Research HPDBSCAN Algorithm Based Wireless Sensor Networks Uncertain Data

With the fast development of wireless communication technology, wireless databasetechnology and integrated electronic technology to be significantly improved and widely used,the wireless sensor networks which with monitoring capability to calculate the storage capacityand data transfer capabilities are very popular at home and abroad in recent years research andapplication areas. Wireless sensor network is capable of real-time monitoring of variousenvironmental perception and gathering network distribution area or monitoring object applied todomestic and foreign researchers in areas such as military, transportation, environmentalmonitoring and agricultural production.The WSN play an important role in the national economicconstruction. Wireless sensor network applications is developing however, it is still subject to anumber of factors, Wireless network instability, poor security, network latency, less energy sensor,around the complex environment that affect the accuracy of the raw data, resulting in theperception of data uncertainty, uncertainty data has become one of the hot academia.Uncertainty data processing algorithms are evolved from classical to determine dataalgorithms, including classic uncertain data algorithms Skyline queries, U-AHC frequentset-based data mining algorithm based on clustering.These algorithms base on clustering to findclusters due to their superiority. Clustering analysis can better find the degree of similaritybetween the data and the formation of cluster type, traditional clustering algorithms to deal withthe uncertainty of data can not be accurately and can not find similar data objects into a cluster inthe data mining process so it can not restore the exact state of the data object.To solve uncertain data in wireless sensor networks, the probability density of the uncertaintyof data distribution probability clustering and Hilbert coding technology mappingmultidimensional data into one-dimensional data space, through improved Hilbert codingtechniques, and this based on the improvement of the uncertainty of the data based on theHilbert-R the tree index HPDBSCAN algorithm clustering. By programming the simulation, therepeated experiments verify the experimental results show that the HPDBSCAN algorithms’pretreatment effect better and more suitable for clustering uncertainty data than PDBSCAN，andFOPTICS clustering algorithm.The main work of this paper is as follows:(1)DBSCAN algorithm on the basis of the classic Hilbert-R tree to improve space utilizationand efficiency index. Hilbert curve has good clustering Hilbert coded values close to thedata points within the same leaf node, close to its spatial location. Data pre-processing, shortenthe time of clusters.(2)Improve the Hilbert encoding value formula, close to the spatial location of the data points to the Hilbert encoding value is close to，data clustering is efficiency and accuracy assured.(3) With the data points Hilbert encoded value difference probability threshold PTImultidimensional complex integration operation is converted into a one-dimensional integraloperation, and reduces the computational costFinally, the work of this thesis is to summary and prospect HPDBSCAN algorithms inwireless sensor networks.