Research On The Processing Of Uncertain Data On Wireless Sensor Network

Wireless Sensor Network is a comprehensive discipline, which is increasing the methods of human access to information and expanding the capacity of the interaction between man and nature. Uncertain data in wireless sensor networks research work is like wildfire, and some crucial technologies in many fields have made considerable progress, but there are still many problems. For the following reasons:(1) a broad application base. Uncertain data mostly has to monitor, the nature urgently need for a good solution to the performance of the program uncertain, but the application itself is a process of change which is increasing complexity to solve the problem. (2) Uncertain data management in wireless sensor networks still has a large number of issues which need to be studied. Since the restrictions on the sensor node itself, between divided network transmission and data acquisition and continuous changes in the physical quantity itself, there are inevitable conflicts. In essence, these data are uncertain data. Therefore, probabilistic models are often used to record data, but for many of the original technologies in traditional database are not application or need to improve for probabilistic models, so the requirements of new data processing also will be born.In this paper, we are based on studies of several key technologies of information fusion in wireless sensor networks, around the data processing algorithms to carry out the research of uncertain data in wireless sensor networks. Following major elements:(1) wireless sensor networks and uncertain data; (2) a new algorithm using the temporal and spatial correlation in uncertain data; (3) sliding window and the key technologies of Top_k query; (4) improved algorithm of framework is based on the buffer.Main achievements are as follows:(1) Through in-depth study of network architecture, analyze the characteristics of wireless sensor networks, and the research framework and the formal description were given,we study some key technologies of the traditional data processing, and the uncertainty of data will be face challenges.(2) Aiming at the uncertainty of tuples, a new algorithm using the temporal and spatial correlation is proposed. To use the historical data of wireless sensor network nodes and combine with Hermite and DESM model, the algorithm can estimate the approximated result of uncertain data by adaptively adjusting the weights of temporal model and spatial model for different applications. The algorithm can adaptive itself by BP algorithm in artificial intelligence, not only make full use of peripheral resources, but also the estimated results has smooth curve, then achieved better estimates of the effect. In the algorithm, the choice of the nearest neighbor nodes have improved, because the data of wireless sensor network's space has little change in short time, so we can be considered to be relatively similar to the stable, by alter the target data to the recent historical data for the comparison, find the nearest neighbor node. This kind of improvement can increase the reliability of wireless networks. Experimental result indicates the algorithm is effective and stable.(3) For the probabilistic model of uncertain data, according to its probability distribution function and the probability range, the queries results which using uncertain data probability model is more credible and more useful than the traditional method of inquiry, and has some value of reference. In this paper, we have introduced the sliding window model and the calculation of various queries, given the definition of the problem; analysis of the compact set:at first, by the difference between two adjacent compact set to eliminate these redundant, achieved compression to redundant data; second, reduced the number of updates compact set when the tuples arrive using a superset; and then use k* H size buffer to hold two new tuples, whenever update the compact set which is play a important role in query of sliding window, batch update the other compact set. Optimize the compact set of methods proposed improved the algorithm's time complexity and space complexity, then combine sliding window with technology of top_k queries, get a unified framework which is met a variety of inquiries and improved query efficiency.The wireless sensor network which from academic into the commercial has great opportunities, but also has challenges since the research is not mature, for example, network transmission error, measurement error, and so can not completely eliminate; the data of nodes is essentially the difference between the real world. Uncertain data processing problems will lead to more attention of scholars, there is still more work to be done and development.