Some principles for developing new data fusion systems

Data fusion systems are being proposed for a wide variety of applications around the world. The desire to obtain the maximum benefit from expensive sensors and information systems has prompted this interest. By applying data fusion methods, more efficient and cost effective systems are hoped to be developed. Various testbeds are being developed to evaluate methods at levels of data abstraction and meaning. Of these testbeds and reviews of existing systems can help in design of new ones. This paper attempts to assist developers of new data fusion systems in a general manner by presenting a few principles that the author and coworkers have gleaned from study and practical experience. The principles are thus subject to debate and scrutiny of time and may require changes and further clarification in the future. These principles are: fuse the people first, know your data sources, know your environment, choosing your data sources, how many sources, use expert knowledge, what algorithms, use levels of abstraction and feedback. Expansion of these principles and their justification are given