A radar using a ULA with a large number of elements(for instance N=150 for 1[deg] beamwidth[1]) is being designed. After the Range and Doppler processing, Digital Beamforing is being performed, yielding a Range-Doppler-Angle cube. The following stages originally should be – CFAR-->Clustering-->Angle Super Resolution. Taking into account the computational burden, wondered whether there are any alternative to applying CFAR over each of the 150 planes of the Cube prior to clustering stage?

[1]https://electronics.stackexchange.com/questions/288530/uniform-linear-array-ula-beamwidth-and-angular-resolution-using-fft