Secrecy Rate Maximization of RIS-assisted SWIPT Systems: A Two-Timescale Beamforming Design Approach

Ming-Min Zhao, Kaidi Xu, Yunlong Cai, Yong Niu, Lajos Hanzo

Secrecy Rate Maximization of RIS-assisted SWIPT Systems: A Two-Timescale Beamforming Design Approach

Abstract

In Simultaneous Wireless Information and Power Transfer (SWIPT) systems, ensuring data security against potential eavesdropping by energy users (EUs) is a critical challenge. This paper presents a novel two-timescale beamforming design for Reconfigurable Intelligent Surface (RIS)-assisted SWIPT networks. By optimizing RIS phase shifts on a long-term scale using channel statistics and base station (BS) beamforming on a short-term scale using instantaneous effective channel state information (CSI), the proposed framework significantly reduces signaling overhead. We develop a Smooth Approximation aided Stochastic Successive Convex Approximation (SA-SSCA) algorithm to maximize the secrecy rate. Simulation results show that this approach achieves high secrecy performance and superior robustness to CSI delays compared to traditional instantaneous CSI-based methods.

Introduction

Simultaneous Wireless Information and Power Transfer (SWIPT) allows a base station (BS) to concurrently deliver data to an Information User (IU) and energy to multiple Energy Users (EUs). However, because EUs are often positioned near the BS to maximize energy harvesting, they can potentially eavesdrop on the IU's private data. Reconfigurable Intelligent Surfaces (RIS) offer a promising solution by creating favorable propagation environments through passive beamforming. Yet, traditional RIS optimization relies on instantaneous CSI for all reflecting elements, leading to prohibitive overhead and computational complexity.

System model of an RIS-aided SWIPT network featuring a BS, RIS, IU, and multiple EUs.

System Model and Problem Formulation

The system consists of a multi-antenna BS, an RIS with $N_r$ elements, one single-antenna IU, and multiple single-antenna EUs. As shown in the system model, the BS transmits a combined signal for information and energy. The RIS reflects signals to enhance the IU's reception while suppressing the signal at the EUs. The primary goal is to maximize the achievable secrecy rate for the IU while satisfying the energy harvesting requirements of the EUs.

Two-Timescale Beamforming Design

To overcome the "CSI tax" associated with instantaneous RIS control, we propose a two-timescale transmission protocol that decouples the optimization based on the varying coherence times of channel statistics and instantaneous realizations.

The proposed two-timescale transmission protocol, illustrating the separation of long-term statistical CSI estimation and short-term effective CSI updates.
  1. Long-term (Slow) Scale: The RIS phase shifts are updated periodically based on channel statistics (e.g., Rician fading parameters and path loss), which remain stable over multiple frames (super-frames).
  2. Short-term (Fast) Scale: The BS's active beamforming vectors for data and energy are updated in every time slot based on the low-dimensional "effective" instantaneous CSI, which is much easier to estimate than the full channel matrix.

Optimization Framework

The resulting optimization problem is a non-convex stochastic program. To solve it, we introduce the SA-SSCA algorithm. A key challenge is the non-differentiable "max" function in the secrecy rate objective, which accounts for the worst-case eavesdropper among the EUs. We employ a Log-Sum-Exp smoothing technique to make the objective tractable. For the short-term subproblem, a combination of the Concave-Convex Procedure (CCCP) and Block Coordinate Descent (BCD) is used. Additionally, a low-complexity heuristic is proposed that balances the effective channel power between the IU and EUs, offering a practical alternative for real-time hardware implementation.

Key Results and Insights

Simulations using 3D coordinate models and Rician fading channels reveal several key findings:

  • Secrecy Performance: The two-timescale approach significantly outperforms benchmarks like random RIS phases, demonstrating the effectiveness of statistical CSI-based passive beamforming.
  • Robustness: Unlike schemes requiring full instantaneous CSI, which suffer performance degradation due to feedback delays, the proposed method is nearly invariant to such delays because the RIS configuration relies on stable statistics.
  • Hardware Feasibility: Using discrete phase shifts (e.g., 2 or 3 bits) yields performance very close to the continuous phase shift ideal, supporting the use of low-cost RIS hardware.
  • Energy Beams as Artificial Noise: The energy signals intended for EUs effectively act as artificial noise for eavesdroppers, further enhancing the secrecy rate without requiring additional power.

Conclusion

The two-timescale beamforming design provides a robust and efficient framework for securing RIS-assisted SWIPT systems. By leveraging channel statistics for RIS control, the system achieves high secrecy rates with minimal overhead. Future research will extend this framework to multi-antenna users and multi-IU scenarios, further exploring the impact of imperfect statistical CSI on long-term performance.

Summary updated

BibTeX
@article{zhao2023secrecy,
  author = {Ming-Min Zhao and Kaidi Xu and Yunlong Cai and Yong Niu and Lajos Hanzo},
  title = {{Secrecy Rate Maximization of RIS-assisted SWIPT Systems: A Two-Timescale Beamforming Design Approach}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2023},
  doi = {10.1109/twc.2022.3225969},
}
Figure