This paper describes the use of cognitive radar networks for the reliable surveillance of an environment where the likelihood of the presence of uncooperative targets is high. Despite the limited effectiveness of amplitude as a radar discriminant, through the clever use of detection through tracking, the detection performance of the noncoherent radar is enhanced. Further improvement in overall performance can be attained through the adaptive illumination of the environment, made possible by linking the receiver to the transmitter via a feedback channel, which, in turn, makes the tracking adaptive, and sensor fusion performed in the central base station