Satellite Edge Computing With Collaborative Computation Offloading: An Intelligent Deep Deterministic Policy Gradient Approach

Enabling a satellite network with edge computing capabilities can complement the advantages further of single terrestrial and provide users full range service. Satellite is potentially indispensable technology for future satellite-terrestrial integrated networks. In this article, three-tier architecture consisting terminal–satellite–cloud proposed, where tasks be processed at three planes intersatellites cooperate to achieve on-board load balancing. Facing varying random task queues different service requirements, we formulate objective problem minimizing system energy consumption under delay resource constraints, jointly optimize offloading decision, communication, allocation variables. Moreover, distribution resources based on reservation mechanism ensure stability link reliability computation process. To adapt dynamic environment, propose an intelligent scheme deep deterministic policy gradient (DDPG) algorithm, which consists several neural networks (DNNs) output both discrete continuous Additionally, by setting selection process legal actions, simultaneous decisions locations allocating multitask concurrency realized. The simulation results show that proposed effectively reduce total ensuring completed demand, outperform benchmark algorithms.