Research For Heterogeneous Wireless Sensor Network Information Fusion Method Of Soil Monitoring

Soil is the material basis and output guaranty of agricultural production and is a precious natural resource of human, microbes living in it can degrade the household and industrial garbage and then include them into the matter cycle, hence keep the ecological balance. However, due to the misuse of the soil in recent years, soil erosion and pollution becomes a more and more serious matter, which gravely threats the ecological environment in our country. It’s necessary to monitor the soil environment quality in real time to improve the quality of agricultural products and to control and manage the soil erosion and pollution. Soil information is multi-dimensional and abundant because of the spatial and temporal variability. We have to integrate all kinds of information that is monitored so we can come to a more accurate result.There is no theoretical breakthrough in the key technology of information fusion, neither a wireless network application system with independent intellectual property rights which can obtain the information of a large area of soil in time. Aiming at that problem, this thesis studies the method of heterogeneous wireless sensor network information integrating to monitor the soil information. And according to the different purpose of monitoring, we mainly do the research on two different aspects, namely, dynamic monitoring and static monitoring of soil.A. Dynamic monitoring of soil information fusion algorithm researchBased on the different focus of attention,the dynamic information monitoring fusion can be decomposed into two issues:1. Single objective monitoring, namely, the issue whether there are exotic objects or not. 2. Multi-target monitoring, namely, the issue how many exotic objects there are. The main research of this part is as follows.1. A HWSN model to monitor the soil dynamics is established. The transmission form of the Multi-objective monitoring model signal is assumed reasonably.2. In the case of heterogeneous WSN, a new research aiming at information fusion algorithm of targeting monitoring has been studied. Firstly, we start from Likelihood Ratio(LR) fusion rules, and deduce the Chair-Varshney fusion rule under the circumstances of single target and high SNR(signal-to-noise ratio), the Maxmum Ratio Combiner(MRC) and Equal Gain Combiner(EGC) fusion rules under the circumstances of low SNR. Then those results are generalized to the multi-target monitoringandfusion.3.singletargetlr-cs(lrincs)andmrc-cs(mrcincs)fusionalgorithmundertheconditionofonlychannelstatistics(cs)isknownareobtained.andtheyaresuccessfullygeneralizedtothemulti-targetmonitoringandfusion.thentherelevantmulti-targetfusionalgorithmarededuced.B.soilstaticmonitoringinformationfusionalgorithmresearchwemainlystudiedtheapplicationofdempster-shafer(d-s)evidencetheoryinthestaticmonitoringinformationfusion.someproblemsmayarisewhend-sevidencetheorywasappliedinthestaticmonitoringinformationfusion;thispartoftheresearchiscarriedoutwithacorrespondingimprovementtoaddresstheseissues.themainresearchcontentsareasfollow.1.tosolvetheproblemthatdempsterevidencesassemblymaycauseaphenomenonofintuitivewhenitisappliedtosoilstaticmonitoringinformationfusion,animproveddssc(d-sonsimilaritycoefficient)algorithmusingweightedsimilaritycoefficientisproposed.throughanalyzingexamplesofsoilstaticmonitoringinformationfusion,theeffectivenessoftheimprovedisproved.bycomparisonwithotherimprovedalgorithm,thesuperiorityoftheimprovedalgorithminthispaperisverified.2.tosolvetheproblemofsoilconflictingevidencemeasurement,the-mbetpunsimmeasuringunitbasedonsimilarmatrixtransformationisproposed,whichavoidsthedefectthatthetraditionalconflictingmeasuringunitkhasandismoresensitiveanddistinguishing,comparedwithothertwoconflictingmeasuringunit,namely,bpadanddifbetp,andrelativelyhaslesscalculationamount.3.takingthelargesamplesizeofsoilintoconsideration,bpa(basicprobabilityassignment)isgeneratedbasedonthefuzzynumberinthefuzzytheory.