Robust Orthogonal Matching Pursuit Detector : Applications to off-grid target issues in MIMO Radar

The paper proposes to solve the off-grid issue arising in sparse processing involving Orthogonal Matching Pursuit (OMP) algorithm. It starts by establishing the Generalized Likelihood Ratio Test using a stochastic model to represent target location uncertainty within one resolution cell. In a second step, a linear approximate of the detector is built, requiring only a grid oversampling of 2 relative to the resolution cell size. Moreover, it involves only the formation of standard delta channels in each dimension, leading to low computational burden without loss of efficiency. The approach is here demonstrated on MIMO radar detection.