Consistency test based on self-support degree and hypothesis testing for multi-sensor data fusion

Focusing on the possibility for observed results with false or erroneous information of sensors which may affect data fusion estimation, a new algorithm is proposed based on self-support degree and hypothesis testing. It is significant to take the consistent test method to identify and remove the observations of failed sensors by checking on the observed results and then the data fusion algorithm could be used for estimation effectively. Based on posteriori probability used in consistent test and the knowledge of hypothesis testing, we regard the problem of consistency test as the hypothesis testing of the difference between two population means. Meantime, the multi-valued problem of multiple sensors is researched. Based on the observed values of different time, each sensor's consistency value of different time about hypothesis testing can be obtained. The results of simulation show the simplicity and effectiveness of the new method based on self-support degree and hypothesis Testing for evaluating the quality of each sensor observations, identifying false and erroneous observed results and providing data fusion estimation with the reliable consistent sensor group.