Adaptive Task Offloading and Resource Allocation for Tasks With Time-Varying Statistical Characteristics in MEC Systems

Considering the random access behavior of mobile devices (MDs), heterogeneous service requirements, and dynamic completion or discarding of tasks, the statistical characteristics of tasks to be scheduled, transmitted, and computed vary over time in practical multiaccess edge computing (MEC) systems. Therefore, it is necessary to design a dynamic task offloading and resource allocation (TORA) strategy to adaptively match these time-varying task characteristics. To achieve the goal, we first formulate the dynamic TORA problem as a Markov decision process (MDP) with a time-varying extension of state space and action space, which cannot be effectively solved by conventional deep reinforcement learning (DRL) algorithms. To address this challenge, we propose a general state–action space adaptive (SASA) DRL framework by exploiting the advantages of the Transformer architecture and its multihead attention (MHA) mechanism. This framework facilitates the integration of available actor–critic DRL algorithms to efficiently solve MDPs with time-varying state and action spaces. Based on the proposed SASA DRL framework, we further develop the SASA-based TORA algorithm, referred to as SASA-TORA, which is adaptable to not only dynamic network conditions but also time-varying statistical characteristics of tasks. Simulation results demonstrate the superiority of SASA-TORA over baseline algorithms and highlight the limitations of conventional DRL algorithms in handling MDPs with time-varying state and action spaces.

• Publication year

  2026

• Venue

  IEEE Internet of Things Journal

• Publication date

  2026-02-01

• Fields of study

  Computer Science, Engineering

• Identifiers
  DOI 10.1109/JIOT.2025.3633937
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• AoI and Energy-Driven Dynamic Cache Updates for Wireless Edge Networks

  2025cited by this paper
• Optimizing task offloading with metaheuristic algorithms across cloud, fog, and edge computing networks: A comprehensive survey and state-of-the-art schemes

  2025cited by this paper
• An advanced deep reinforcement learning algorithm for three-layer D2D-edge-cloud computing architecture for efficient task offloading in the Internet of Things

  2024cited by this paper
• A novel offloading strategy for multi-user optimization in blockchain-enabled Mobile Edge Computing networks for improved Internet of Things performance

  2024cited by this paper
• Multi-Objective Secure Task Offloading Strategy for Blockchain-Enabled IoV-MEC Systems: A Double Deep Q-Network Approach

  2024cited by this paper
• Achieving Fast Environment Adaptation of DRL-Based Computation Offloading in Mobile Edge Computing

  2024influential reference
• Cooperative Partial Task Offloading and Resource Allocation for IIoT Based on Decentralized Multiagent Deep Reinforcement Learning

  2024cited by this paper
• Decentralized Scheduling for Concurrent Tasks in Mobile Edge Computing via Deep Reinforcement Learning

  2024influential reference
• Deep Learning-Assisted Online Task Offloading for Latency Minimization in Heterogeneous Mobile Edge

  2024influential reference
• Fully Decentralized Task Offloading in Multi-Access Edge Computing Systems

  2024cited by this paper
• Joint Task Offloading and Resource Allocation for Vehicular Edge Computing With Result Feedback Delay

  2023influential reference
• Offloading Using Traditional Optimization and Machine Learning in Federated Cloud–Edge–Fog Systems: A Survey

  2023cited by this paper
• Deep Reinforcement Learning Based Cooperative Partial Task Offloading and Resource Allocation for IIoT Applications

  2023influential reference
• Large-Scale Computation Offloading Using a Multi-Agent Reinforcement Learning in Heterogeneous Multi-Access Edge Computing

  2023influential reference
• Reinforcement Learning-Based Deadline and Battery-Aware Offloading in Smart Farm IoT-UAV Networks

  2022cited by this paper
• Wireless Powered Mobile Edge Computing: Dynamic Resource Allocation and Throughput Maximization

  2022cited by this paper
• Machine and Deep Learning for Resource Allocation in Multi-Access Edge Computing: A Survey

  2022cited by this paper
• MLP-Mixer: An all-MLP Architecture for Vision

  2021cited by this paper
• Resource Scheduling in Edge Computing: A Survey

  2021cited by this paper
• Transformer in Transformer

  2021cited by this paper
• Delay-Limited Computation Offloading for MEC-Assisted Mobile Blockchain Networks

  2021cited by this paper
• Collaborative Cloud-Edge-End Task Offloading in Mobile-Edge Computing Networks With Limited Communication Capability

  2021cited by this paper
• Optimal Status Update for Caching Enabled IoT Networks: A Dueling Deep R-Network Approach

  2021cited by this paper
• Joint Optimization Strategy of Computation Offloading and Resource Allocation in Multi-Access Edge Computing Environment

  2020cited by this paper
• Optimal Mobile Computation Offloading with Hard Deadline Constraints

  2020cited by this paper
• UAV-to-Device Underlay Communications: Age of Information Minimization by Multi-Agent Deep Reinforcement Learning

  2020cited by this paper
• Mobility-Aware Multi-User Offloading Optimization for Mobile Edge Computing

  2020cited by this paper
• Dynamic Task Offloading and Resource Allocation for Mobile-Edge Computing in Dense Cloud RAN

  2020influential reference
• A Game-Based Computation Offloading Method in Vehicular Multiaccess Edge Computing Networks

  2020cited by this paper
• Dynamic Service Migration in Mobile Edge Computing Based on Markov Decision Process

  2019cited by this paper
• Deep Reinforcement Learning for Offloading and Resource Allocation in Vehicle Edge Computing and Networks

  2019influential reference
• Dynamic Task Offloading and Resource Allocation for Ultra-Reliable Low-Latency Edge Computing

  2018cited by this paper
• Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor

  2018influential reference
• Adaptive Federated Learning in Resource Constrained Edge Computing Systems

  2018cited by this paper
• Resource Management in Fog/Edge Computing

  2018cited by this paper
• EMM: Energy-Aware Mobility Management for Mobile Edge Computing in Ultra Dense Networks

  2017cited by this paper
• A Survey on Mobile Edge Computing: The Communication Perspective

  2017cited by this paper
• Proximal Policy Optimization Algorithms

  2017influential reference
• Attention is All you Need

  2017cited by this paper
• Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks

  2016cited by this paper
• Deep Residual Learning for Image Recognition

  2015cited by this paper
• Continuous control with deep reinforcement learning

  2015influential reference
• Internet of Things: A Survey on Enabling Technologies, Protocols, and Applications

  2015cited by this paper
• Quantized Convolutional Neural Networks for Mobile Devices

  2015cited by this paper
• Human-level control through deep reinforcement learning

  2015influential reference
• Sequence to Sequence Learning with Neural Networks

  2014cited by this paper
• Reinforcement Learning: An Introduction

  1998cited by this paper

• No citing papers are available for this paper.