Approach Toward Simulation Modeling With Reinforced Offloading for Heterogeneous IoT Terminals Under a Novel STECN

Satellite terrestrial edge computing network (STECN) has significant potential application deploying mobile edge computing (MEC) in offloading the computational tasks by the heterogeneous Internet of Thing (IoT) terminals under terrestrial network (TN). In this article, we present a methodological approach toward modeling the simulation environment for a novel multitier multiconstellation STECN where the IoT user equipment (UE) from both aviation space and ground will offload computing tasks into the edge satellite network for processing. We designed a network model, a communication and coverage time model and a computing model under the proposed STECN. We proposed two algorithms with three offloading schemes addressing variability in latency tolerance by heterogeneous UEs. We optimized the allocation of communication and computing resources by the satellites under the model by adopting a deep deterministic policy gradient (DDPG) algorithm with an actor-critic network for training and learning. We approached simulation modeling by designing and developing relevant modules, simulation architecture, and workflow. We incorporated techniques to fine tune the training system under specific evaluation matrices so that our simulation model can be followed by others in the domain.