Lyapunov-Based Network Slicing and Resource Optimization in Air-Ground Integrated Networks

The air-ground integrated network (AGIN) architecture is the key support for future 6 G to realize ubiquitous network connectivity. For the diversity and heterogeneity of the AGIN, we need to divide the slicing resources for different services. To cope with the suddenness and uncertainty of business arrivals, a flexible resource slice management solution is required. This paper models the joint optimization problem of service request scheduling, network slicing, and resource allocation as a multi-stage stochastic mixed-integer nonlinear programming (MINLP) problem. The goal is to maximize the system utility under long-term user data queue stability constraints and minimize the number of re-slicing times by rational resource allocation when service requests change dynamically. In order to solve the coupling problem of service request scheduling, network slicing, and resource allocation in different time frame decisions, we propose an online control algorithm named ORSRM. It first applies Lyapunov optimization to decouple the MINLP problem into a series of per time slot deterministic subproblems. Then a clustering algorithm is used to determine the UAV location and the service request scheduling decisions are generated based on deep reinforcement learning (DRL). Finally, the per time slot system utility is maximized by optimizing the upper bound of the Lyapunov drift plus penalty function. Simulation results show that the proposed algorithm can ensure system stability while meeting user needs and try to avoid network re-slicing caused by changes in service requirements as much as possible.