Skip to Main Content
Performance assessment (PA) is an important component of a cognitive radar system. In addition to allowing evaluation of candidate radar operating parameters and signal processing algorithm sequences on a near real-time or synoptic basis, it can also allow determination of the performance envelope of the radar system by defining the surveillance area over which the system is expected to detect targets. The use of operational PA metrics for over-the-horizon (OTH) radar is typically limited by the availability of truth data, real targets, and real-time signal processing capacity. This paper presents a computationally efficient PA module suitable for real-time operational cognitive radar use. The PA module uses synthetic targets to calculate detection and false alarm probabilities, which are used to estimate a novel “probability difference” (PD) metric. The PD metric provides several advantages over receiver operating characteristic (ROC) derived metrics, and allows the ability to explicitly incorporate target location accuracy. The PA module is demonstrated in the context of near real-time selection of adaptive processing algorithm appropriate for the prevailing environmental conditions.