Lyapunov-Based Stochastic Resource Allocation for Integrated Satellite-Terrestrial Networks

The integration of a satellite network with a terrestrial network, supporting optimized network selection and service continuity, is essential to meet the requirements of sixth-generation (6 G) wireless networks. In a typical scenario with both terrestrial and satellite services, a subset, referred to as terrestrial users (t-users), can communicate only with the terrestrial base station (BS), while dual users (d-users) can transmit information to both the BS and an available LEO satellite. Thus, we focus on an integrated satellite-terrestrial uplink network, assuming that non-orthogonal multiple access (NOMA) is used in the terrestrial sub-network, which is motivated by the existence of non-line-of-sight channels between the BS and the users. We also investigate the use of two communication protocols for the satellite sub-network, considering both orthogonal multiple access (OMA) and NOMA. We define a utility function based on the achievable rates of the users and formulate an optimization problem to maximize the long-term utility of the network, taking into account constraints on the maximum capacity of the BS and the LEO satellite, as well as short- and long-term power consumption. We apply Lyapunov theory to decompose it into multiple single-time slot sub-problems, and finally transform it into convex form and solve it effectively. The network selection by the d-users as well as the optimal resource allocation for all users are also investigated. The simulation results show that d-users prefer to communicate with the LEO satellite rather than the BS. The probability of this selection is strongly influenced by the chosen communication scheme and the trajectory of the LEO satellite. Furthermore, there is an improvement in utility when the LEO satellite is close to its zenith position, while users achieve higher utility when the NOMA protocol is used in the satellite sub-network compared to the OMA counterpart.