Phaseless Diagnosis and Pattern Correction of Faulty Antenna Arrays via Advanced Bayesian Compressive Sensing Approaches

This paper presents an integrated approach for diagnosing and correcting faults in antenna arrays using a Bayesian compressive sensing (BCS) method. The proposed diagnostic technique effectively identifies both ON-OFF and partial faults with limited phaseless measurement data. By linearizing the nonlinear inverse problem through a phaseless mapping method and employing a multitask BCS (MT-BCS) algorithm, the solution accounts for statistical correlations between the real and imaginary parts of sparse unknowns, ensuring robust diagnoses from highly coherent near-field measurements. Meanwhile, to address the detected faults effectively, a novel pattern correction method within an alternate projection framework is further developed to recover the pattern features with minimal corrections. This method features a modified forward projection rule to accelerate convergence and utilizes a BCS algorithm during backward projection to find sparse correction vectors. In addition, an innovative termination criterion is introduced to avoid trapping in local minima. Comprehensive numerical experiments demonstrate the effectiveness and efficiency of the proposed integrated approach in diagnosing various fault types and correcting radiation patterns. The results indicate that the method offers a promising solution for real-time online correction of large-scale antenna arrays.