Data Aggregation Based On Clustering And Wavelet Compression In WSNs

As one of the most important technologies in the 21 st century,sensor networks have already had a greatly impact on our lives.We can utilize the Internet of things to manage the sensing data acquired through sensors.Compared with the sensor networks,the Internet of things is more diverse and can concretely realize the function of the sensor networks.While a large number of sensor nodes are deployed in different ways in various environments,there are many diversities and temporal correlations between sensors.In order to extract more effective information,we need to add the data aggregation technology in the process of data collection,which aims to realize more efficient transmissions by reducing the duplicate data to achieve the monitoring requirement of application.Due to the constraints of cost and volume,the nodes in wireless sensor network are generally energy-constrained.The wireless sensor network can benefit from data fusion technology,which can satisfy the requirement of the application design.In this thesis,we study the characteristics,principles and corresponding performance metrics of the typical spatio-temporal correlation data aggregation algorithms,we propose the corresponding data collection and clustering algorithm according to the temporal and spatial correlation between the application requirement and the collected data.Moreover,we propose the corresponding lifting wavelet data redundancy algorithm for the redundancy among the collected data.The main contributions of this thesis can be summarized as follow:(1)By considering the temporal and spatial correlation of the acquired data among neighboring nodes in wireless sensor networks,we propose a data collection clustering model for wireless sensor networks based on application requirements.Through the data dependency of the adjacent area,a node with high residual energy and strong data representation is selected as a cluster member participating in the data collection in the sensor network,and dynamically adjust the network cluster according to the change of the residual energy and the real-time data size and detection of abnormal node data changes.The simulation experiment results show that the clustering model can save more energy under the requirement of a certain changeable data transmission.(2)Aiming at the temporal correlation at different time points and the spatial correlation among the different cluster members,we propose an effective eliminable lifting wavelet data compression algorithm.This algorithm can effectively remove a large amount of data redundancy,because some wavelet coefficients are discarded,which can guarantee the accuracy of data recovery while occupying only a small amount of storage space.