Research On Topology Control Algorithm Based On Machine Learning In Free Space Optical Communications

Free-space optical communication is an optical communication technology that uses light propagating in free space to wirelessly transmit data for telecommunications or computer networking.It has the advantages of large bandwidth,high speed,high security,flexible networking and low cost.It has become a research hotspot in the field of communication.However,due to the narrow FSO beam,the difficulty of terminal integration and the limitation of node degree,as the first step of communication,there are some urgent problems to be solved in FSO networking.At present,the research of FSO networking is not mature,and the corresponding reports are relatively few at home and abroad.Because the FSO requires the Line of Sight and the alignment requirements of the transceiver terminals are higher,At the same time,unlike traditional wireless networks,the FSO link power is limited and affected by the atmospheric channel.At the receiver,the signal power will be reduced,so the communication distance will be limited,which will affect the network scale and the scalability of the network.Therefore,the traditional networking technology in Mesh and Ad-hoc can not be moved to FSO network.Reasonable and effective network topology and good routing algorithm can make up for the defects of FSO single link characteristics,and also have a significant impact on the performance of communication system and the survival of the entire network.This paper first describes the background significance and key technologies of free space optical communication,then designs the node model,energy model and network model of FSO network for FSO network,and introduces the Tyson polygons and machine learning used in this paper.On this basis,this paper proposes a Topology Control Based Machine Learning Algorithm(TCBMLA)algorithm based on hierarchical distributed network model,combined with Delaunay triangulation(LDT)algorithm and machine learning,taking into account the mobility of cluster heads and lower nodes.The algorithm is called TC-ML algorithm.The topology control algorithm can not only create a topology with high connectivity and strong robustness,but also realize topology reconstruction.The topology control algorithm includes two algorithms.The first is called topology formation algorithm,which includes the algorithm of the upper layer topology formation and thealgorithm of the lower level topology formation.The algorithm of the upper topology formation mainly constructs the backbone network based on Voronoi diagram,and the algorithm of the lower level topology formation is mainly based on the average degree of the node predicted by machine learning as the reference topology.The second is called dynamic management algorithm,which reconfiguring the upper topology by exploiting the local dynamics of Tyson polygons and redistributing the underlying topology by using machine learning prediction algorithm.On this basis,the connection criteria of nodes are studied,including the selection weight of nodes and the maximum access degree of nodes.The derivation formula is given to avoid overloading of nodes and excessive energy consumption.In this paper,TC-ML algorithm is simulated and analyzed.The results show that TC-ML algorithm can form a FSO network with high connectivity,full coverage and automatic reconfigurable topology.