Multipixel, Multidimensional Laser Radar System Performance

The superb angular, range, and Doppler resolutions of coherent laser radars have led developers to design imaging radars in multiple measurement dimensions. Designing processors to detect targets in the images generally proceeds in an ad hoc fashion and it is difficult to predict the performance of the resulting processors. This paper proposes simplified statistical models for the target, radar, and signals then uses classical detection theory to derive quasi-optimal processors which take advantage of the multipixel, multidimensional nature of the image. The target model is of a radar looking down at a vertical target against a uniform, sloping background. The paper also presents the receiver operating characteristics (ROCs) for the resulting generalized likelihood ratio test (GLRT) processors. The receivers may use any combination of intensity, range, and Doppler measurements. The target reflectivity, range, and angular location are unknown and the background reflectivity is also unknown. The forms of the quasi-optimal receivers provide analytical confirmation of the principles used in many ad hoc processors. The ROCs not only give bounds on the performance of any ad hoc processors and prove the range-only processors are usually superior to the intensity-only processors, but go on to predict how much better and under what conditions. The ROCs also predict how performance changes as a function of resolution in one or several measurement dimensions.