Spotting Faults Over the Spectrum: Fast and Online Antenna Array Fault Diagnosis for Multi-Carrier Precoding Systems

We present a new online method to diagnose faulty elements in an antenna array quickly and reliably while a wireless communication system is under regular operation. Existing methods for fault diagnosis do not usually allow simultaneous operation and diagnosis, leading to downtime, which becomes a severe constraint for modern wireless systems. Our approach is based on exploiting an extra degree of freedom available in the frequency domain in modern multicarrier hybrid/digital precoding architectures by allocating a very small part of the available spectrum for diagnosis while the rest is used for regular operation. Measurements can be made in a single shot by using optimized inputs, thereby minimally disrupting regular operation. We address both fully and partially connected RF precoding architectures of hybrid precoders and introduce one and two-stage fault diagnosis procedures, respectively. While the faulty antennas for fully connected hybrid architectures or digital architectures are diagnosed employing ideas from sparse recovery, additional ideas from block-sparse recovery are used in the case of a partially connected architecture. Our results indicate that online array fault diagnosis is possible in both digital and hybrid precoding architectures by introducing simple digital operations into the system and by using a single fixed receiver measurement setup.