Abstract
This letter investigates the problem of detecting a point-like target embedded in Gaussian environment for multiple input multiple output (MIMO) radar, for the case where there are not assumed the presence of training data. Specifically, consider a specifical case where MIMO radar dwell is a short duration so that we can’t collect enough snapshots of test data. In this respect, a positive covariance matrix can’t be estimated by the existed receive data. To end it, this paper proposes a reduced-dimension approach via pre-multiplying the recorded data with the receive steering vector. Based on the transformed data model, one new detector has been designed in accordance with the criterion of generalized likelihood ratio test (GLRT), namely RD-GLRT. Further, this paper proves that RD-GLRT exhibits the constant false alarm rate (CFAR) property against covariance matrix. Numerical simulation results validate the effectiveness of proposed theory.

This publication has 20 references indexed in Scilit: