Effects of Known and Unknown Antenna Position Errors on MVDR

We assess the impact of known and unknown position errors on two variations of the Minimum Variance Distortionless Response (MVDR) algorithm specifically for measurements using multipass, swarms or constellations that involve non-rigid and varying sensor offsets. We consider this in the context of the radar systems and autonomous aircraft used by Center for the Remote Sensing of Ice Sheets (CReSIS) for conducting measurements of snow and ice in polar regions. We find that unknown position errors affect all algorithms in a uniform way and that the performance impact is not related to known position errors. We also find that MVDR is largely unaffected by known position errors except when the number of elements is small. However, self-nulling can result in substantially worse performance than even the simple nonadaptive periodogram method.