Using Neural Networks for switch failure correction in frequency reconfigurable antenna arrays

This paper presents a new technique to correct switch failures in a frequency reconfigurable antenna array. The correction procedure requires at the beginning a detection of the failed switch. The detection process relies on integrating sensing lines within the array substrate to monitor the switch failure. The proposed correction technique is based on using Neural Network (NN) to overcome the failed switch. This is achieved by making the trained NN searches for other switch combinations that give the same array behavior.