Pilot Spoofing Attack on the Downlink of Cell-Free Massive MIMO: From the Perspective of Adversaries

Weiyang Xu, Yuan Zhang, Ruiguang Wang, Hien Quoc Ngo, Wei Xiang

Abstract

The channel hardening effect is less pronounced in the cell-free massive multiple-input multiple-output (mMIMO) system compared to its cellular counterpart, making it necessary to estimate the downlink effective channel gains to ensure decent performance. However, the downlink training inadvertently creates an opportunity for adversarial nodes to launch pilot spoofing attacks (PSAs). This paper investigates the impact of such attacks from an adversarial perspective, deriving closed-form expressions for achievable rates and optimizing attack power to minimize legitimate system performance.

Introduction: The Vulnerability of Cell-Free mMIMO

Cell-Free Massive MIMO (mMIMO) is a promising architecture for 6G, offering superior coverage and spectral efficiency by deploying a large number of distributed Access Points (APs) to serve users. Unlike traditional cellular systems where "channel hardening" allows users to rely on statistical channel information for data detection, cell-free systems exhibit much weaker hardening. Consequently, these systems must employ a downlink training phase where APs transmit pilot signals, allowing users to estimate their effective channels for coherent data detection.

However, this reliance on downlink training introduces a critical security vulnerability. Because pilot sequences are often public or predictable, adversarial nodes can exploit this phase to "poison" the user's channel estimates, leading to massive errors and dropped data rates.

The Attack Framework

The researchers analyze the threat from the perspective of an adversary asking: "How can an attacker cause the most damage?" The attack typically unfolds in two distinct phases: uplink training and downlink training.

System model for uplink and downlink training phases under a pilot spoofing attack.

As illustrated in the figure above, the adversary first listens during the Uplink Training phase (a) to estimate the channel between the malicious APs and the target users. In the subsequent Downlink Training phase (b), the malicious APs transmit the same pilot sequences as the legitimate APs, but precoded based on their earlier estimates. This results in a "poisoned" channel estimate at the user side. When the legitimate data arrives, the user applies an incorrect filter to decode it, resulting in a significant performance drop.

Beyond the Pilot Spoofing Attack (PSA), the authors also consider a Data Phase Interference strategy. In this scenario, adversaries remain silent during training to avoid detection but inject precoded random noise during the actual data transmission phase to jam the legitimate signal.

Analytical Methods and Optimization

To quantify the impact, the authors derived rigorous closed-form expressions for the per-user achievable downlink rate under both PSA and data-phase interference. These expressions account for the contamination of channel estimates and the resulting mismatch in the decoding process.

The study employs min-max optimization to determine the most damaging strategy for the adversary. By using techniques like Sequential Convex Approximation (SCA) and the bisection method, the adversary can optimally distribute its power budget across different users to minimize the maximum achievable rate in the system. This "intelligent" attack is shown to be far more effective than a simple "brute-force" noise injection.

Key Insights and Results

The research yields several critical insights into the security of 6G networks:

  • The Training Trap: While necessary for performance, downlink training provides a predictable window for attackers. The study shows that in a system with 128 legitimate APs, just 32 adversarial nodes can slash the downlink sum rate by nearly 50% (from ~28 bits/s/Hz to ~15 bits/s/Hz).
  • Estimation Poisoning vs. Jamming: Surprisingly, the analysis reveals that attacking the data transmission phase with precoded random noise is often more effective at reducing the total system rate than spoofing pilots, provided the attacker has a fixed power budget.
  • Power Asymmetry: Simply increasing the transmit power of legitimate APs is not a silver bullet. Because the user’s channel estimate itself is contaminated, the user will still incorrectly decode the signal, even if it is stronger.
  • Adversarial Collusion: The attack's effectiveness peaks when adversarial APs act in collusion, optimizing their power allocation factors to target specific users strategically rather than blasting noise blindly.

Conclusion and Future Directions

This work highlights a significant hurdle for the deployment of Cell-Free mMIMO. While the paper focuses on the attack's mechanics and optimization, it underscores the urgent need for robust defense mechanisms. Future research must prioritize the development of robust detection methods—such as identifying anomalies in received pilot power—and implementing "pilot-hopping" or cryptographic pilot schemes to defend these promising 6G networks against sophisticated physical-layer threats.

Summary updated

Abstract

The channel hardening effect is less pronounced in the cell-free massive multiple-input multiple-output (mMIMO) system compared to its cellular counterpart, making it necessary to estimate the downlink effective channel gains to ensure decent performance. However, the downlink training inadvertently creates an opportunity for adversarial nodes to launch pilot spoofing attacks (PSAs). First, we demonstrate that adversarial distributed access points (APs) can severely degrade the achievable downlink rate. They achieve this by estimating their channels to users in the uplink training phase and then precoding and sending the same pilot sequences as those used by legitimate APs during the downlink training phase. Then, the impact of the downlink PSA is investigated by rigorously deriving a closed-form expression of the per-user achievable downlink rate. By employing the min-max criterion to optimize the power allocation coefficients, the maximum per-user achievable rate of downlink transmission is minimized from the perspective of adversarial APs. As an alternative to the downlink PSA, adversarial APs may opt to precode random interference during the downlink data transmission phase in order to disrupt legitimate communications. In this scenario, the achievable downlink rate is derived, and then power optimization algorithms are also developed. We present numerical results to showcase the detrimental impact of the downlink PSA and compare the effects of these two types of attacks.
BibTeX
@article{xu2024pilot,
  author = {Weiyang Xu and Yuan Zhang and Ruiguang Wang and Hien Quoc Ngo and Wei Xiang},
  title = {{Pilot Spoofing Attack on the Downlink of Cell-Free Massive MIMO: From the Perspective of Adversaries}},
  journal = {IEEE Trans. Inf. Forensics Secur.},
  year = {2024},
  url = {https://openreview.net/forum?id=Uk7TMY37Ik},
}
Figure