{"title":"Dependent task offloading in multi-access edge computing: A GCN augmented deep reinforcement learning approach","authors":"Liqiong Chen, Xinyuan Yang, Huaiying Sun, Xiuchao Yu, Kaiwen Zhi","doi":"10.1016/j.comnet.2025.111445","DOIUrl":null,"url":null,"abstract":"<div><div>Multi-access edge computing (MEC) is a promising distributed computing paradigm that reduces the delayed energy cost (DECC) of users by offloading user-generated application tasks to the network edge. Most of the user-generated tasks contain a series of subtasks with dependencies, how to effectively offload these interdependent tasks and reduce DECC is a key issue. In addition, most existing learning-based approaches are inadequate in dynamically modeling MEC environments and cannot effectively characterize heterogeneous MEC environments. To this end, this paper models the multiuser task offloading problem as a Markov Decision Process (MDP). First, to address the challenges of heterogeneous MEC environments, we propose a multi-dependent task offloading algorithm with state embeddings. This algorithm uses commonality and dissimilarity components to capture the interactions between user and the MEC environment, providing robust state representation. Secondly, we introduce the strategy gradient theorem of the Stackelberg game to optimize the offloading decision. Finally, extensive experiments show that our proposed method significantly reduces DECC compared to existing methods.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"269 ","pages":"Article 111445"},"PeriodicalIF":4.4000,"publicationDate":"2025-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Computer Networks","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1389128625004128","RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, HARDWARE & ARCHITECTURE","Score":null,"Total":0}
引用次数: 0
Abstract
Multi-access edge computing (MEC) is a promising distributed computing paradigm that reduces the delayed energy cost (DECC) of users by offloading user-generated application tasks to the network edge. Most of the user-generated tasks contain a series of subtasks with dependencies, how to effectively offload these interdependent tasks and reduce DECC is a key issue. In addition, most existing learning-based approaches are inadequate in dynamically modeling MEC environments and cannot effectively characterize heterogeneous MEC environments. To this end, this paper models the multiuser task offloading problem as a Markov Decision Process (MDP). First, to address the challenges of heterogeneous MEC environments, we propose a multi-dependent task offloading algorithm with state embeddings. This algorithm uses commonality and dissimilarity components to capture the interactions between user and the MEC environment, providing robust state representation. Secondly, we introduce the strategy gradient theorem of the Stackelberg game to optimize the offloading decision. Finally, extensive experiments show that our proposed method significantly reduces DECC compared to existing methods.
期刊介绍:
Computer Networks is an international, archival journal providing a publication vehicle for complete coverage of all topics of interest to those involved in the computer communications networking area. The audience includes researchers, managers and operators of networks as well as designers and implementors. The Editorial Board will consider any material for publication that is of interest to those groups.