Near-Optimal Modulo-and-Forward Scheme for the Untrusted Relay Channel

Shengli Zhang, Lisheng Fan, Mugen Peng, H. Vincent Poor

TL;DR

This paper introduces a "Modulo-and-Forward" (MF) scheme for untrusted relay networks that overcomes the "interference power dilemma" of traditional methods. By using nested lattice codes and a modulo operation at the relay, the authors prove that this approach stays within 1/2 bit of the secrecy capacity and significantly outperforms standard Amplify-and-Forward (AF) techniques in both secure degrees of freedom and diversity gain.

What Problem Does This Paper Address?

In many wireless scenarios, a source ($S$) must use a relay ($R$) to reach a destination ($D$), but the relay itself might be "untrusted"—meaning it shouldn't be able to decode the confidential message. To protect the message, the destination can send "artificial noise" (AN) to jam the relay while the source is transmitting.

The standard approach is Amplify-and-Forward (AF), where the relay simply boosts and retransmits whatever it hears. This leads to the interference power dilemma:

  1. To keep the message secret, the destination sends strong noise.
  2. Because the relay has a limited power budget, it wastes most of its energy retransmitting that useless noise back to the destination.
  3. This leaves very little power for the actual message, severely limiting the transmission rate.

While Decode-and-Forward (DF) can solve this, it requires extremely strict (and often impractical) synchronization of signal phase, amplitude, and timing. The authors seek a middle ground: a scheme that is robust like AF but efficient like DF.

Methods

The authors propose a Modulo-and-Forward (MF) scheme based on nested lattice encoding.

  1. At the Source: The message is encoded using a nested lattice code. A "dither" signal (random noise known to all parties) is added to ensure the transmitted signal is uniformly distributed.
  2. At the Relay: Instead of just amplifying the received signal (which contains the message + destination's noise + channel noise), the relay performs a modulo operation relative to the source's lattice. This "wraps" the signal into a bounded region.
  3. The "Aha" Moment: This modulo operation effectively strips away the massive power overhead of the destination's artificial noise while preserving the underlying message structure. The relay's transmit power remains constant and focused on the message, regardless of how loud the destination's jamming was.
  4. At the Destination: Since the destination knows exactly what noise it sent, it can cancel its own interference and use "inflated lattice decoding" to recover the original message.

Datasets

Not discussed in the paper. This is a theoretical and analytical study. The performance is evaluated through mathematical proofs and numerical simulations based on standard Gaussian and fading channel models rather than empirical datasets.

Key Insights and Ideas

  • Solving the Power Dilemma: The MF scheme decouples the security requirement (how much noise the destination sends) from the transmission efficiency (how much power the relay uses for the message).
  • Near-Optimality: The authors prove that for any channel realization, the MF scheme achieves a secrecy rate within 1/2 bit of the theoretical secrecy capacity.
  • Synchronization Requirements: Unlike DF, which requires perfect phase and amplitude synchronization, MF only requires symbol-level time synchronization, making it much more feasible for real-world hardware.
  • Degrees of Freedom (G-SDoF): In high-signal-to-noise (SNR) environments, MF achieves the full "generalized secure degrees of freedom," whereas AF's performance collapses as the destination increases its noise power to maintain secrecy.

Key Results

  • Secrecy Capacity: MF approaches the secrecy capacity within 0.5 bits. In contrast, the AF scheme's rate is strictly capped by the interference it must carry.
  • Secure Diversity Gain (G-SDG): In fading channels without Channel State Information at the transmitter (CSIT), the MF scheme achieves a full diversity gain of order 1. The AF scheme is limited to a gain of 1/2 at most.
  • Outage Probability: The "total outage" (either the destination can't decode or the relay can decode) is significantly lower for MF. Numerical results show that the connection outage probability for MF is roughly half that of AF for the same target rate.

Limitations and Future Work

  • Synchronization: While less demanding than DF, MF still requires symbol-level synchronization. The impact of timing jitters or frequency offsets in a practical implementation isn't fully explored here.
  • Lattice Complexity: Implementing high-dimensional nested lattice codes can be computationally expensive for low-power IoT devices or high-speed links.
  • Channel Models: The study focuses on single-antenna nodes. Extending this to Multi-Input Multi-Output (MIMO) systems or multi-relay networks remains an open area for further research.

Summary updated

BibTeX
@article{zhang2016near,
  author = {Shengli Zhang and Lisheng Fan and Mugen Peng and H. Vincent Poor},
  title = {{Near-Optimal Modulo-and-Forward Scheme for the Untrusted Relay Channel}},
  journal = {IEEE Trans. Information Theory},
  year = {2016},
  doi = {10.1109/tit.2016.2530080},
}
Figure