Research On Multi-sensor Data Fusion In Wireless Sensor Networks

As the rapid improvement of sensors' capacity in sening,storage and computation,wireless sensor networks(WSN)have been widely applied in industrial monitoring,disaster monitoring,intelligent control and many other fields.However,data generated by sensors is vast and wide range,which is also uncertain and from multiple sources.Decision fusion methods of multi-sensor data fusion can not only reduce the amount of data transmission and energy consumption but also effectively improve the data processing accuracy.So,researching on multi-sensor data fusion methods,expecially decision fusion methods are of great value and meaningful.Based on the detailed analysis on multi sensor decision fusion in wireless sensor networks and relative theories,the evidence theory application on multi sensor decision fusion methods in wireless sensor networks is the key research content,mainly includes the following two aspects.(1)For the application of target classification decision in WSN,the reliabilityprobability(RP)decision fusion algorithm based on evidence theory is proposed and then,a simple and explicit fusion rule is derived.The unceartainty of decisions' probability is measured by the belief function.And based the simple support evidence belief frame,a reliability-probability basic belief assignment(BBA)construction algorithm is put forward.First,the local decision's reliability is measured by the relative distance between targets' data and the sample data.Then,based the reliability-probability BBA algorithm,the corresponding appropriated BBA is assigned to each local decision.Finally,through the Dempster fusion rule,the explicit formula of global belief assignment is deduced and the explict novel fusion rule is clearly derived.The new decision fusion method just only needs to send the local decision information to the decision center,which not only effectively improves the accuracy of the target classification decision-making,but also greatly decreases the amount of data transmission.(2)Due to the operating environment of the decision-making system is unknow,the data collected by sensors is not absolutely reliable and correct.Conflict data between decision results is inevitable.In order to solve the conflicting data fusion problem,the mutual support degree between evidence and the uncertainty of the evidence itself are both considered.And according to the averge weighted evidence method,a conflict evidence combination rule based on evidence distance and uncertainty measurement is proposed.First,through the evidence distance,calculate the relative distance between evidence,which represents the conflict degree between evidence.Then,the whole evidence is divided into two parts: the credible evidence and the incredible evidence.Then,a novel belief entropy is applied to measure the uncertainty of the evidence.Finally,the weight of each evidence is calculated and used to modify the evidence before using the Dempster's combination rule.Finally,numerical examples' results prove that the proposed method can effectively handle conflicting evidence with better convergence.