Abstract:
Network-centric warfare (NCW) and the interoperability of joint and coalition forces are among the future warfighting concepts identified
by defence. To realise the goals of interoperability and shared situation awareness for NCW, it has long been acknowledged that
data fusion is a key enabling technology. Typically, however, distributed data fusion, which is relevant to NCW, and the fusion of disparate
types of uncertain data, which is relevant to interoperability, have been investigated separately. Ideally, for shared situation
awareness, the system should be capable of performing both aspects of data fusion. In this paper, these facets of data fusion are considered
in unison for the automatic target identification problem. In particular, novel Bayesian and generalised Bayesian algorithms
are formulated for fusing estimates of target identity generated by local heterogeneous data fusion systems in a network, each of which
expresses target identity estimates as either finite probability distributions or Dempster–Shafer belief functions. An example drawn from
the literature is used to illustrate the algorithms and their relative performances are assessed in the context of the example to identify
issues of possible relevance to distributed target identification in a more general setting.
