Performance of Multi-Criterion Acquisition Sensor Systems
Robert T. Frankot · AIAA Guidance, Navigation, and Control Conference and Exhibit · 2008
This paper addresses the performance of sensor systems that automatically recognize, acquire, and commit resources to objects in accordance with an external cue. Such capability is of utility in remote sensing, surveillance, fire control, and missile systems. Object acquisition is posed as a task of selecting from among multiple detected objects based on the consistency of their sensed position, velocity, and signature with that of an external target cue. The probability of correct acquisition/selection is formulated as an order statistic of a joint discriminant that fuses these multiple features. Special cases are evaluated for several object selection rules to provide system level performance as a function of subsystem performance parameters such as position and velocity estimation accuracy, detection and false alarm rates, and automatic target recognition (ATR) performance for the acquisition sensor. The effect of deferred commitment (i.e. selection of the K≠1 best ranking candidates) is modeled. A performance estimation and target information gain/loss accounting approach is presented. System performance modeling is thereby reduced to simple spreadsheet calculations that include loss tables for non-ideal effects. Monte Carlo simulations verify that the superposition of gain and loss factors accurately approximates performance over a broad range of parameters.