Source-linked AI summary
RIS-Assisted Green Secure Communications: Active RIS or Passive RIS?
Weigang Lv, Jiale Bai, Qingli Yan, Hui-Ming Wang
TL;DR
Double fading limits the secrecy gain of passive RIS, while active RIS can amplify reflected signals. The paper optimizes transmitter beamforming and RIS reflection with a penalty-based alternating method; simulations report lower energy consumption for active RIS than passive RIS at matched secrecy rate.
Problem
Double fading limits the performance gain of passive RIS, while active RIS may also reduce fading on the eavesdropper path and increase information leakage.
Source: §I · p. 1Method
A penalty-based alternating minimization algorithm optimizes transmitter beamforming and the RIS reflection matrix for secure transmission.
Source: §I · p. 1Results
About 20% passive-RIS energy reduction versus less than 35% for active RIS when N rises from 10 to 60, with PI=10mW and target secrecy rate R̄=10.
Source: §IV · p. 5Takeaways & Limitations
At the same secrecy rate, active RIS consumes less energy than passive RIS and can reduce the number of reflecting units needed.
Source: §IV · p. 5, §V · p. 5
Abstract
from arXiv · showhide
Reconfigurable Intelligent Surface (RIS) is one of the promising techniques for 6G wireless communications, and recently has also been shown to be able to improve secure communications. However, there is a "double fading" effect in the reflection link between base station and user, thus passive RIS only achieves a negligible secrecy gain in typical communications scenarios.In this letter, we propose an active RIS-aided multi-antenna physical layer secrecy transmission scheme, where the active RIS can amplify the signal actively. Our aim is to minimize the transmit power subject to the constraint of secrecy rate. To solve the non-convex optimization problem, a penalty-based alternating minimization (AltMin) algorithm is proposed to optimize both the beamformer at the transmitter and the reflection matrix at RIS. Simulation results show that active RIS can resist the impact of "double fading" effect effectively, and is more energy efficient than passive RIS.
I. INTRODUCTION
RIS is studied as a low-power approach to secure communications, but double fading limits the gain of passive RIS. This letter considers active RIS for multi-antenna secure transmission and compares its power use with passive RIS under the same secrecy-rate constraint.
I. INTRODUCTION
- Double fading limits RIS performance because reflected signals experience large-scale fading on both the base-station–RIS and RIS–user paths.Active RIS elements can adjust phase and amplify signals simultaneously; active RIS is also reported to differ fundamentally from a conventional relay.Source: §I · p. 1
- Active RIS can also reduce double fading on the RIS–eavesdropper reflection path, potentially increasing information leakage in the broadcast channel.The proposed setting considers Alice communicating with Bob through active RIS in the presence of Eve.Source: §I · p. 1
- Active RIS is proposed for a multi-antenna wiretap channel to reduce transmitter power while meeting the same secrecy-rate constraint used for passive RIS.The system includes a single-antenna user and eavesdropper.Source: §I · p. 1
- A penalty-based alternating algorithm optimizes the transmitter beamformer and RIS reflection matrix, using successive convex approximation to handle the rank-one constraint.The authors report that active RIS uses less energy than passive RIS at the same secrecy rate and weakens double fading's influence.Source: §I · p. 1
II. CHANNEL MODEL AND PROBLEM FORMULATION
The paper models an active-RIS MISO wiretap system with a multi-antenna Alice, single-antenna Bob and Eve, and direct and RIS-assisted links. It formulates transmitter-power minimization under a secrecy-rate target and an RIS amplification-power budget.
II. CHANNEL MODEL AND PROBLEM FORMULATION
- The MISO wiretap system has an M-antenna Alice, single-antenna Bob and Eve, and an active RIS with N reflecting elements.The RIS matrix Q is diagonal, with each element's phase shift and amplification factor; an active element may have amplification factor greater than one.Source: §II · p. 2
- The model includes direct Alice–Bob and Alice–Eve links, plus Alice–RIS, RIS–Bob, and RIS–Eve links.The received signals at Bob and Eve are formulated separately.Source: §II · p. 2
- Active RIS requires additional power and generates thermal noise, while the analysis assumes Alice has instantaneous CSI for all links.The CSI assumption is used as a benchmark for possible practical schemes.Source: §II · p. 2
- The design minimizes total base-station transmit power subject to a target secrecy rate and a maximum RIS amplification-power budget.The secrecy rate is the positive part of Bob's achievable rate minus Eve's.Source: §II · p. 2
III. ALTERNATING MINIMIZATION ALGORITHM
The proposed penalty-based AltMin algorithm alternates between designing the transmitter beamformer and the RIS reflection matrix to solve the non-convex problem.
III. ALTERNATING MINIMIZATION ALGORITHM
- The algorithm alternates between optimizing beamformer w and RIS matrix Q, solving one subproblem with the other variable fixed.For each subproblem, it uses a penalty-based approach with SCA; the stated output is a KKT solution of P1.Source: §III · p. 2
- For fixed Q, the method penalizes an equivalent representation of the rank-one constraint and applies SCA rather than relying on Gaussian randomization.For fixed w, it introduces δ to avoid a feasibility check and applies the penalty-based SCA method to optimize Q.Source: §III · p. 2
- The algorithm's alternating optimization is stated to yield a KKT solution of P1.Source: §III · p. 2
A. Optimal Secure Transmit Beamformer w
With Q fixed, the beamformer subproblem is relaxed to an SDP, then a penalty-based SCA procedure handles the rank-one constraint and iteratively refines the solution.
