Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Secure Resource Allocation via Constrained Deep Reinforcement Learning

Jianfei Sun, Qiang Gao, Cong Wu, Yuxian Li, Jiacheng Wang, Dusit Niyato

TL;DR

The paper tackles secure, efficient resource allocation in serverless multi-cloud edge environments under IoT/6G demands. It introduces SARMTO, a framework built on action-constrained DRL (AC-DQN) with an MDP formulation, adaptive security, and robust optimization techniques to jointly optimize task offloading, resources, and security. Key contributions include embedding feasibility constraints within the learning process, modeling security overhead, and demonstrating substantial improvements in system cost and energy efficiency across diverse scenarios. The findings suggest SARMTO enables scalable, secure resource management for heterogeneous edge-cloud ecosystems, with practical implications for next-generation IoT and edge applications. $C(\pi)=\alpha_1 T(\pi)+\alpha_2 E(\pi)$ and related security overheads underpin the optimization, highlighting the framework's balance between latency, energy, and protection.

Abstract

The proliferation of Internet of Things (IoT) devices and the advent of 6G technologies have introduced computationally intensive tasks that often surpass the processing capabilities of user devices. Efficient and secure resource allocation in serverless multi-cloud edge computing environments is essential for supporting these demands and advancing distributed computing. However, existing solutions frequently struggle with the complexity of multi-cloud infrastructures, robust security integration, and effective application of traditional deep reinforcement learning (DRL) techniques under system constraints. To address these challenges, we present SARMTO, a novel framework that integrates an action-constrained DRL model. SARMTO dynamically balances resource allocation, task offloading, security, and performance by utilizing a Markov decision process formulation, an adaptive security mechanism, and sophisticated optimization techniques. Extensive simulations across varying scenarios, including different task loads, data sizes, and MEC capacities, show that SARMTO consistently outperforms five baseline approaches, achieving up to a 40% reduction in system costs and a 41.5% improvement in energy efficiency over state-of-the-art methods. These enhancements highlight SARMTO's potential to revolutionize resource management in intricate distributed computing environments, opening the door to more efficient and secure IoT and edge computing applications.

Secure Resource Allocation via Constrained Deep Reinforcement Learning

TL;DR

The paper tackles secure, efficient resource allocation in serverless multi-cloud edge environments under IoT/6G demands. It introduces SARMTO, a framework built on action-constrained DRL (AC-DQN) with an MDP formulation, adaptive security, and robust optimization techniques to jointly optimize task offloading, resources, and security. Key contributions include embedding feasibility constraints within the learning process, modeling security overhead, and demonstrating substantial improvements in system cost and energy efficiency across diverse scenarios. The findings suggest SARMTO enables scalable, secure resource management for heterogeneous edge-cloud ecosystems, with practical implications for next-generation IoT and edge applications. and related security overheads underpin the optimization, highlighting the framework's balance between latency, energy, and protection.

Abstract

The proliferation of Internet of Things (IoT) devices and the advent of 6G technologies have introduced computationally intensive tasks that often surpass the processing capabilities of user devices. Efficient and secure resource allocation in serverless multi-cloud edge computing environments is essential for supporting these demands and advancing distributed computing. However, existing solutions frequently struggle with the complexity of multi-cloud infrastructures, robust security integration, and effective application of traditional deep reinforcement learning (DRL) techniques under system constraints. To address these challenges, we present SARMTO, a novel framework that integrates an action-constrained DRL model. SARMTO dynamically balances resource allocation, task offloading, security, and performance by utilizing a Markov decision process formulation, an adaptive security mechanism, and sophisticated optimization techniques. Extensive simulations across varying scenarios, including different task loads, data sizes, and MEC capacities, show that SARMTO consistently outperforms five baseline approaches, achieving up to a 40% reduction in system costs and a 41.5% improvement in energy efficiency over state-of-the-art methods. These enhancements highlight SARMTO's potential to revolutionize resource management in intricate distributed computing environments, opening the door to more efficient and secure IoT and edge computing applications.
Paper Structure (17 sections, 15 equations, 4 figures, 1 table, 2 algorithms)

This paper contains 17 sections, 15 equations, 4 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. System Model and Problem Formulation
  4. Design of SARMTO
  5. Overview
  6. MDP Formulation
  7. AC-DQN
  8. Security Mechanism Integration
  9. Adaptive Exploration Strategy
  10. Performance Optimization Techniques
  11. Experimental Results
  12. Evaluation Setup
  13. Overall Performance
  14. Impact of Average Data Size
  15. Energy Consumption Analysis
  16. ...and 2 more sections

Figures (4)

  • Figure 1: Serverless multi-cloud edge computing model
  • Figure 2: Performance under different MEC computational capacities, including system cost (a), task completion time (b), energy efficiency (c), and offloading rate (d)
  • Figure 3: Performance under different number of tasks, including system cost (a) and average delay (b)
  • Figure 4: Performance under different data sizes, including system cost (a) and energy consumption (b)