Studies On Data Collection Method For AUV Based On K-Means In Underwater Sensor Networks

The Underwater sensor networks(UWSNs)are underwater monitoring network systems made up of sensor nodes with acoustic communication and computing ability.UWSNs can be widely used in the area of the survey of marine resource,water environment monitoring and water environment monitoring.So the domestic and foreign research institutions have paid highly importance to it gradually.UWSNs have specific character of sparsely deployed sensors.So it is difficult for the data transmission and aggregation through multi-hop routing of sensors.To fulfill the task of collecting all the data,a solution that involves the use of autonomous underwater vehicle(AUV)has been widely applied,which the AUV's movement in the networks,visitation to the sparsely deployed sensors and reducing the energy consumption are key problems on the premise that all the sensor nodes' data can be collected.In this thesis,we proposed a data collection method for AUV based on k-means in underwater sensor networks.The validity of the method is proved by large numbers of simulations.The main contents and innovations of this thesis are as follows:(1)A data collection method for autonomous underwater vehicle based on k-means in underwater sensor networks is proposed.Firstly,the UWSNs are divided by related K-Means algorithm and the final division result,several “network subclusters”,is formed with the constraint of communication radius of sensors.Secondly,we make the center of the network subclusters as the data collection points.Lastly,the optimal path is formed based on the data collection points.Extensive experiments have demonstrated that the proposed method can effectively plan the path for autonomous underwater vehicle to gather all sensors' data with advantages such as short length of the path and data collection time and high energy efficiency.(2)We design a dynamic adjustment mechanism of network subclusters based on class distance.The migration of sensor nodes may occur passively or actively because of the complex underwater environment.The sensor nodes will compute the class distance of the nearby subclusters based on their location.Then they can get the probabilities of joining the nearby subclusters and realize the adjustment of network subclusters.The AUV will make a new path planning based on the new network division and complete the task of data collection.The validity of the adjustment mechanism is proved by simulation.The research work and results of this paper have certain promotion and reference value for the development of the data aggregation theory and application in underwater sensor networks...