Study of Method for Data Fusion Assurance in Wireless Sensor Network
|School||East China University of Science and Technology|
|Course||Applied Computer Technology|
|Keywords||Data fusion CDMA election intrusion detection|
In recent years, with the continuous development of wireless sensor network (WSN), increasingly importance has been attached to security problems. As the deep-seated technical in WSN, the security problem about data fusion has become the focus of study at home and abroad. Since be charged with heavy responsibilities of fusing data, and its fusion data always contains a great deal of valid information, the fusion node becomes the focus of attack in WSN. How to ensure the safety of fusion nodes, and more important, how to ensure the fusion result correct and valid, appears to be a hot issue that catches many scholars’ attention in recent years.Both the witness-based approach proposed by Du et al. and the direct-voting approach proposed by Pai et al played a certain part of data fusion assurance, however, there are still many shortcomings, for example, the time for judgment of whether a copy of fusion result is valid or not is too long, the handing method of abnormal behavior nodes including normal failure nodes and intrusive nodes is inappropriate.This thesis proposes a fast and tolerant voting mechanism (FTVM) which will definitively overcome the mentioned weak points. This mechanism adopts the MAC protocol based on CDMA technical, which makes it possible for some fusion nodes to communicate with the base station in a channel simultaneously without arousing data collision, and plus the approach of choosing witness nodes by base station, many witness nodes could poll together, which will accelerate the election by a substantial degree. Besides these, this thesis also proposes a method of intrusion detection which is used to determine whether a fusion node is intruded or not through supervising the behavior of fusion nodes by base station for a long term. With it, the network tolerance is strengthened clearly. In the end of this thesis, a series of experiments and results analysis are shown to verify the efficiency of FTVM.