کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

Proximal Policy Optimization for Tri-Objective Hybrid Beamforming in BS-Only THz Integrated Sensing and Communication Systems

نویسندگان :

Alireza Mard Shoorijeh¹ Khashayar Saremi Tabrizi² Fatemeh Alia³ Bahareh Akhbari⁴

1- دانشگاه صنعتی خواجه نصیرالدین طوسی 2- دانشگاه صنعتی خواجه نصیرالدین طوسی 3- دانشگاه صنعتی خواجه نصیرالدین طوسی 4- دانشگاه صنعتی خواجه نصیرالدین طوسی

کلمات کلیدی :

Terahertz Communications،Integrated Sensing and Communications (ISAC)،Hybrid Beamforming،Deep Rein- forcement Learning (DRL)،Proximal Policy Optimization (PPO)،Energy Efficiency (EE)،Multi-Objective Optimization

چکیده :

This study proposes a Deep Reinforcement Learning (DRL) framework for joint Hybrid Multi-Beam Beamforming (BF) and power allocation in a highly constrained Base Station (BS)-only Terahertz (THz) Integrated Sensing and Communication (ISAC) small cell, addressing the critical 6G mandate for integrated high-rate services. Moving beyond the prevalent reliance on external reflective devices (IRS/STAR-RIS) to mitigate severe path loss, our core objective is to achieve a balanced, long-term trade-off among three competing performance metrics: Communication Sum-Rate (Rsum), Sensing Quality (SNRSense), and Energy Efficiency (EE). The complex optimization problem, featuring multi-class mobile users, is modeled as a Markov Decision Process (MDP) and solved using the stable Proximal Policy Optimization (PPO) algorithm and its distributed variant, DPPO. Analysis confirms that the proposed PPO/DPPO methodology learns an optimal policy for dynamic resource sharing, achieving significant performance gains across all three objectives compared to classical hybrid beamforming techniques and simple rule-based baselines, with DPPO demonstrating accelerated convergence vital for adapting to the volatile THz channel environment.