Direction-of-Arrival Estimation of Ultra-Wideband Signals in Narrowband Interference Environment Based on Power Inversion and Complex-Valued Neural Networks

We propose two-stage null-steering direction-of-arrival (DoA) estimation of ultra wideband (UWB) signals with power inversion algorithm and complex spatio-temporal neural network (CVSTNN). This method can estimate DoA more accurately than conventional methods in narrowband interference (NBI) environment. For null steering in UWB systems, it is necessary to adjust the amplitude and phase of tapped delay lines (TDLs) of CVSTNN. However, with a conventional CVSTNN, it often fails to estimate the arrival direction because of the NBI. We aim to reduce the influence of NBI in the learning process to avoid falling into a local solution by setting the initial weights of the TDLs with power inversion. In simulation results, it is shown that the two-stage method can realize higher DoA estimation accuracy than conventional methods.