Goodput maximization for large language model edge inference: A two-phase maskable PPO approach
Author Identifier (ORCID)
Abstract
This letter presents a novel two-phase maskable proximal policy optimization (TP-MPPO) algorithm, which maximizes the system goodput counting request throughput with strict service level objective (SLO) compliance for large language model (LLM) inference services in wireless edge networks. In the first phase of TP-MPPO, we optimize the task offloading decisions by MPPO with action masking mechanism, effectively avoiding exploring invalid actions and reducing the action space. In the second phase, closed-form solutions are derived for uplink bandwidth allocation; a greedy algorithm is designed for downlink bandwidth allocation to provide immediate rewards for the MPPO in the next round. The two stages alternate till convergence. Simulation results demonstrate that TP-MPPO can improve the system reward by 33.3%-87.5% compared to its benchmarks and achieve the highest goodput.
Keywords
edge inference, large language model, resource allocation, task offloading
Document Type
Journal Article
Date of Publication
1-1-2026
E-ISSN
21622345
ISSN
21622337
Volume
15
Publication Title
IEEE Wireless Communications Letters
Publisher
IEEE
School
School of Engineering
Funding Information
Shanghai Municipal Science and Technology Commission Foundation (Grant Number: 25DP1500300 and 24DP1501001)
Copyright
subscription content
First Page
4400
Last Page
4404
Recommended Citation
Chen, X., Zhang, Q., Ni, W., Zhang, S., & Sun, Y. (2026). Goodput maximization for large language model edge inference: A two-phase maskable PPO approach. IEEE Wireless Communications Letters, 15, 4400–4404. https://doi.org/10.1109/LWC.2026.3718001