Research On Node Localization Of Underwater Wireless Sensor Network Based On Machine Learning

Underwater node positioning is an important foundation for the application of Underwater Wireless Sensor Networks(UWSN).Accurate node location information is a prerequisite for the effective data collection of sensor nodes.However,due to the complex underwater environment,the deployment of underwater nodes is sparser than that of land nodes,the structure of the underwater network is uneven,and the nodes are difficult to recover,which brings difficulties to the positioning of underwater nodes.Machine learning node positioning can improve the accuracy of node positioning and the robustness of the algorithm.Therefore,based on the machine learning algorithm,this paper optimizes the traditional ranging positioning algorithm and the DV-Hop positioning algorithm.The main research contents are as follows.(1)Research on the support vector machine algorithm model and the topological structure of UWSN.Taking the edge length of the network area,the ranging error and the number of anchor nodes into account in the node location algorithm,an underwater 3D node location algorithm based on improved weighted least squares support vector machine is proposed.The algorithm divides the underwater three-dimensional space into several cubes,and uses the distance vector between the anchor node and the grid intersection point as the training set for training.The multi-class pattern recognition method is used for classification,and the distance vector from the unknown node to the anchor node is used as the test set to determine the node coordinates.Experimental simulation shows that the algorithm can effectively solve the positioning problems caused by underwater environment noise and sparse anchor nodes.(2)Aiming at the low positioning accuracy of the DV-Hop positioning algorithm caused by the three-dimensional structure of the underwater wireless sensor network and the communication differences of anchor nodes,a DV-Hop positioning algorithm based on random forest optimization is proposed.Firstly,the distance vector and the number of hops between anchor nodes are obtained by DV-Hop positioning algorithm,and a random forest model of anchor nodes is established.Then obtain the number of hops between the unknown node and all anchor nodes,and input the model to predict the distance relationship between the unknown node and the anchor node.Finally,the maximum likelihood estimation method is used to estimate the position coordinates of the nodes.Experimental simulation shows that the algorithm effectively improves the positioning accuracy of the DV-Hop positioning algorithm in an underwater environment.