Iterative partial computation offloading algorithm for LEO satellite edge computing served by MEO satellite

This paper studies the energy-efficient optimization problem for low earth orbit (LEO) satellite edge computing served by medium earth orbit (MEO) satellite when partial computation offloading strategy is adopted. An optimization model with the goal of minimizing the total energy consumption of LEO satellites is established, with constraints of task tolerable delay, transmit power, and CPU-cycle frequency, forming a challenging mixed-integer and nonlinear problem. Based on the non-convexity of the optimization problem and the coupling variables, the problem is decomposed into three manageable subproblems as local computing subproblem, task offloading subproblem, and resource allocation subproblem, which are solved using analytical method, convex optimization, and Lagrange dual decomposition method, respectively. Based on the solution of the subproblems, alternative iterations are performed to converge to the optimal value. The simulation results show that when using the partial computation offloading optimization algorithm, the total energy consumption required for LEO satellites is significantly lower than that of local computing and full offloading strategies.