Lagrangian-Based Energy-Efficient Route Learning Considering Expected Guaranteed Delay for Satellite Network

With the rapid development of satellite network, route problem has gained much attention in these years to ensure the service quality. However, due to the uncertain transmission requirements, limited energy generation and battery capacity, the optimal route path for the satellite network is non-trivial to be searched. We consider this important problem in this paper and make the following contributions. First, this problem is formulated as a constrained stochastic shortest path model to capture the uncertain transmission requirements. Besides reducing the energy consumption during routing, this model incorporates the expected guaranteed delay constraint to ensure service quality. Second, a Lagrangian-based distributed route learning algorithm is developed to search the optimal route path. By Lagrangian relaxation, the proposed model can be transformed into a bi-level optimization model. The upper-level searches the optimal multiplier while the lower-level makes distributed forward decisions among satellites. Third, the performance improvement of the proposed route algorithm is theoretically proved to ensure the routing convergence. The validations of the energy saving, the end-to-end delay and the convergence of the proposed method are systematically investigated via numerical experiments.