Integrating artificial intelligence (AI) with mobile edge computing, edge intelligence (EI) has emerged as a new paradigm for 5G and beyond 5G (B5G) systems. The integration of EI and heterogeneous networks (e.g., mobile and wireless local area networks) also raises new concerns about security and privacy. This article examines two important security aspects of EI-empowered, heterogeneous, B5G networks, that is, authentication and trust-evaluation-based compromised user equipment (UE) detection. Technical challenges are discussed. A new edge-computing-enabled, unified authentication framework is developed, which authenticates UEs consistently via heterogeneous networks with UEs' privacy preserved. A new trust-evaluation-based compromised UE detection method is developed based on reinforcement learning to prevent compromised UEs from launching internal attacks. Case studies show that the new frameworks improve the authentication efficiency and detection rate of compromised UEs, as compared to technologies specified by current 5G standards. The frameworks have great potential to secure B5G in future heterogeneous network environments.

中文翻译：

---

5G 之外的异构系统的边缘智能授权、统一身份验证和信任评估


将人工智能 (AI) 与移动边缘计算相结合，边缘智能 (EI) 已成为 5G 及超 5G (B5G) 系统的新范例。 EI 和异构网络（例如移动和无线局域网）的集成也引发了对安全和隐私的新担忧。本文研究了 EI 授权的异构 B5G 网络的两个重要安全方面，即基于身份验证和信任评估的受损用户设备 (UE) 检测。讨论了技术挑战。开发了一种新的支持边缘计算的统一身份验证框架，该框架可以通过异构网络对 UE 进行一致的身份验证，同时保护 UE 的隐私。基于强化学习，开发了一种新的基于信任评估的受损UE检测方法，以防止受损UE发起内部攻击。案例研究表明，与当前 5G 标准规定的技术相比，新框架提高了受感染 UE 的身份验证效率和检测率。这些框架在未来异构网络环境中保护 B5G 方面具有巨大潜力。