Large Timescale Optimization for Communications Over Aerial Ad Hoc Networks With Predetermined Trajectories

This paper studies a delay-tolerant data transportation problem over an aerial ad hoc network where the aerial nodes have predetermined trajectories. The challenge is how to exploit the large-scale channel information predicted from the predetermined trajectories of the aerial nodes to optimize for the communication strategy in a distributive way. The objective is to minimize both the communication energy and the communication time to control the interference leakage to the ground. Most existing approaches for aerial network communications require intensive centralized coordination, but the trajectory information may not be globally available. To tackle these issues, this paper develops a large timescale two-layer optimization strategy using a game theoretical approach. In the inner layer, a mixed timescale optimization on the power allocation and transmission timing is formulated, which is converted into a one-parameter optimization with an optimality guarantee for a deterministic proxy of the original problem. In the outer layer, a handover time game is formulated, and a neighbor coordinate response strategy based on local information exchange is developed, demonstrating rapid convergence and near-global optimality in simulations. Numerical experiments demonstrate that, under large timescale optimization, an order of magnitude of cost saving can be achieved.