DRL-Based URLLC-Constraint and Energy-Efficient Task Offloading for Internet of Health Things.

Area of Science:

• Telecommunications
• Health Informatics
• Computer Engineering

Background:

• The Internet of Health Things (IoHT) paradigm offers e-Health solutions by offloading tasks to edge servers.
• Existing computation offloading methods face challenges in meeting ultra-reliable and low-latency communication (URLLC) demands for IoHT.

Purpose of the Study:

• To propose a novel Multi-Agent Soft-Actor-Critic-discrete based URLLC-constrained task offloading and resource allocation (MASACDUA) scheme.
• To maximize throughput and minimize power consumption in IoHT systems under long-term URLLC constraints.

Main Methods:

• Formulating URLLC constraints using extreme value theory.
• Employing Lyapunov optimization to decompose the problem into task offloading and resource allocation.
• Utilizing MASAC-discrete for task offloading and a queue backlog-aware algorithm for resource allocation.

Main Results:

• The MASACDUA scheme demonstrates superior performance compared to traditional Deep Reinforcement Learning (DRL) algorithms.
• Achieves significant improvements in delay and bound violation probability under various IoHT conditions.
• Effectively balances throughput maximization and power consumption minimization.

Conclusions:

• The proposed MASACDUA scheme effectively addresses the challenges of task offloading in URLLC-constrained IoHT environments.
• Offers a robust solution for enhancing the performance and reliability of e-Health applications.
• Provides a foundation for future research in intelligent resource management for IoHT.