The conventional device authentication of wireless networks usually relies on a security server and centralized process, leading to long latency and risk of single-point of failure. While these challenges might be mitigated by collaborative authentication schemes, their performance remains limited by the rigidity of data collection and aggregated result. They also tend to ignore attacker localization in the collaborative authentication process. To overcome these challenges, a novel collaborative authentication scheme is proposed, where multiple edge devices act as cooperative peers to assist the service provider in distributively authenticating its users by estimating their received signal strength indicator (RSSI) and mobility trajectory (TRA). More explicitly, a distributed learning-based collaborative authentication algorithm is conceived, where the cooperative peers update their authentication models locally, thus the network congestion and response time remain low. Moreover, a situation-aware secure group update algorithm is proposed for autonomously refreshing the set of cooperative peers in the dynamic environment. We also develop an algorithm for localizing a malicious user by the cooperative peers once it is identified. The simulation results demonstrate that the proposed scheme is eminently suitable for both indoor and outdoor communication scenarios, and outperforms some existing benchmark schemes.

翻译：传统无线网络中的设备认证通常依赖安全服务器与集中式流程，这导致长时延与单点故障风险。虽然协作认证方案可缓解这些问题，但其性能仍受限于数据采集与聚合结果的僵化性，且往往忽略协作认证过程中的攻击者定位。为克服上述挑战，本文提出一种新型协作认证方案，其中多个边缘设备作为协作对等节点，通过估计用户接收信号强度指示（RSSI）与移动轨迹（TRA），以分布式方式协助服务提供商完成用户认证。具体而言，本文设计了一种基于分布式学习的协作认证算法，协作对等节点可在本地更新其认证模型，从而保持低网络拥塞与低响应时间。此外，针对动态环境，本文提出一种态势感知的安全组更新算法，用于自主刷新协作对等节点集合。我们还开发了一种算法，使得协作对等节点在识别恶意用户后能够对其进行定位。仿真结果表明，所提方案在室内外通信场景中均具有高度适用性，且优于部分现有基准方案。