Research On Data Aggregation Scheme In Dynamic Wireless Sensor Network

In recent years,the popularity of IoT applications has brought unprecedented attention to the research in the field of Wireless Sensor networks(WSNs).The WSNs is constructed by a large-scale sensor nodes connected by wireless links in a critical monitoring region.These sensor nodes periodically sense the state of the monitoring region and aggregate the perceived data into the central processing unit through wireless transmissions,which realizes the real-time monitoring in unattended situations.Obviously,aggregating the data is one of the most important functions of WSN.Meanwhile,due to the limited energy capacity of sensor nodes,research on improving the energy efficiency of data aggregation process has important theoretical and practical significance.The existing data aggregation schemes based on compressive-encoding use the compressive sensing algorithm to map the original data into the low-dimension during transmission process,which reduces the number of data quantity passing in the network and saves the energy consumption.However,there are also many shortcomings by considering the characteristic of the WSN:First,the existing schemes cannot fully incorperate the network topology features to design a more detailed data aggregation scheme,which further improve the efficiency of the data transmission.Secondly,the existing schemes are mostly based on static scenarios,ignoring the actual dynamic feature of wireless sensor networks,which causes many problems in practical applications.In view of the shortcomings of the above existing schemes,the research content of this thesis is as follows:1)For the problem of high data transmission required by traditional data compression scheme based on compressive sensing algorithm under tree topology,a data aggregation scheme combining tree topology features is proposed,which is solved by disassembling the tree topology into multiple independent data transmission paths.Thus a measurement matrix with a lower dimension is obtained,which further reduces the amount of data transmission during data aggregation and saves network energy consumption.2)For the problem that the dynamic wireless sensor network results in a large amount of update energy consumption due to the increasing number of nodes,thus a measurement matrix based on Vandermonde matrix and corresponding network expansion scheme are proposed.At the same time,the design of key parameters of the measurement matrix is given under the actual data frame transmission process.The simulation results show that the scheme has higher energy efficiency in the dynamic data aggregation scenario.3)For the dynamic wireless sensor network,due to the increase of the number of nodes,the existing data aggregation scheme based on compressive sensing encoding has higher energy consumption,and an efficient update and expansion scheme for the existing measurement matrix is proposed.Based on ensuring the original matrix structure is unchanged,the optimal encoding vector is allocated to the nodes newly added to the scene,which results in effectively coping with the small-scale expansion of the scene.4)For the imbalance of dynamic wireless sensor network topology due to the increase of the number of nodes,the location update scheme of the sink node is proposed,and the autonomous pathfinding task of the sink node is realized by the reinforcement learning algorithm.Finally,the performance of the proposed scheme is verified in multiple simulation scenarios.