Direction Finding for Multiple Wideband Chirp Signal Sources using Blind Signal Separation and Matched Filtering

This paper proposes a new scheme to estimate the direction-of-arrival (DOA) of multiple wideband chirp signals. The proposed scheme comprises three steps: 1) preprocessing of the mixed signal using independent component analysis in order to obtain each wideband chirp signal; 2) matched filtering in order to transform the multiple wideband chirp signals into narrowband signals; and 3) the estimation of final DOAs using the conventional subspace-based method. In order to validate the effectiveness of the proposed scheme, we carried out extensive experiments, including three different scenarios: non-overlapped in the frequency domain, overlapped in the frequency domain, and nonlinear chirp signal cases. From the experimental results, we can conclude that the proposed scheme provides not only accurate results in terms of root-mean-squared error (RMSE), but also the lowest computational complexity compared to the previous methods.