Research On Correlation And Key Technology In Wireless Sensor Networks

Wireless sensor networks(WSNs) is different from the traditional network, its limited energy, storage space, computing speed, communication distance and band width request that the research in the field must focus on how to minimize node workload, including traffic, computation, storage, etc., in order to achieve the goal of saving resources, and to extend the network lifetime. In WSNs, there are correlations existing among both the nodes and data, involving spatial and temporal correlation. Recently, some literatures began to focus on the correlation of data types, and correlation plays an important role in WSNs energy-saving technologies. Correlation research focused on the correlation mining and evaluation, the data query and fusion based on correlation, the protocol stack design considering correlation, the estimation of missing values using correlation, and some other key technologies. This paper can be divided into three parts as follows.1. Correlation research. Exploit the correlation characteristics to save energy, improve data quality, reduce the delay in WSNs. The paper first study the concept of correlation. By reviewing the development status and application requirements of WSNs about the latest multi-type data, we propose the multi-type attribute correlation concept to reveal the possibility of multi-type data can be fused, analyze the manifestation of correlation, and discussed behavior and data characteristics existing in the correlation. Behavior characteristics, the paper discusses the relationship between correlation and traffic load, the proposed behavior correlation is divided into local correlation (LC) and global (GC), the behavior evaluation model is established, and based on this, the evaluation methods of the spatial, temporal and multi-type attribute correlation are established respectively, the local and global correlation evaluation methods are as well as established based on them. Data-correlated characteristics, the paper describes the evaluation methods of spatial, temporal and multi-type attribute correlation in detail.2. An adaptive MAC protocol based on correlation. According to the relevant evaluation model, this paper proposes an adaptive MAC protocol based on correlation (CAS-MAC). CAS-MAC is from AS-MAC. The paper first makes an analysis about the flaws of AS-MAC protocol in terms of both theoretical and experimental conditions, and then adjusts the frame structure ratio of AS-MAC to adapt to impaction on the MAC protocol performance by different load. In addition, CAS-MAC introduces a determine mechanism of the timeliness of data packets for dealing with the adverse effects of the failure packet. The simulation results show that AS-MAC protocol can improve the adaptive mechanism of AS-MAC protocol, and enhance the ability of MAC protocol to adapt to the ultra-high load and unstable environments.3. Data fusion based on correlation. Based on previous evaluation algorithm, and considering the characteristics of multi-type data, this paper first presented a second evaluation method of Floyd algorithm (Floyd), accurate and simplified representation of the correlation. The data subscript is coding into three-dimensional in accordance with the spatial-temporal-type attribute correlation firstly, and three-dimensional subscript data is transformed into two-dimensional matrix with a three-dimensional, and further into the one-dimensional sequence; then in accordance with the correlation transfer rules, a second evaluation of the correlation Floyd algorithm to calculate the maximum correlation between any data. Secondly, based on the second correlation evaluation, Floyd-Artificial Fish selection algorithm is proposed, including an "optimal solution-optimal solution dimensions" of two-way feedback mechanism. In the mechanism, Floyd algorithm makes up for the defects of artificial fish swarm algorithm on the convergence. Using artificial fish swarm algorithm for solving the optimal solution when the stability of Floyd algorithm and midpoint is difficult to determine the dimensions of the defect to optimize. As the dimension is difficult to determine in the Floyd algorithm, the mechanism optimize. Finally, the paper tentatively proposed a data prediction algorithm named Floyd-Artificial-Fish-BPNN, using the results of Freud-Artificial Fish data selection algorithm as the input of the algorithm, and from Freud-The artificial fish swarm algorithm trained BPNN has been estimated value and the relative error. The simulation results show that the algorithm can effectively improve the efficiency and quality of data selection.This paper made a detailed analysis and elaboration about the status quo and characteristics of correlation and proposed a correlation mining and evaluation methods. Correlation is applied in energy-saving technologies of wireless sensor networks. Finally, we summarize the paper.