Rank-Order Adaptive CFAR: Performance Bounds and Efficient Implementation

The uniformly most powerful (UMP) rank-order analog to the cell-averaging constant false alarm rate (CA-CFAR) detector is developed for both the single and multi-scan case. The detector is capable of false alarm control in arbitrary noise while achieving maximum detection probability for the case of a Swerling II target in Gaussian noise. Relationships with earlier tests are discussed. A divide-and-conquer method is developed that allows for rapid implementation of a large class of two-sample rank tests.