Altruistic Approaches to Cooperative Estimation

摘要

We present a novel approach to cooperative estimation for two sub-systems possessing the same information. Instead of the two systemsultimately yielding an identical, joint estimate { here, each of the twosubsystems yields a dierent estimate than the other. The estimatorsare derived based on the underlying assumption that the global missionis accomplished if at least one of the estimators succeeds in providing asatisfactory estimate. A new notion of optimality is dened, that gener-alizes the common denition of minimum mean square error optimality.In the Gaussian case, explicit expressions for the optimal estimators arefound, that only require calculation of the largest eigenvalue of the con-ditional covariance matrix and its corresponding eigenvector. A simula-tion study demonstrates that the performance improvement, resultingfrom using the new approach, depends on the problem's dimension andstatistical distribution.