Deployment Optimization Of Monitoring Sensor Networks For Marine Ranching

Marine fisheries in China have developed rapidly.However,due to the overfishing and extensive farming regardless of the environment,as well as coastal urban construction and industrial development,the coastal ecological environment has been destroyed.China has implemented marine ranching construction to solve the conflicts between marine ecological environment and development issues.After decades of development,large-scale output has been achieved in coastal provinces such as Liaoning,Shandong,Zhejiang and Guangdong.However,the construction of marine ranching in China is still in the initial stage.Strengthening the informatization of marine ranching is an urgent requirement for the transformation and upgrading of marine pastures.As an artificial fishery,there are a large number of natural seabed animals and plants such as fish,algae and aquatic plants in the marine pastures.Artificial reefs and other artificial objects are also present.The monitoring of these animal and plant conditions has important research significance.At present,most of China's marine ranching monitoring technology is still in the process of collecting marine data by using ships and platforms.The marine monitoring network is constructed by mooring buoys and sensors to obtain marine pasture information in real time.The level of cognitive management has broad application prospects in the future real-time monitoring of ecological environment and scientific farming.Firstly,this thesis analyzes the characteristics of marine sensing monitoring network and various communication technologies,and proposes a sensor monitoring network architecture for marine ranching,which can timely and accurately grasp the environmental data in marine pastures,provide scientific basis for disaster warning and environmental monitoring.Secondly,the deployment problem of the marine pasture monitoring sensor network is described as the Sensor Networks Deployment and Optimization(SNDO)problem,and the Integer Linear Programming(ILP)model is established.In a series of small and medium-scale scenarios,the mathematical planning optimizer Gurobi is used to solve the problem,and the accurate network deployment scheme is obtained.The validity of the ILP model is verified and the deploylent of small and medium-sized networks can be guided.In large-scale scenarios,as the variables and constraints in the ILP model increase,the solution time of the ILP model increases exponentially,and some large-scale mathematical programming problems cannot be solved by Gurobi.Finally,for the problem that Gurobi can't solve,we use Genetic Algorithm(GA)to solve the SNDO problem,analyze the shortcomings of traditional GA in solving SNDO problem and improve GA.This paper proposes use greedy strategy to quickly generate the initial population of genetic algorithm to solve the problem that GA is too dependent on the initial population;use local search strategy to enhance GA's local search ability to reduce the redundancy of sensor nodes in the network.Finally,we verify that the improved GA has good solution through experiments of various scales and can be applied to solve large-scale SNDO problems.