INTELLIGENT RESOURCE ALLOCATION IN 6G OFDMANETWORKS: PROXIMAL POLICY OPTIMIZATION APPROACH WITH MULTI-OBJECTIVE OPTIMIZATION

Abstract

The transition to sixth-generation (6G) wireless networks brings forth a complex landscape of resource management challenges, driven by demanding applications like extended reality and holographic communications. These applications require data rates exceeding 1 T bps and ultra-low latency, pushing current Orthogonal Frequency Division Multiple Access (OFDMA) systems to their limits. The joint optimization of power allocation and resource block assignment is an NP-hard problem that defies traditional computational methods. This paper proposes a novel framework utilizing Deep Reinforcement Learning (DRL), specifically

Proximal Policy Optimization (PPO), to autonomously manage resources. By integrating a multi-objective reward function, our PPO-DRL agent balances energy efficiency with strict Quality of Service (QoS) requirements. Simulation results demonstrate that our approach achieves a 446% improvement in energy efficiency over Lagrangian methods and maintains QoS satisfaction above 96%, all while operating with a decision latency below 1 millisecond.