Research On Dynamic Time Slot Allocation And Network Planning For Low-orbit Satellite Data Link

The low-orbit satellite data link adds low-orbit satellite nodes on the basis of the Link-16 data link to relay and forward data link messages,thereby increasing its transmission distance and adapting to the mode of cross-domain joint operations and increasing defensive distance in modern warfare.Since LEO satellite datalink system is based on Time Division Multiple Access(TDMA),the allocation of time slot resources has a significant impact on its network throughput and other indicators,so dynamic time slot allocation algorithm has always been the focus of research on datalink system.In addition,network planning is the key of LEO satellite datalink network management and the basis of datalink networking.A reasonable network planning can guarantee the reliability and security of the network to a great extent.Therefore,the design of dynamic time slot allocation schemes with high throughput and efficient network planning software can effectively improve the performance of LEO satellite data link networks and the efficiency of network planning.This thesis mainly studies the dynamic time slot allocation algorithm of the low-orbit satellite data link and designs and develops the network planning software.The research contents are as follows:(1)In the time slot prediction stage,this thesis proposes a Sparse Adaboost-ESN(SA-ESN)time slot prediction algorithm based on Echo State Network(ESN)with sparsification.For the Network Participation Group(NPG)with nonlinear characteristics of time slot data,this thesis uses the SA-ESN time slot prediction algorithm for its time slot prediction.The algorithm uses the sparse Adaboost algorithm as an integration framework and ESN as a weak learner,which can reduce the integration scale of the model without affecting the prediction accuracy of the algorithm.The simulation results show that compared with Adaboost-ANN and ESN machine learning algorithms,the SA-ESN time slot prediction algorithm proposed in this thesis has a smaller root mean square error,relatively stable prediction effect,and a more significant improvement in prediction accuracy.(2)In the time slot allocation phase,a time slot allocation algorithm based on Improved Graph Coloring Theory(IGCT)is proposed in this thesis.The IGCT algorithm adds constraints to the traditional graph coloring problem to achieve the effect that all vertices connected to any one vertex are also colored differently with the minimum number of colorings,based on the fact that two vertices on the same edge are colored differently.Using IGCT algorithm for time slot allocation,it is possible to allocate different time slots for nodes within two hops and the same time slots for nodes beyond two hops when the network topology is known.Thus,the utilization of time slots can be improved and the throughput of the network can be increased.Simulation analysis is performed using OMNe T++ network simulation software.The results show that the IGCT time slot allocation algorithm proposed in this thesis improves the average throughput of nodes at different network sizes compared with the traditional Five Phase Reservation Protocol(FPRP)and Dynamic TDMA Protocol(D-TDMA).(3)To address the problems of long network planning cycle and large computation for LEO satellite data chain,this thesis carries out the design of network planning process and the development of network planning software.Firstly,on the basis of the existing data link network planning system,we summarized the network planning elements and designed a combat mission-based network planning process.Then,based on the network planning process,we designed the network planning software modules and developed each module in Qt creator.The network planning software developed in this thesis can be used for network planning of LEO satellite data links to reduce the workload of network planners and shorten the network planning cycle.