Multi-sensor Information Fusion Methods For Integrated Identification Of Friend Or Foe Based On Evidential Networks

Identification of friend-or-foe (IFF) is to distinguish and validate the allegiance of a target by all available techniques and methods in the space-time area required by the war. The improvement of modern information technology promotes the fast development of new-style radars, visible-light sensor, infrared sensor and electronic support measures (ESM). These sensors enhance the ability of battlefield conductor to obtain information of target, and greatly enrich the information sources which can be used to identificate the allegiance of target. How to characterize legitimately and fusion effectively the uncertainty target allegiance information of all kinds of different levels of the sensor-output, and then give the identifiable result of the target allegiance consistently, is one of the core theoretical problems of the integrated identification of friend or foe (IIFF). Based on the characteristics of the typical IFF sources and the research on theories and methods of evidential networks, proposed a kind of approach which based on evidential networks'multi-source IFF information fusion, provided a useful information integrated fusion approach.First, introduced the JDL data fusion model, analysised the levels'features and the multi-structure characteristics. Indicated the location of IFF issues in this model. In connection with the features of typical cooperative and non-cooperative sensors ,and the external sources to obtain target information, analysised and introduced the ability of obtain target identification information or these typical sources. The identifying information comes from different sources, and with different types and multi-level natures, combined with JDL information fusion model, proposed a multi-source hierarchical IFF information fusion mechanism.Second, introduced the basic definition of the evidential network, the key elements, generalized structure model and parametric modeling processes; analysised deeply about the process of the evidential network modeling from the two sides of qualitative level: undirected graph and directed graph, which based joint belief function, conditional belief function as parameter model. Summarized the characteristics of the evidential network, indicated its'advantages in the level of uncertainty and hierarchical information fusion processing .On this basis, with the issue of IIFF, proposed and constructed two types of IFF evidential networks model about multi-source information fusion. First, proposed a multi-source IFF information evidential networks fusion method, which as non-directed graph to topology structure, as joint belief function to parameter model, and it realized the design of network's structure, and studied the reasoning algorithms and solution ideas that based extension and Marginalization as basic operation. In connection with the problem of combination explosion in the joint belief reasoning, explored the evidential networks reasoning method based on VBS theory, built a BJT structure diagram. Then, raised the evidential network modeling means, which as directed graph to topology structure, as conditional belief function to parameter model, analysised the forward reasoning and backward reasoning algorithm deeply. With the typical scenario for target IIFF problem, proposed evidential networks based on joint belief function and conditional belief function, verified the effectiveness of the fusion and reasoning approach according to the simulation, this approach will play an important role in IIFF systems.