Akademik Veri Yönetim Sistemi
32nd European Signal Processing Conference (EUSIPCO), Lyon, Fransa, 26 - 30 Ağustos 2024, ss.2712-2716
We consider the problem of parameter estimation under constraints on error probabilities and uncertainty regarding the prior probabilities of the hypotheses. By combining the average estimation cost with respect to some guessed a priori distribution with the maximum estimation cost, the joint optimal estimators and detector that minimize the resulting estimation performance metric subject to individual constraints on miss detection and false alarm probabilities are derived. Based on the belief in the prior estimate, the proposed framework provides the flexibility to strike any desired balance between the performance of the average estimation cost based scheme, which is employed in the presence of perfect prior information, and that of the minimax scheme, which is employed in the absence of any prior information. Numerical results indicate that the proposed method yields lower estimation cost compared to the classical method that treats detection and estimation separately.