Fundamentals of Physical Layer Anonymous Communications: Sender Detection and Anonymous Precoding

Zhongxiang Wei, Fan Liu, Christos Masouros, H. Vincent Poor

Abstract

In wireless networks, the physical layer (PHY) often leaks identity information through unique MIMO channel "fingerprints," enabling a receiver to identify a sender even when upper-layer identifiers are hidden. This paper investigates the fundamentals of PHY-anonymous communications. We first model the threat by proposing sender detection strategies based on Maximum Likelihood and Maximum Norm criteria. To counter these threats, we develop two anonymous precoding techniques: Interference Suppression (ISA) and Constructive Interference (CIA). These strategies aim to maximize the receiver's detection error while ensuring reliable data decoding. Our results show that these precoders can effectively mask sender identity, achieving high detection error rates without significantly compromising communication performance.

Introduction

Traditional network security focuses on anonymity at the upper layers of the protocol stack, using techniques like onion routing or IP address encryption. However, the physical layer remains a significant source of privacy leakage. Every wireless link possesses a unique spatial signature dictated by the specific Multiple-Input Multiple-Output (MIMO) channel between the transmitter and receiver. A sophisticated receiver can exploit these "fingerprints" to unmask a sender, posing risks in sensitive applications such as private cloud computing or anonymous e-health reporting.

System model for physical layer anonymous communications involving multiple potential senders and a receiver performing detection and decoding.

This research addresses the challenge of creating PHY-anonymous links. As illustrated in the system model, the interaction is framed as a game between a receiver attempting to detect the sender and a sender using precoding to hide their identity while maintaining data integrity.

Sender Detection and Anonymous Precoding

The paper defines two primary components: the attack (detection) and the defense (precoding).

Sender Detection Strategies

A receiver acting as an adversary can employ different strategies to identify the sender among $K$ potential users:

  • Maximum Likelihood Estimation (MLE): Suitable for scenarios where the receiver has more antennas than the sender ($N_r \ge N_t$). It leverages known channel state information (CSI) to find the best fit for the received signal.
  • Maximum Norm (M-Norm): A lower-complexity alternative for cases with fewer receiver antennas, relying on the statistical signal strength properties of the channels.

Anonymous Precoding Techniques

To protect anonymity, the sender must scramble the signal's spatial signature. We propose two methods:

  1. Interference-Suppression based Anonymous (ISA) Precoder: This approach treats inter-antenna interference as a nuisance to be suppressed. It optimizes the precoding matrix using Semi-Definite Relaxation (SDR) to ensure the transmitted signal does not match the sender's known channel signature.
  2. Constructive-Interference based Anonymous (CIA) Precoder: This more advanced method leverages interference constructively. By pushing signals into "constructive regions" of the constellation (e.g., QPSK), it facilitates easier data decoding for the receiver while simultaneously making the sender's identity harder to distinguish.

Performance Evaluation

The proposed framework was evaluated using Monte-Carlo simulations over Rayleigh block-fading channels. The study compared the ISA and CIA precoders against standard benchmarks like SVD-MIMO and MMSE.

Key results include:

  • High Anonymity: In a network of five potential senders, the anonymous precoders maintained a Detection Error Rate (DER) of nearly 0.8—approaching the theoretical limit of random guessing. In contrast, standard non-anonymous methods allowed for nearly 100% detection accuracy.
  • Reliable Communication: Despite the anonymity constraints, the Symbol Error Rate (SER) for the CIA precoder remained very close to high-performance, non-anonymous benchmarks.

Discussion and Conclusion

A critical finding of this work is the Anonymity-Diversity Trade-off: perfect receive diversity and perfect anonymity are mutually exclusive. Since diversity relies on identifying the channel to clean up the signal, it inherently reveals the sender's identity. To overcome this, the sender must perform Transmit Phase Equalization, allowing the receiver to decode data "blindly" without needing to know the specific channel.

The CIA precoder stands out for its efficiency and ability to use interference as a "cloak," boosting data rates while hiding spatial signatures. Future research will extend these concepts to multi-user simultaneous transmission and explore scenarios with imperfect CSI at the receiver.

Summary updated

BibTeX
@article{wei2022fundamentals,
  author = {Zhongxiang Wei and Fan Liu and Christos Masouros and H. Vincent Poor},
  title = {{Fundamentals of Physical Layer Anonymous Communications: Sender Detection and Anonymous Precoding}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2022},
  doi = {10.1109/twc.2021.3093722},
}
Figure