Localization ROC Analysis for Multiband Spectrum Sensing in Cognitive Radios

Recent attention in cognitive radio applications has created interest in developing strategies for detecting and localizing primary users in order for secondary users to utilize the empty spectrum bands. The main contribution of this paper is to define optimal decision strategies for multiband spectrum localization and use several localization receiver operating characteristic (LROC) curves to evaluate performance of these decision strategies. The decision strategies investigated are motivated by the lesion detection problem in medical imaging, and include optimality criteria such as maximizing the area under the curve (AUC) of the LROC and Free Response ROC (FROC) curves. Performance of these optimal strategies are compared with well-known decision strategies, including maximum energy detection.