Support Vector Machines Based Mutual Interference Mitigation for Millimeter-Wave Radars

With the intelligent development of vehicles, the number of vehicles equipped with millimeter-wave (mmWave) radars is increasing, and the possibility of interference between radars is rising dramatically. In automatic driving, it will be common for target detection to be affected by multiple interfering radars. Addressing the mutual interference challenges, an adaptive interference detection method based on support vector machines (SVMs) is proposed. First, a window selection is performed on the received signal and features describing the difference between the normal signal and the interference are extracted. Then, we use a nonlinear SVM to distinguish between the interference and the normal signal. After completing the localization of the interference, we use an autoregressive (AR) prediction model to reconstruct the target echo signal. Results from both multiple interference simulation scenarios and real experimental scenarios show that the accuracy of interference localization and the effect of interference mitigation of the proposed method outperforms the mainstream methods.