$\ell_{p}$-Norm Based Capon Filter for Robust DOA Estimation
Yi Yang, Yuan Chen · 2024
The estimation of the Direction of Arrival (DOA) plays a pivotal role in the domain of intelligent vehicular systems, serving to enhance situational awareness, augment safety measures, refine sensor optimization, and elevate the overall driving experience for both passengers and road users alike. Within this context, the Capon filter, also renowned as the minimum variance distortionless response technique, is prevalently adopted for the computation of DOA estimations. This is achieved through the minimization of the output power whilst concurrently preserving the integrity of the signal's response. Such a methodology exhibits commendable precision in pinpointing the origins of signals amidst environments characterized by Gaussian noise. Nonetheless, its efficacy wanes significantly under conditions dominated by heavy-tailed noise distributions, attributable to its diminished resilience against outliers. In light of this, the present study introduces an enhanced variant of the Capon filter, designated as the$\ell_{p}$-Capon filter. This innovative approach entails the substitution of the conventional output power with the$p$-th power of the output signal, specifically within the range of$1 < p < 2$. Empirical simulations delineated herein substantiate the$\ell_{p}$-Capon filter's superior outlier resistance capabilities, thereby underscoring its potential applicability in advancing DOA estimation methodologies.