AIGC-Assisted Digital Watermark Services in Low-Earth Orbit Satellite-Terrestrial Edge Networks

Abstract

Low Earth Orbit (LEO) satellite communication is a crucial component of future 6G communication networks, contributing to the development of an integrated satellite-terrestrial network. In the forthcoming satellite-to-ground network, the idle computational resources of LEO satellites can serve as edge servers, delivering intelligent task computation services to ground users. Existing research on satellite-to-ground computation primarily focuses on designing efficient task scheduling algorithms to provide straightforward computation services to ground users. This study aims to integrate satellite edge networks with Artificial Intelligence-Generated Content (AIGC) technology to offer personalized AIGC services to ground users, such as customized digital watermarking services. Firstly, we propose a satellite-to-ground edge network architecture, enabling bidirectional communication between visible LEO satellites and ground users. Each LEO satellite is equipped with intelligent algorithms supporting various AIGC-assisted digital watermarking technologies with different precision levels. Secondly, considering metrics like satellite visibility, satellite-to-ground communication stability, digital watermark quality, satellite-to-ground communication time, digital watermarking time, and ground user energy consumption, we construct an AIGC-assisted digital watermarking model based on the satellite-to-ground edge network. Finally, we introduce a reinforcement learning-based task scheduling algorithm to obtain an optimal strategy. Experimental results demonstrate that our approach effectively meets the watermark generation needs of ground users, achieving a well-balanced trade-off between generation time and user energy consumption. We anticipate that this work will provide an effective solution for the intelligent services in satellite-to-ground edge networks.