Redundancy in Multi-source Information and Its Impact on Uncertainty

This paper explores the relationship between the uncertainty of information (UoI) and information entropy as applied to multiple-source data fusion (MSDF). Many MSDF methods maximize system-wide entropy by minimizing source-data redundancy. However, the potential for uncertainty in the system provides a role for redundancy to confirm or validate the sensory inputs. While the relationship between uncertainty and entropy is neither wholly dependent nor independent, it is sufficiently complex to require modeling for each MSDF system. A one-dimensional model of redundancy versus entropy will not suffice when considering the UoI. The concept of UoI includes identifying the category associated with uncertainty. Thus, considering the redundancy within one category in relation to multiple categories may further reduce overall uncertainty. This paper proposes using utility functions, as well as a two-dimensional model with certainty on the x-axis and entropy on the y-axis, as tools for optimizing redundancy to benefit certainty and entropy maximally.