Improvement And Application Of DS Evidence Theory In Multi-sensor Data Fusion

With the deepening of the network warfare , data fusion technology becomes more and more important and becomes one of the great technical challenges of combat capability. In this paper, with the backdrop of National defense scientific research projects based on Ministry of Education, we major in multi-sensor data fusion algorithm of DS evidence theory. Content as follows:①The systemOn the basis of analysis of the system, we describe the system specially on the key technologies such as electronic detection, distributed interferenceand and intrduce the main research subjects of this article, multi-sensor data fusion.②Analysis and comparison of the two DS algorithm :weight-based distribution and matrix- based evidence theoryAfter the analysis of the DS theory Algorithm in time used and ability in conflict resolution,we found that the weight-based distribution evidence theory can be successful in conflict resolution, but it leads to a huge computation,which is not appliable to strong real-time requirement in battlefield . Matrix- based evidence theory can greatly shortened the time integration ,but when there is conflict between two evidence, the results of this formula may be perverse.③The improved DS algorithm based on Matrix analysisThe classic synthesis of DS evidence theory has two problem.One is it is bad to deal with conflicting evidence, the other is it demands excessive calculation.So we use the matrix analysis to improve the weight-based distribution DS algorithm . This algorithm will be used to resolve the conflict by the idea, that distribute the probability of evidence dy the average level of support proposition; using analysis matrix to overcome the problem of huge calculation and to reduce the burden of the node with limited ability.Improved algorithm will be applied in simulated battlefield in this paper to identify the aerial targets. Integrate the multi-radar sensors data to improve the accuracy in single cycle vertically . Integrate Radar, ESM, IFF sensor data to enhance range of identification horizontally. Integrate multi-sensor data fusion to improve the accuracy of discriminant in multi cycle . Through simulation and experiment prove that the analysis of this paper not only can effectively solve the conflict, but also greatly reduce the computing time.