Distributed Clustering Technology Research And Its Applications

Clustering analysis whose strong advantages drawing the attention of many researchers has been successfully applied in many areas at present. With the rapid development of Internet, however, data has begun into distributed storage to satisfy the explosive increase of the information. It is almost impossible for the data from different sites to be focused on one central point due to limitations of network bandwidth, privacy protection and memory capacity. After previous studies about clustering technique summarized, the following tasks have been carried out in this paper.DK-Dmeans algorithm which is a modified version of K-Dmeans algorithm is proposed to reduce the communication consumption. Time efficiency of the algorithm is evaluated through simulation experiments and the results show that this algorithm achieves the effect as common K-means.HII DDM based onⅡDDM is designed according to a hierarchical idea with decomposing a large system into several smaller subsystems to address the issue for the defects ofⅡDDM. With the comparison of simply usingⅡDDM model, the new model has stronger flexibility, more efficiency of the implementation, and less communication consumption, especially for large-scale data clustering.An intrusion detection system for wireless sensor network has been designed, which is based on the new distributed clustering algorithm and the model. This system is on the basis of multi-Agent distributed architecture to obtain good recognition results with very promising application prospects.