Bayesian Sensor Fusion for Minimum-Cost I.D. Declaration
Abstract
Current operational procedures for I.D. sensor fusion tend to be rigid, stereotypical, and based on unvalidated assumptions about the performance of specific I.D. discriminant sources, such as Q&A IFF, NCTR, etc. To be optimum, any I.D. sensor fusion procedure must be explicitly sensitive to several factors. These include: (1) the objective performance of the specific sensors in use, (2) the costs of making different kinds of I.D. declaration error, and (3) the friend/hostile force ratio at the specific time and place of CID declaration. In this paper we describe an analytic methodology for developing and evaluating an I.D. sensor fusion procedure that is sensitive to these factors and which can be shown to yield the lowest risk I.D. declarations, in the sense of having the lowest expected cost due to errors of all kinds. This approach to fusion analysis allows the performance of individual sensors to be reflected directly in CID effectiveness and allows the connection between individual sensor performance and the overall system performance requirement to be traced. By permitting direct visibility of and access to the often "hidden" assumptions about force ratio and cost of error, it can promote the development of fusion rules suited to the joint operations and coalition forces environment. Operationally, it would permit I.D. declaration rules to be quickly adapted to the threat, forces, and sensor performance appropriate to a particular theater. It is possible that procedures based on this analysis could be applied within existing and developmental computer-based command and decision systems.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1999
- Accession Number
- ADA371089
Entities
People
- James M. Ralston
Organizations
- Institute for Defense Analyses