ACE-PPO and MDE-IDS: a novel approach for efficiency and security in 5G optical networks

Abstract This paper proposes a novel dual-strategy intelligent framework to enhance the efficiency and security of high-speed 5G optical networks. The framework, labeled as attention and curriculum learning enhanced proximal policy optimization (ACE-PPO), is introduced for efficiency optimization. It manages optical signal processing beamforming and power control by focusing on key features like user locations and channel conditions, improving spectral efficiency and signal quality without requiring the overheads of explicit channel state information (CSI). For security, an intrusion detection system (MDE-IDS) based on a mutual distillation ensemble model is proposed. It synergistically combines a temporal convolutional network, a convolutional neural network, and a transformer, using a cross-attention gate and mutual distillation to boost robustness against sophisticated cyber threats in a 5G optical network. Simulations show ACE-PPO outperforms baselines, achieving a 14 % higher signal to interference plus noise ratio (SINR) and 5 % lower power consumption, while MDE-IDS provides higher detection accuracy. The framework forms a holistic approach for intelligent and resilient 5G optical networks.