Research On Wireless Sensor Network Algorithm Based On Data Compression

Wireless Sensor Network is a self-organized communication network, which is deployed by many sensor nodes in certain area. It monitors environment and collects information of the monitoring object through sensors working together. These data will be processed and sent to the relevant monitor via wireless communication and chain structure by self-organizing LAN. In the practical application, some nodes will be placed the place where the environment is harsh, and hard to reach, and in typical protocol, like the LEACH and the HEED, a lot of energy is used in the process of transmitting data. So how to improve the utilization efficiency of energy, and extending the life cycle of the network becomes very import.For the above problems, the WSN data compression model based on clustering routing and the PCA algorithm are presented in this paper. It reduces a lot of transmission data by using multiple-PCA data compression. The compressed data is transmitted in the hierarchical chain, which reduces the energy consumption in the process of data transmission effectively. Firstly this paper introduces the research state in wireless sensor networks. Then the data compression is introduced in detail from the time domain, space domain and time-space domain. Finally, through the MATLAB simulation demonstrates the feasibility of this algorithm.In the cluster head selection algorithm, it determines the cluster number and the cluster radius of each layer in the three routing layer according to the hierarchical routing rules and the weights of living nodes in networks. Firstly it calculates each node weights of round, sorts them in descending order by size, and the largest weight node will be selected as the third layer cluster head, then the second layer cluster head and the first layer cluster head are chosen according to the weights and the distance between same layer head, which must be greater than the cluster radius; secondly all nodes and the first layer cluster head join to last layer head, which is the closest one to it. Finally, when part of the cluster head energy depletion, the cluster head will be reallocated, and this mechanism ensures the cluster heads are uniformly distributed in the monitoring area. In the data compression algorithm, the cluster head compress the data matrix which is composed by the data from the cluster nodes. Then the mean in each column, the corresponding center difference matrix, the covariance matrix, the eigenvalues and the eigenvectors are got, and making the eigenvalues in descending order, getting the k eigenvalues which the value is more than 85% in the total characteristic value. With the original data matrix and the corresponding eigenvector matrix of k eigenvalue multiplication, the time domain compression matrix is got, and then sends it to the second cluster head for space domain compression. This algorithm compresses the data alternately in time-space domain, which reduces the transportation of relevant data. Finally it introduces the definition, classification and detection algorithm for abnormal data.Theoretical analysis and simulation using MATLAB show that data compression algorithm can reduce the node redundant data transmission and energy consumption. In the same lifetime it has more living nodes and increases the energy efficiency compared with LEACH, HEED protocol, so that it can extend the life cycle of the network. Finally, the paper has a summary and optimization tips for the proposed algorithm.