Multisensor Distributed Interval Estimation Fusion

This paper mainly discusses multi-sensor interval estimation fusion.We obtain the results as follows:(1) When n sensor estimates {I1,,I2,....., In} along with their confidence degrees { } are given and they are independent of each other, an optimal interval estimation fusion can be obtained based on a combination rule and the two popular optimization criteria. Moreover, adding the condition that at most f or at least f out of {I1,I2,....., In} arefaulty, we can extend the results above to obtain a conditional combination rule and the corresponding optimal fault-tolerant interval estimation fusion. In addition, Marzullo' s fault-tolerant interval estimation fusion is a special case of our method.(2)when n sensor estimates {I1, I2,.....,In} without the information of their confidence degrees are given and at most / out of {I1, I2,.... , I n} are faulty, an optimal fault-tolerant interval estimation fusion .which gives interval output and the corresponding belief level can be obtained based on belief function, the prior information that a sensor usually has a smaller probability of being faulty than being normal, and the three optimization criteria. This method not only shortens the output intervalgiven by Marzullo method, but also its output is more reliable than that given by Prasad, lyengar, Kashyap and Madan. Marzullo' method is also a special case of ours.(3)We show Marzullo function satisfies neither Lipschitz condition nor local Lipschitz condition by two examples.