cs.NINov 13, 2026

PUMA-Former: Uncertainty-Aware Agentic Computation Offloading in MEO-LEO Satellite Networks With Incomplete Information

Authors: Yi Jing, Chunxiao Jiang, Jiawei Wang, Jiachen Sun

Organizations: Department of Electronic Engineering, Tsinghua University, Beijing 100084, China · State Key Laboratory of Space Network and Communications, Tsinghua University, Beijing 100084, China, and also with the Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing 100084, China

Abstract

Satellite edge computing distributes mobile workloads across moving processors whose service capacity must be coordinated over time. In medium Earth orbit (MEO)--low Earth orbit (LEO) networks, periodic reports leave CPU states stale or missing, making offloading sensitive to uncertain capacity and shared-destination congestion. We propose PUMA-Former, an uncertainty-aware agentic framework for cooperative offloading under incomplete information that balances completed work, service delay, and execution cost. A MEO coordinator maintains distinct-report memory and closes the loop between available evidence, capacity estimates, allocation commands, and execution feedback. Latent-delay attention weights plausible measurement-age hypotheses, and historical innovations refine CPU estimates around the latest received report. Queue-error-aware fusion incorporates auxiliary pressure while accounting for backlog uncertainty. The inferred resource moments condition a structured Transformer allocator for joint destination, partition, bandwidth, CPU, and power decisions. Bounded risk-aware calibration then adjusts workload splits to reflect uncertainty and shared destination load. In traffic-driven development evaluation, PUMA-Former increases admitted compute-work completion by 8.62%, reduces mean modeled service delay by 34.78%, and reduces mean episode-P95 delay by 71.08% relative to ATLA-adapt. It also improves completion and normalized energy over fixed-half round-robin offloading. Service gains over the learning baselines involve higher normalized energy.

Figures & tables

Explore similar work

CardsList
  1. Update for Decisions, Not Freshness: Goal-Oriented Status Updating and Selective Offloading at the Network Edge

    Sep 1, 2026Jianpeng Qi, Qiyang Zhang, Chao Liu +5Partially Observable RLCooperative MARL

  2. Multi-Turn Reasoning LLMs for Task Offloading in Mobile Edge Computing

    Apr 8, 2026Ning Yang, Chuangxin Cheng, Haijun ZhangEdge ComputingLLM Reasoning

  3. The Operable Pareto Front: Distilling Offline Search into Run-Time Control for Multi-Objective UAV Edge-Computing Scheduling

    Sep 16, 2026Qiao Liao, Zhiyong Feng, Bin Wu +1Trajectory OptimizationEdge Computing