Age-Critical Joint Communication and Computation Offloading for Satellite-Integrated Internet

The emerging sixth-generation (6G) networks are expected to deliver ubiquitous intelligent services through satellite-integrated Internet, addressing the stringent demands of computation-intensive and delay-sensitive applications in extreme and infrastructure-sparse environments. Existing offloading approaches often fail to achieve age-critical resource allocation in dynamic satellite-integrated Internet, potentially leading to outdated information and inefficient resource utilization. To address this, we propose an age-critical joint communication and computation offloading scheme, where low Earth orbit satellites cover several base stations and many ground users. Then, we formulate a comprehensive optimization problem, which aims to minimize the average peak age of information (PAoI), subject to energy consumption constraints. We propose an online Lyapunov-enhanced decoupled actor-critic (LDA) scheme. Specifically, our LDA scheme decomposes the original long-term problem into manageable per-frame subproblems. It employs a deep neural network integrated with an order-preserving quantization approach to effectively decouple offloading decisions, and utilizes low-complexity iterative algorithms for resource allocation. Extensive simulations demonstrate that our LDA scheme substantially reduces the average PAoI compared to the related deep reinforcement learning schemes, while satisfying energy consumption constraints and reducing up to 25% complexity.