Managing deceitful reports with the transferable belief model

Suppose a deceiver interacts with the communication channel between a sensor and its end user. The deceiver is a liar, and tells systematically the false. So when he/she says that the measurement is in A, it means it is not in A. If the end user knew the deceiver is active, it could be easy to rebuilt truth. In practice, all the end user knows it that ´maybe´ the deceiver is active. Representing such a complex situation within the transferable belief model (TBM) is quite straightforward thanks to the concept of the negation of a belief function. If m^Ω is basic belief assignment define on Ω, its negation m~^Ω is a basic belief assignment whose masses satisfy: for all A⊆Ω, m~^Ω(A)=m^Ω(A~) where A~ is the complement of A relative to Ω. Mathematically this negation allows to build a De Morgan algebra when using the conjunctive and disjunctive combination rules as defined within the TBM framework. We illustrate how this concept of negation can be used to manage possibly deceitful sensor reports. For multisensor data fusion systems, it is also possible to assess which of the sensor reports might be deceitful.