Interpolated root-Music for transversal automotive radar traffic monitoring

One of the key requirements for future automotive radar safety systems is to have a wide field of view along with high angular resolution and accuracy. The combination of a conformal array and a root-MUSIC algorithm can provide these properties but root-MUSIC requires an uniform linear array (ULA) structure and high SNR in order to work. In this paper the ULA requirement is handled by transforming the real array data into data from a set of virtual linear arrays by a linear interpolation procedure. High SNR is achieved by first detecting the targets and then use the Fourier transformed signal at the target Doppler frequencies for the angular estimation problem. The system performance is analysed by implementing an advanced simulation model with the radar targets moving in a transversal trajectory testing the system at the extreme angles near ±90°. The tracking performance is finally evaluated by implementing an extended Kalman filter. The results show that by due to the high SNR the angular estimation works properly even though we have a small number of signal snapshots, also target ambiguities are resolved and the estimated trajectories are accurate enough to find which lanes the targets are moving in.