Research On Decision Fusion Of Wireless Sensor Networks Based On D-S Evidence Theory

Due to the diversity,complexity,uncertainty and redundancy of the representation of information obtained by sensors,the energy consumption of the sensor node is increased,thereby reducing the life cycle of the network.Thus,the wireless sensor network decision fusion technology emerges as the times require.It may not only reduce the amount of data transmission and the transmission energy consumption,but also improve the processing accuracy of sensor network data effectively,which has very important practical significance.Recently,the Dempster-Shafer(D-S)evidence theory is widely used to solve decision fusion problems in wireless sensor networks.However,its basic probability assignment function model is usually difficult to be constructed accurately.Besides,making highly conflicted evidences fusion usually obtains some abnormal results,which lead to wrong decision-makings.Thus,three decision fusion methods are proposed in this paper based on D-S evidence theory.1.Since existing evidence conflict measures may not distinguish between single subset and multiple subsets effectively,an evidence composition approach based on the improved evidence distance is proposed.Firstly,on the basis of the Jousselme evidence distance,rationally divide the similarity Jaccard coefficient matrix.Through calculating each evidence weight to correct the source of evidences,the result of the fusion decision is finally obtained.Simulation results show that the proposed approach can not only effectively characterize the evidence conflict of the sensor,but also improve the accuracy of decision fusion.2.Traditional evidence theory fusion rules usually may not distribute evidence conflicts reasonably,which makes the result of decision fusion is often contrary to the facts.In view this,a multi-sensor decision fusion approach based on improved evidence theory fusion rules is proposed and applied to the target recognition.Firstly,an extended conflict coefficient is defined,which not only has the advantage of the small calculation amount of the global conflict coefficient of the traditional evidence theory,but also can adapt to a variety of real situations.Secondly,according to the idea of "redistribution of conflicts",a new rule of evidence theory fusion is finally constructed.Simulation results show that compared with existing approaches,the decision fusion rules of the proposed approach are simple and easy to implement.They have the fasted convergence speed.3.Taking the intelligent greenhouse environment control under the wireless sensor network as the application background,a data preprocessing and decision fusion approach based on D-S evidence theory is proposed.Firstly,the greenhouse environmental data collected by various sensor nodes is used to preprocess the outliers(correct adaptively and clustering).Secondly,a basic probability assignment approach based on weighted similarity is proposed,and the decision fusion is finally carried out within the framework of evidence theory.Simulation results show that the approach has higher accuracy and effectively reduces the uncertainty and decision risk of the fusion results.