Precoder Design for mmWave UAV Communications with Physical Layer Security

Sung Joon Maeng, Yavuz Yapici, Ismail Güvenç, Huaiyu Dai, Arupjyoti Bhuyan

Introduction

The deployment of Unmanned Aerial Vehicles (UAVs) as flying base stations (UAV-BS) offers a flexible solution for on-demand wireless coverage. However, the broadcast nature of wireless signals makes these systems inherently vulnerable to eavesdropping. While traditional security relies on upper-layer encryption, Physical Layer Security (PLS) provides an additional defense by exploiting the inherent characteristics of the wireless channel. In millimeter-wave (mmWave) communications, the high directionality of signals can be leveraged to "steer" information away from unauthorized receivers.

System Model and Problem Formulation

We consider a UAV-BS equipped with $N$ antennas serving $K$ single-antenna users in the presence of a multi-antenna passive eavesdropper ("Eve"). The primary challenge is to design precoders—the algorithms mapping data to antenna elements—that maximize the sum secrecy rate while managing the trade-off between information signal power and Artificial Noise (AN) power.

System model of a UAV-BS communicating with users in the presence of an eavesdropper.

Proposed Precoding Strategies

This research moves beyond conventional "blind" precoding, which ignores the eavesdropper's location, by proposing two strategies aware of Eve's Channel State Information (CSI):

  1. Linear Precoding (Full vs. Limited CSI): By utilizing Singular Value Decomposition (SVD) to identify the dominant direction of Eve's channel, the researchers treat the eavesdropper as a "virtual" user. Applying Zero-Forcing (ZF) or Regularized Zero-Forcing (RZF) to this virtual matrix allows the precoder to naturally suppress signals in Eve's direction.
  2. Nonlinear Precoder Design: This core innovation formulates a non-convex optimization problem to minimize total transmit power while meeting specific Signal-to-Interference-plus-Noise Ratio (SINR) targets for users and a maximum leakage threshold for Eve. This problem is transformed into a solvable Second-Order Cone Programming (SOCP) format.

Performance Evaluation and Key Insights

Simulation results based on the 3GPP Urban Micro (UMi) mmWave path-loss model reveal several critical findings regarding the efficiency of these strategies.

Sum secrecy rate vs. power splitting factor $\phi$ for an underloaded scenario ($K=4$).
  • The "AN-Free" Advantage: In underloaded scenarios (where there are fewer users than antennas, e.g., $K=4$), knowing Eve's CSI allows the system to achieve peak secrecy without any artificial noise (where the power splitting factor $\phi=1$). As shown in the figure, the proposed precoding schemes continue to improve as more power is allocated to data, whereas conventional schemes suffer a performance drop.
  • Robustness in Overloaded Scenarios: In cases where $K=32$ (overloaded), conventional linear precoders fail almost completely. In contrast, the proposed nonlinear SOCP approach maintains a robust sum secrecy rate of approximately 8 bps/Hz.
  • Limited CSI Effectiveness: Even if the UAV only knows the Line-of-Sight (LoS) angle to the eavesdropper rather than full channel info, the system achieves a substantial secrecy boost over conventional methods.

Conclusion

This work demonstrates that "smarter" precoding, informed by eavesdropper CSI, can significantly enhance the security of mmWave UAV communications without the power overhead of traditional artificial noise. Future research will address the practical challenges of estimating the channels of truly passive eavesdroppers and reducing the computational complexity of real-time SOCP optimization on UAV platforms.

Summary updated

BibTeX
@inproceedings{maeng2020precoder,
  author = {Sung Joon Maeng and Yavuz Yapici and Ismail Güvenç and Huaiyu Dai and Arupjyoti Bhuyan},
  title = {{Precoder Design for mmWave UAV Communications with Physical Layer Security}},
  booktitle = {SPAWC},
  year = {2020},
  doi = {10.1109/spawc48557.2020.9154268},
}
Figure