Performance adaptive resource allocation in a time-varying environment

This paper presents a performance adaptive algorithm for the allocation of information gathering (probing) resources and action resources in a time-varying environment. A suboptimal procedure that preserves the essential features of stochastic dynamic programming is developed. The information gathering resources are allocated using the prior probability density function of the subsequent observations in order to evaluate the value of future information. This value is measured in terms of the improvement in the performance index when the action resources are allocated subsequently with more accurate, even though still imperfect, information.