Fuzzy and Dempster-Shafer evidential reasoning fusion methods for deriving action from surveillance observations

Fuzzy reasoning and evidential reasoning using Dempster Shafer (D-S) rules of combination are compared in the fusion of identity attribute information to automatically monitor the proposition that a target is a friend or foe. The comparative results show the advantage of using the Dempster-Shafer methodology over fuzzy reasoning in this particular application. The simulated problem consists of 21 different scenarios of sequential information derived from various knowledge sources arriving at an intelligence centre at various times. The centre will track the target observation data digitally on a geographic raster map and monitor the likelihood of each target sequence as either a friend or foe. The sources of information may be derived from fired and variable locations in a 1000 square kilometre area which contains a potential enemy target such as a military base, vital assets (communication towers), parkland, scrubland and a road network