Enhancing Physical Layer Communication Security through Generative AI with Mixture of Experts

TL;DR

The paper tackles physical layer security in wireless networks by leveraging Generative AI (GAI) and addressing its limitations with Mixture of Experts (MoE). It develops an MoE-enabled GAI framework for network optimization problems and demonstrates its effectiveness through a cooperative friendly jamming case study, achieving superior secrecy rate $SR$ and secure energy efficiency $S_e$ compared to baselines. Key contributions include a taxonomy of GAI applications for confidentiality, integrity, and availability, a MoE-based security framework with a defined architecture and workflow, and a case study showing improved convergence and multi-metric tradeoffs with MoE. The work suggests practical avenues for zero-trust security, real-time anomaly detection, and privacy-preserving MoE strategies in future wireless networks.

Abstract

AI technologies have become more widely adopted in wireless communications. As an emerging type of AI technologies, the generative artificial intelligence (GAI) gains lots of attention in communication security. Due to its powerful learning ability, GAI models have demonstrated superiority over conventional AI methods. However, GAI still has several limitations, including high computational complexity and limited adaptability. Mixture of Experts (MoE), which uses multiple expert models for prediction through a gate mechanism, proposes possible solutions. Firstly, we review GAI model's applications in physical layer communication security, discuss limitations, and explore how MoE can help GAI overcome these limitations. Furthermore, we propose an MoE-enabled GAI framework for network optimization problems for communication security. To demonstrate the framework's effectiveness, we provide a case study in a cooperative friendly jamming scenario. The experimental results show that the MoE-enabled framework effectively assists the GAI algorithm, solves its limitations, and enhances communication security.

Figures (4)

• Figure 1: The overall of GAI for physical layer communication security. Part A illustrates the structure of GAI models. Part B presents security issues faced by physical layer communications. In Part C, we introduce three examples based on GAI for communication security. These three models are based on VAE erdemir2022privacy, DM wu2023cddm, and GAN cai2020spectrum, respectively. Part D demonstrates important physical layer communication attack scenarios, including eavesdropping, jamming, and anomaly detection.
• Figure 2: The structure of MoE. In Part A, we illustrate the main components and working mechanism of the MoE framework. Part B presents an example of using the MoE framework to learn the different characteristic distributions of multiple data. Part C demonstrates how to learn different features of data for multiple tasks via MoE structure.
• Figure 3: The workflow of the proposed framework. In Part A, the proposed framework solves optimization problems through reinforcement learning following 4 steps: 1). Task Setting; 2). MoE Learning; 3). Denoising Diffusion; 4). Action Evaluation. Part B illustrates the cooperative friendly jamming optimization problem considered in Section \ref{['Case']}. The figure introduces the detail scenario and optimization problem settings including optimization objectives, constraints, etc.
• Figure 4: Performance comparison of three methods in cooperative friendly jamming scenario.