A New Inconsistent Context Fusion Algorithm Based On Bp Neural Network And Modified Dst

As the number of various sensors grows fast in real applications such as smart city and intelligent agriculture, context-aware systems would acquire raw context information from dynamic, asynchronous and heterogeneous context providers, but multi-source information usually leads to the situation uncertainty of the system entities involved, which is harmful to appropriate services, and specially the inconsistency is a kind of main uncertainty problems and should be processed properly. A new inconsistent context fusion algorithm based on back propagation (BP) neural network and modified Dempster-Shafer theory (DST) combination rule is proposed in this paper to eliminate the inconsistency to the greatest extent and obtain accurate recognition results. Through the BP neural network, the situations of entities can be recognized effectively, and based on the modified combination rule, the recognition results can be fused legitimately and meaningfully. In order to verify the performance of the proposed algorithm, several experiments under different error rates of context information sources are conducted in the personal identity verification (PIV) application scenario. The experimental results show that the proposed BP neural network and modified DST based inconsistent context fusion algorithm can obtain good performance in most cases.