Topology Optimization And Intelligent Access Control Of Stereoscopic Marine Monitoring Networks

The development of Stereoscopic Marine Monitoring Networks(SMMNs)is essential to implementing marine economic powerful nation strategy and improving the comprehensive marine strength.It provides information and data services for the understanding,development,utilization,protection and management of the ocean.It also plays an important role in promoting marine scientific research,improving maritime emergency responsiveness,promoting economic and social development of coastal areas,and safeguarding national marine rights and interests.However,the development of marine monitoring networks is far behind that of terrestrial monitoring networks.Thus,it is urgent to conduct research on key issues in the deployment of marine monitoring networks.Network topology optimization and access control are the primary research issues to be addressed in the deployment and construction of the SMMNs.In this thesis,we have investigated the SMMN architecture,network topology optimization framework and access control mechanism.The main contributions of this thesis can be summarized as follow.Firstly,under the SMMN architecture,a multi-objective optimization(STMO)problem formulation is established,with the objectives of minimizing network deployment costs while achieving the maximal network lifetime subject to the constraints of tree-based network topology,connectivity,data flow and node battery capacity.To solve the formulated STMO problem,two algorithms based on greedy strategy and ant colony intelligence strategy are proposed respectively,and the time complexity is analyzed.A large series of simulations are conducted to verify the effectiveness and efficiency of the algorithms.Secondly,to address the limited energy problem of SMMN network node,we integrate the tidal energy harvesting technology into the SMMN architecture to construct a SMMN model(so-called EH-SMMN).Considering the characteristics of node energy harvesting,large underwater acoustic propagation loss and limited network bandwidth,together with constraints of tree-based network topology and data flow,an EH-SMMN topology optimization(ESTO)problem formulation is established with the objective of minimizing network deployment costs.Two algorithms based on path searching and set covering are proposed respectively,and complexity analysis and simulation experiments are performed.Numerical simulation results show that the proposed SC-ESTO algorithm can approach the optimal solution with polynomial complexity.Finally,we study the access control problem in the underwater acoustic sub-networks which has the largest number of network nodes within the multi-tiered architecture of the SMMN.A generic Markov chain model is developed to analyze the performance of the EH-SMMN with energy harvesting,capturing the salient features of underwater communication channel and the stochastic energy harvesting process.By jointly considering Value of Information(Vol),throughput and fairness,a network access parameter optimization problem is formulated with the objective of maximizing the Vol weighted throughput.A deep reinforcement learning based intelligent SMMN access control(DRL-ISAC)approach is further proposed to autonomously adapt the access parameter.Numerical results show that the proposed learning algorithm can significantly outperform the fixed transmission probability-based access algorithm and the random polling algorithm.