A Dynamic Beam Switching Algorithm for LEO Satellite Based on Communication Traffic Volume Prediction

This paper proposes a dynamic beam-switching algorithm that utilizes traffic volume prediction to address the high complexity of the current beam-switching algorithm and the wastage of onboard resources in low-orbit satellite constellations. Firstly, we model the signal intensity distribution of a single satellite coverage area. Then, considering the obvious spatiotemporal periodicity of the communication traffic volume, Convolutional Long Short Term Memory neural network (Conv-LSTM) is used to accurately analyze the regional communication traffic volume prediction. Finally, we define the importance of the beam service according to the signal intensity distribution and communication traffic volume, so as to select the beam that needs to be switched off. Based on the proposed algorithm, when the service area of the satellite is switched, the algorithm can preset the beam state at the next moment according to the prediction result, thereby reducing the number of beam adjustments. The simulation results show that the algorithm takes into account the signal blind area of the beam coverage area and the ground communication traffic, effectively reducing the waste of beam resources and computational complexity.