Adaptive microwave photonic angle-of-arrival estimation based on BiGRU-CNN [Invited]

An adaptive microwave photonic angle-of-arrival (AOA) estimation approach based on a convolutional neural network with bidirectional gated recurrent unit (BiGRU-CNN) is proposed and demonstrated. Compared the previously reported AOA methods phase-to-power mapping, method unnecessary to know frequency of signal under test (SUT) in advance. The envelope voltage correlation matrix obtained from dual-drive Mach–Zehnder modulator (N-DDMZM, N > 2) optical interferometer arrays first, then estimations are performed different signals aid BiGRU-CNN. A three-DDMZM-based experiment carried out assess performance at three frequencies, mean absolute error only 0.1545°.