Research On Data Aggregation Algorithm Of Decision Level In Wireless Sensor Networks

Wireless sensor network is self-organizing multi-hop wireless network system and is widely applied in the scientific, medical, commercial, defense and other areas. It is got a great attention at home and abroad as new information technology. Most sensor nodes of wireless sensor networks are powered by batteries, soenergy of network is very limited. Data aggregation technology is used to deal with lots of initial data, get rid of redundant and reduce the data transmission. In addition, the technology must process integrated operations to data and improve data accuracy. In addition due to the easy failure of sensor nodes, data fusion technology requires more than data, integrated operations, improve data accuracy.First, the evidence when combined operations in the computational complexity and the number of operations is related to focal element, so reduce the computing time of evidence combination can reduce the number of focal element computational complexity, thus reducing energy consumption, extend the node Survival time. Since wireless sensor networks with high redundancy of data characteristics, this article proposes the use of pre-treatment method based on the voting mechanism to reduce the combined operations in the evidence the number of focal element. Existing algorithms, although the truncated preprocessing algorithm can also achieve this effect to some extent, but sometimes it will be the impact of a node can not achieve satisfactory results. The use of the proposed voting mechanism based on the decision matrix preconditioning approach to pretreatment to prevent this from happening.Second, because the sensor nodes work in more complex environments, some nodes will be electromagnetic interference, vibration and other disturbances, which affected the perceived accuracy of the data. If you do not have this kind of data nodes removed, then the fusion will influence decisions in the data accuracy of the results. Sensing using wireless sensor networks with high redundancy of data characteristics, this thesis presents the pretreatment criteria based on Dike Song. This method can exclude large data errors. And error handling than other methods used in, Dike Song guidelines less computation, so the calculation of energy consumption and computing time are small, which can lower the cost of improving the accuracy of fusion.