This paper presents a decision support analysis framework for selecting screening devices, theirsensitivity parameters and detection thresholds in a two-device system for automated, non5intrusive inspection of checked baggage at airports. The analysis procedure assumes that theoutput signals of a screening device, when presented with a baggage with or without securitythreat, may be modeled by probability distributions. Compared with the pre-specified thresholds,the probability of making wrong decisions (false clear and false alarm) for the baggage may thenbe estimated. An objective function is then formulated to minimize the total expected cost ofscreening errors. Based on this framework, two decision criteria, one based on the minimumtotal expected cost and another one incorporates the operating characteristic curve of the system,are recommended. The application of the analysis framework is demonstrated with ahypothetical problem based on published data.
展开▼