Efficient on-board signaling processing for satellite-terrestrial integrated core networks

Integrating Low-Earth Orbit (LEO) satellite constellations with terrestrial mobile networks can achieve global coverage and complement terrestrial networks. The inherent mobility of satellites induces frequent handovers of User Equipment (UE), generating massive signaling. Coupled with limited satellite resources, the Network Functions (NFs) deployed on satellites cannot process these signaling promptly, leading to increased queuing time. Additionally, the movement of on-board NFs increases the distance to UE, extending propagation delay. Extended Procedure Completion Time (PCT) of control plane procedures degrades user plane Quality of Service (QoS). To address the above challenges, we propose a satellite-terrestrial integrated core network architecture to enhance signaling processing performance. Firstly, we redesign the control plane network functions and introduce a Satellite-Ground Synergy Method (SGSM), categorizing signaling into time-sensitive and time-tolerant types. The former is processed on-board, while the latter is handled terrestrially, utilizing a designed UE context synchronization mechanism. Furthermore, migration is employed to counteract the movement. We devise a migration procedure to reduce transferred data during migration. Moreover, we model instance migration as a Markov Decision Process and proposed an online NFs migration algorithm based on deep reinforcement learning to determine migration timing and target satellites. Extensive experiments demonstrate that the proposed methods significantly reduce queuing time and the volume of transferred data, while also exhibiting superior performance in terms of propagation delay and the frequency of migrations.