this study, we use the Particle Filters to obtain the estimation distribution, and we proposed a gridless DOA estimation method within an adaptive compressive sensing framework using this distribution. We compared the performances of random and designed measurement matrices. We demonstrated an estimation performance increase via a set of simulations depending on the measurement noise.