Information Fusion System design space characterization by Design of Experiments

An Information Fusion System (IFS) is comprised of multiple independent and distributed systems which collaborate with one another while performing Low-Level Information Fusion (LLIF) and High-Level Information Fusion (HLIF) functions. The design variables of an IFS range from selecting algorithms and techniques that provide LLIF and HLIF functionality, to orchestrating the distribution and collaboration of independent systems (namely the IFS architecture) which provides the information fusion capability. Hence, the design space of an IFS constitutes an extensive set of heterogeneous design variables. It is imperative for the systems designers and decision makers to characterize this design space in order to evaluate implications of IFS design decisions. The objective of this paper is to introduce a structured methodology for characterizing the design space of an IFS, where characterization means both identifying the significant design variables and quantifying their impact. First, a design space taxonomy of IFS is developed which facilitates identification of various IFS design variables. Second, the quantification challenges that arise from the nature of heterogeneous design variables, the inevitable coupling of LLIF and HLIF functions, and the distribution of multiple systems are discussed. Finally, a design space quantification strategy based on Design of Experiments (DoE) is formulated and suitable DoE methods which address all the IFS design space quantification challenges are identified.