Improving the performance of adaptive arrays in nonstationary environments through data-adaptive training

Adaptive array algorithms based on sample matrix inversion require the availability of a secondary data set to "train" the adaptive filter. Numerous data-independent rules have been proposed for selecting this training data. However, such rules often perform poorly in highly nonstationary environments. In this paper, we present data-adaptive techniques for selecting the training data. The techniques, called power selected training and power selected de-emphasis, use measurements of the interference environment to select training data. This paper describes the algorithms, as well as optimality, complexity, and performance on recorded radar data.