Research Of Data Fusion In Wireless Body Sensor Network Based On Fuzzy Sets And D-S Evidence Theory

With the development of information technology, the wireless communication plays more and more important role in the economic development and life. The development of wireless network shows the Ubiquitous and heterogeneous feature. The FP7 launched by EU in 2008 increased the researching intensity of next-generation wireless network in order to develop to wireless cognitive network on the basis of the integration of the network.As the application of wireless cognitive network, wireless Body Sensor Network play a more great role in medical care, care of special people and community medicare.however, as an important public application network, Body Sensor Network is still in elementary phrase, especially the Body Sensor Network based on wearable sensor node,the research is mainly on how to construct system architecture and service platform, the research faces lots of technological challenges.Wireless Body Sensor Network is also a kind of wireless sensor network, it has the feature of wireless sensor network such as the limitation of computation and communicating ability, the main disadvantage is that its battery can't be changed frequently, so the main researching point is how to prolong the lifetime of the sensor network.The data fusion structure first preprocesses the human's body's feature data, then decide whether the data is normal or abnormal, and take different measures according to different data.For the correct data, we can abstract the user's vital signs, the paper compares the centroid algorithm, center algorithm and maximum likelihood algorithm, we choose the centroid algorithm according to the sensor's computing ability and storage limitation.Evidence theory is an important researching direction of data integration, because it can describe the uncertainty of the material, transform the uncertainty of proposition into uncertainty of set. Reasoning through the set's theory can satisfy the condition lower than probability theory. This paper also introduce the fuzzy sets' theory including the relationship of membership function and basic trust distribution function。The abnormal data can be used to judge whether it results from the body itself or the environment factors. We use fuzzy set theory to define basic trust distribution function considering the influence of the outer environment. When integrating the evidence,we use the distance of the evidence, improve the Josselme distance formula through grid degree of nearness of fuzzy set. In the analysis of the instance, we compare the Josselme's algorithm and ours, the result shows that our algorithm show good time complexity and it also solves the problem that the features'false positive resulted from outer environment. Error data is used to decide whether the reliability is too low, whether it should be replaced. The bathtub curve theory shows that the failure rate of component first decline and then increase gradually, we judge whether we need replace the component through failure rate and the changing trend.