MIMO Radar Moving Target Detection in Clutter Using Supervised Learning

To detect a potential moving target in presence of clutter, the statistical method such as the generalized likelihood ratio test (GLRT) may be computationally expensive, due to unknown target and clutter parameters. This paper exploits the potential of supervised learning, which can have the computational complexity under control, for multiple-input-multiple-output (MIMO) radar target detection in combating clutter. Compound Gaussian clutter with unknown parameters is considered for generating synthetic data. The measurements at all receivers are fed to a fully connected network (FCN) at the fusion center, where detection is viewed as a binary classification problem. The detection performance of the FCN-based detector is analyzed for scenarios with various MIMO radar settings under different clutter parameters. It is shown that the FCN-based detector can outperform the traditional GLRT detector in the tested cases.