A. Optimal Secure Transmit Beamformer w
- When Q is fixed, P1 reduces to P2; relaxing its rank-one constraint produces an SDP whose solution may not satisfy that constraint.The rank-one condition is represented through equality between the trace and largest eigenvalue, which holds exactly for a rank-one matrix.Source: A. Optimal Secure Transmit Beamformer w · pp. 2–3
- A penalty factor incorporates the equivalent rank-one condition into the objective, and SCA iteratively optimizes an approximate surrogate.The text states that for sufficiently small η, P3 and P4 have the same optimal solution.Source: A. Optimal Secure Transmit Beamformer w · p. 3
- The penalty-based SCA sequence forms a non-increasing objective that converges to a stationary point of P4.The method monotonically tightens an upper bound on P4's optimal value; the beamformer is obtained from the largest-eigenvalue eigenvector of W.Source: A. Optimal Secure Transmit Beamformer w · p. 3
B. Optimizing Q with given w
With the beamformer fixed, the RIS reflection matrix is optimized through a lifted rank-one formulation. Penalty-based rank handling and successive convex approximation yield a convex subproblem for recovering the RIS matrix.
B. Optimizing Q with given w
- The RIS matrix is optimized for a fixed beamformer by lifting q into U and imposing a rank-one constraint.The vector q contains each RIS element’s amplitude and phase.Source: B. Optimizing Q with given w · p. 3
- A variable δ is added to avoid a feasibility-check subproblem during optimization.The modified formulation includes a secrecy-rate constraint with δ and δ ≥ 0.Source: B. Optimizing Q with given w · pp. 3–4
- The rank-one constraint is handled with a penalty-based method, and SCA addresses the remaining non-convex constraint using a surrogate at U(t).The surrogate uses a first-order Taylor approximation.Source: B. Optimizing Q with given w · pp. 3–4
- The resulting P9 subproblem is convex; after solving it, the RIS matrix Q is reconstructed from the optimal U.The penalty-based SCA procedure iterates on U until its trace changes by no more than ε2.Source: B. Optimizing Q with given w · p. 4
2. Calculate U(t+1) in P9 with given U(t)
The SCA iteration stops when consecutive U solutions’ traces differ by no more than ε2, then outputs U as the optimal rank-one solution for P9.
2. Calculate U(t+1) in P9 with given U(t)
- The iteration stops when consecutive traces differ by at most ε2 and outputs U as P9’s optimal rank-one solution.Source: §2 · p. 4
C. Overall Algorithm
The alternating minimization objective is nonincreasing and bounded below, so the algorithm has monotonic convergence. Its two subproblem costs scale cubically with M and N+1, respectively.
C. Overall Algorithm
- The AltMin objective is nonincreasing and bounded below by zero, guaranteeing monotonic convergence.The beamformer and RIS subproblem complexities are about O(M^3) and O((N+1)^3), respectively.Source: C. Overall Algorithm · p. 4
IV. SIMULATION RESULTS
Simulations evaluate the penalty-based AltMin algorithm and compare active and passive RIS energy consumption under secrecy-rate and deployment settings. Active RIS uses less energy than passive RIS in the reported comparisons, with benefits also observed as the reflecting-element count and RIS location vary.
IV. SIMULATION RESULTS
- The penalty-based AltMin algorithm for optimizing Q and w converges under R̄ = 10, and Fig. 2 is reported to verify its convergence and effectiveness.The simulation results are averaged over 100 channel realizations; the channels use large- and small-scale fading, with Rician small-scale fading assumed for all channels.Source: §IV · p. 4, Figure 2 · p. 4
- At the same secrecy rate, active RIS consumes less energy than passive RIS, and increasing the RIS power allowance from 10 mW to 20 mW yields further transmit-power savings.Although active RIS uses extra power for signal amplification, the paper reports that this is only a small part of total consumption.Source: §IV · p. 5, Figure 3 · p. 5
- With secrecy rate R̄ = 10, increasing reflecting elements N from 10 to 60 reduces energy consumption by about 20% for passive RIS and 35% for active RIS at PI=10mW.The authors conclude that active RIS can save reflecting elements while achieving a better performance gain than passive RIS.Source: §IV · p. 5, Figure 4 · p. 5
- As the RIS position parameter ds increases, total energy consumption decreases; the simulations indicate better performance when the RIS is near Bob.This location comparison uses M=4, N=8, and R̄ = 10.Source: §IV · p. 5, Figure 5 · p. 5
V. CONCLUSION
The conclusion reports an active RIS-aided multi-antenna secrecy scheme that minimizes transmit power subject to a secrecy-rate constraint. Simulations show better energy efficiency, and the authors note possible reductions in RIS size and deployment difficulty.
V. CONCLUSION
- The proposed active RIS-aided multi-antenna secrecy scheme minimizes transmit power subject to a secrecy-rate constraint.Source: §V · p. 5
- Simulations show that the active RIS-aided scheme achieves better energy efficiency.Source: §V · p. 5
- Active RIS can reduce the number of reflecting units, thereby reducing RIS area and making practical deployment easier.Source: §V · p. 5