Auto-Scaling Service Strategy for Distributed Satellite System

In this paper, we investigate a scenario where a distributed satellite system (DSS) serves as a distributed computing platform within the space-terrestrial integrated network (STIN), catering to ground users. To enhance the scalability of the distributed computing system and adapt to variations in user request traffic, we propose an auto-scaling service strategy based on traffic prediction. The prediction is implemented using the long short-term memory recurrent neural network (LSTM-RNN) method. Numerical results are presented to demonstrate the superiority of our proposed traffic prediction algorithm compared to existing approaches, while experimental findings confirm that our auto-scaling service strategy effectively enhances system performance in terms of service quality.