EAGER: Detection and Mitigation of Pilot Contamination Attacks and Related Issues in Massive MIMO Systems

移动数据流量继续以指数级速度增长。为了应对这种数据挑战，最近提出了大规模MIMO(多输入多输出)系统技术，其中基站使用大量的天线，允许同时为多个用户服务。它被认为是未来5G无线系统的关键推动因素之一。虽然最近的原型已经证明了它的可行性，但在大规模MIMO得以部署之前，仍有许多重大的研究挑战需要解决。大规模MIMO的成功运行关键依赖于对基站和终端用户之间的信道状态信息的了解。实际上，这将在训练阶段期间获得，其中用户向基站发送单独的导频信号。这一阶段具有挑战性，因为大量终端用户导致导频重复使用，导致导频污染，并且易受恶意窃听者的攻击，恶意窃听者可能通过传输相同的导频信号来欺骗合法用户。该项目的重点是检测和防御飞行员污染攻击的方法。本文研究了检测和防御来自主动窃听者和欺骗中继器的飞行员污染攻击的创新方法。引入的一个关键的变革性想法是合法用户对导频序列的自我污染，以便于检测活跃的窃听者。本项目在以下研究推进的背景下探索了这一想法的各种后果。(1)主动窃听者对飞行员污染攻击的检测：假设已知训练序列集(而不是其他任何知识)，则可以检测到是否有一个或多个训练序列受到攻击。正在开发基于数据关联函数的源枚举方法。(2)联合获取合法用户和窃听者的信道状态信息，以促进有效的波束成形设计，以增强合法用户的接收，同时降低窃听者的接收。(3)通过欺骗中继攻击检测和缓解主动窃听，其中欺骗中继以全双工模式操作，并且在时分双工上行链路操作中简单地放大并将来自合法用户的信号转发到基站。

Mobile data traffic continues to grow at an exponential rate. To meet this data challenge, massive MIMO (multiple-input multiple-output) system technology has recently been proposed where the base station employs a large number of antennas, allowing many users to be served simultaneously. It is regarded as one of the key enablers of future 5G wireless systems. While recent prototypes have demonstrated its feasibility, many significant research challenges remain to be addressed before massive MIMO can be deployed. Successful operation of massive MIMO depends critically on knowledge of the channel state information between the base station and the end users. In practice this would be acquired during the training phase where the users send individual pilot signals to the base station. This phase is challenging due to a large number of end users which lead to pilot reuse causing pilot contamination, and due to vulnerability to attacks by malicious eavesdroppers who may spoof legitimate users by transmitting identical pilot signals. This project is focused on methods to detect and defend against pilot contamination attacks. Innovative approaches to detect and defend against pilot contamination attacks from active eavesdroppers as well as from spoofing relays are investigated in this research. A key transformative idea introduced is that of self-contamination of pilot sequences by legitimate users to facilitate detection of active eavesdroppers. This project explores various ramifications of this idea in the context of the following research thrusts. (1) Detection of pilot contamination attacks by active eavesdroppers: Assuming the knowledge of the set of training sequences (and nothing else), can one detect whether one or more training sequences are under attack. Source enumeration methods based on data correlation function are being exploited. (2) Joint acquisition of channel state information for both legitimate users and eavesdroppers to facilitate effective beamforming designs to enhance reception at legitimate users while degrading reception at eavesdroppers. (3) Detection and mitigation of active eavesdropping via spoofing relay attack where a spoofing relay operates in a full-duplex mode and simply amplifies and forwards the signal from a legitimate user to the base station in a time-division duplex uplink operation.