Currently working on a FMCW MIMO radar RX pipe consisting of the following modules:
1. 2D CFAR – processing the Range Doppler images per Angle.
2. 3D clustering – clustering the 3D cube generated from the 2D CFAR outputs.
3. DOA(super resolution).
Wondered what are the cons and pros of changing the order between the DOA and the 3D clustering modules.
In case of Cluster-->DOA - fewer targets will burden the DOA(large computation effort), the angle resolution for the clustering will be however lower. As a result, several targets can be clustered into a single one.
In case of DOA-->Cluster, targets angle separation is better, however, DOA related computation cost is high.
Taking into account that the EKF will evantually seperaate the targets - seems that the Cluster-->DOA order is preferable.
It appears that you have already answered the question, there will be some trade off.
Perhaps, you can quantify the trade off DOA resolution angle vs. computation time.
In other words, how much is saved if the architecture is changed?
Then, the systems engineer or product owner should be able to determine if the trade off is meeting product requirements.