ABSTRACT
In Chapters 5 and 6, although various extensions to ideal estimation fusion models have been investigated, the model knowledge required for the estimation and fusion can be always known exactly, and the MSE is still an appropriate optimization criterion. If the uncertainty of model knowledge is ignored, many conventinal estimation results given precise model knowledge may lead to poor performance (see, for instance, Ben-Tal et al. (2009)). However, in many practical applications, the model knowledge may not be known exactly. These facts motivate us to further research the estimation fusion with the uncertainty of model knowledge, i.e., only some set information of the model knowledge can be provided. Thus, we call the estimation fusion with such uncertain model knowledge as robust estimation fusion.
