On the Capacity Region of Reconfigurable Intelligent Surface Assisted Symbiotic Radios

Qianqian Zhang, Hu Zhou, Ying-Chang Liang, Sumei Sun, Wei Zhang, H. Vincent Poor

Introduction

As the telecommunications industry transitions toward 6G, the "spectrum crunch" remains a primary hurdle. To meet the demands for higher data rates and massive device connectivity within limited frequency bands, researchers are exploring Symbiotic Radio (SR). In an SR system, a passive secondary transmitter "hitchhikes" on the signal of an active primary transmitter. Reconfigurable Intelligent Surfaces (RIS) are ideal candidates for this role; while typically used for passive beamforming to bypass obstacles, they can also modulate their own data onto reflected signals.

This paper investigates the fundamental capacity region—the set of all achievable data rate pairs—for an RIS-assisted symbiotic radio system. Unlike traditional additive channels, this setup creates a multiplicative interaction between the primary and secondary signals, posing unique theoretical challenges.

The Multiplicative Multiple Access Channel (M-MAC)

The researchers model the communication environment as a Multiplicative Multiple Access Channel (M-MAC). In this scenario, the direct path between the primary transmitter and the receiver is assumed to be blocked, making the RIS the sole conduit for information.

The secondary signal is not added to the primary signal but is instead multiplied by it during the reflection process. Furthermore, the RIS is a passive device, meaning it cannot amplify the signal. This imposes "passive reflection constraints" that differ significantly from the power constraints found in standard information theory. The study specifically compares two hardware configurations:

  1. Phase-only adjustment: The RIS modulates data by shifting the phase of the reflected signal.
  2. Phase and amplitude adjustment: The RIS modulates data by varying both the phase and the magnitude (attenuation) of the reflection.

Optimal Signal Distributions

A core contribution of this work is determining the "optimal input distributions"—the statistical characteristics the signals must possess to maximize throughput. In traditional additive white Gaussian noise (AWGN) channels, Gaussian distributions are typically optimal. However, the multiplicative nature of the SR channel changes the requirements:

  • Primary Link Optimization: For the secondary link to achieve its maximum potential, the primary transmitter should ideally send a signal with a constant amplitude (similar to a pure sine wave).
  • Sum-Rate Capacity: To maximize the total throughput of both users (the sum-rate), the primary signal's amplitude should follow a continuous Rayleigh distribution.
  • Boundary Points: For other points on the capacity region boundary, the optimal signal is often "discrete," meaning the transmitter should utilize a finite set of specific power levels rather than a continuous range.

The Impact of Amplitude Control

The analysis reveals that allowing the RIS to adjust the amplitude of reflections, rather than just the phase, is a significant advantage. This capability increases the "degrees of freedom" for the secondary link, effectively doubling the slope of the rate increase at high power levels.

Key findings include:

  • Capacity Expansion: Systems with amplitude control exhibit a significantly larger capacity region. At high Signal-to-Noise Ratios (SNR), this configuration provides a gain of approximately 4.34 dB over phase-only systems.
  • Rate Trade-offs: The capacity region is strictly convex. While users can trade off rates smoothly, there are clear diminishing returns when attempting to push either the primary or secondary user to their absolute theoretical limit.
  • Primary Dominance: Because the secondary user relies entirely on the primary user's energy for its reflections, the maximum achievable rate for the primary user is inherently higher than that of the secondary user.

Limitations and Future Directions

While this research provides a rigorous theoretical foundation, several practical challenges remain:

  • Direct Link Interference: The current model assumes the direct link is blocked. Real-world applications may involve a hybrid additive-multiplicative model where a weak direct signal is present.
  • Channel State Information (CSI): The study assumes perfect CSI. In practice, estimating the channel parameters for a passive RIS is complex and overhead-intensive.
  • MIMO Scaling: The current focus is on single-antenna devices. Future research should explore how these capacity regions scale with Multi-Input Multi-Output (MIMO) arrays, which are standard in 5G and 6G architectures.

Conclusion

By characterizing the capacity region of RIS-assisted symbiotic radios through the lens of the M-MAC, this paper demonstrates that RIS can do more than just improve coverage—it can serve as an efficient secondary data source. The transition from phase-only to phase-and-amplitude modulation represents a vital step in maximizing the spectral efficiency of future wireless networks.

Summary updated

BibTeX
@article{zhang2023capacity,
  author = {Qianqian Zhang and Hu Zhou and Ying-Chang Liang and Sumei Sun and Wei Zhang and H. Vincent Poor},
  title = {{On the Capacity Region of Reconfigurable Intelligent Surface Assisted Symbiotic Radios}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2023},
}
Figure