Discrimination of Transmitters in Equally-Spaced DDMA MIMO Radar Using a Correlation Feature and Neural Networks

In this paper, we explore the use of neural networks in the discrimination of radar transmitters in frequency-modulated continuous wave (FMCW) multiple-input multiple-output (MIMO) radars. The transmitters are equally spaced in Doppler through the use of Doppler-division multiple access (DDMA). The transmitters are then separated during the signal processing to realize a wide virtual aperture. Although the equal Doppler spacing works well with single targets, this may produce partial or even full overlap when the difference in the Doppler shift of two targets is a multiple of the spacing. We separate the transmitters in both the single and the overlapping two-target scenarios by the use of a cross-correlation feature, implemented as a circular convolution operation, as an input to neural networks. Experimental results in a radar simulator and a traffic monitoring radar system show the effectiveness of the proposed transmitter discrimination.