Secure symbol-level precoding for reconfigurable intelligent surface-aided cell-free networks

Abstract

With the improvement in communication network density, inter-cell interference has become severe. Due to the blurring of boundaries, cell-free networks are considered as a solution. However, it faces some challenges, such as high energy consumption due to the deployment of a large number of base stations, and security issues in complex communication scenarios. To tackle these issues, we propose a novel modeling scheme involving symbol-level precoding and reconfigurable intelligent surfaces (RIS). This solution can reduce the base station transmission power while ensuring communication layer security. Then, we decompose the non-convex problem of modeling into two sub-problems and solve them iteratively. The first sub-problem is to design symbol-level precoding which can be realized by an efficient gradient descent algorithm. The second sub-problem is about solving the reflection coefficients of RIS, which can be obtained by a Riemann conjugate gradient algorithm. In simulations, our proposed method outperforms benchmark methods. While ensuring physical layer security, the power consumption of cell-free networks has been reduced by 6 dBm.

Appendices

Appendix A: Real-valued formula conversion

By the definition of Eq. (10), the following equation is transformed into a real-valued equation

$$\begin{aligned} \begin{aligned}&\Re \left\{ \sum \limits _{b=1}^{B}{\tilde{{\textbf{H}}}_{b,k,p}^{H}{{{\textbf{x}}}_{m}}}{{e}^{-j\angle {{s}_{m,k}}}} \right\} \\&\quad = \Re \left\{ \tilde{{\textbf{H}}}_k^{T}{{{\textbf{x}}}_{m}} \right\} = \Re \left\{ \tilde{{\textbf{H}}}_k \right\} ^T\Re \left\{ {{{\textbf{x}}}_{m}} \right\} ^T - \Im \left\{ \tilde{{\textbf{H}}}_k \right\} ^T\Im \left\{ {{{\textbf{x}}}_{m}} \right\} ^T = \bar{{\textbf{H}}}_k^T{{\varvec{\Delta }}_1}{\bar{{\textbf{x}}}_m} \end{aligned} \end{aligned}$$

Similarly,

$$\begin{aligned} \begin{aligned}&\Im \left\{ \sum \limits _{b=1}^{B}{\tilde{{\textbf{H}}}_{b,k,p}^{H}{{{\textbf{x}}}_{m}}}{{e}^{-j\angle {{s}_{m,k}}}} \right\} =\bar{{\textbf{H}}}_k^T{{\varvec{\Delta }}_2}{\bar{{\textbf{x}}}_m} \\&\Re \left\{ \sum \limits _{b=1}^{B}{\tilde{{\textbf{H}}}_{b,e,p}^{H}{{{\textbf{x}}}_{m}}}{{e}^{-j\angle {{s}_{m,1}+\pi /2}}} \right\} =\bar{{\textbf{H}}}_e^T{{\varvec{\Delta }}_1}{\bar{{\textbf{x}}}_m} \\&\Im \left\{ \sum \limits _{b=1}^{B}{\tilde{{\textbf{H}}}_{b,e,p}^{H}{{{\textbf{x}}}_{m}}}{{e}^{-j\angle {{s}_{m,1}+\pi /2}}} \right\} =\bar{{\textbf{H}}}_e^T{{\varvec{\Delta }}_2}{\bar{{\textbf{x}}}_m} \end{aligned} \end{aligned}$$

Appendix B: Random distribution methods

The location coordinates of the User and the Eve are randomly generated using the following equations

$$\begin{aligned} \begin{aligned}&x = x_0+R_0L_\textrm{amp}\cos {(\textrm{angle})}\\&y = y_0+R_0L_\textrm{amp}\sin {(\textrm{angle})}\\ \end{aligned} \end{aligned}$$

Where $x_0$ and $y_0$ are the coordinates of the center point, $R_0$ is the range radius, $L_\textrm{amp}$ is a random number between 0 and 1, and angle is a random angle.