Source-linked AI summary
RIS-Aided Integrated Sensing and Communication: Joint Beamforming and Reflection Design
Honghao Luo, Rang Liu, Ming Li, Qian Liu
TL;DR
RIS-assisted ISAC addresses spectrum-sharing sensing and communication, where practical channel degradation can limit performance. The paper jointly optimizes transmit beamforming and RIS reflection coefficients using an alternating algorithm. In the considered ISAC system, the proposed scheme reports about 4 dB improvement over other schemes.
Problem
Practical channel degradation can leave MIMO-based ISAC performance unsatisfactory, motivating RIS assistance for joint sensing and communication.
Source: §I · p. 1Method
The paper jointly optimizes transmit beamforming and RIS reflection coefficients using an alternating algorithm combining MM, penalty-based, and manifold optimization methods.
Source: §I · p. 1Results
About 4 dB performance improvement over other schemes is reported for the proposed scheme in the considered ISAC system.
Source: §IV · p. 5Takeaways & Limitations
In the simulations, schemes with RIS outperform those without RIS in both radar-only and ISAC systems, with additional NLoS links providing passive beamforming gain.
Source: §IV · p. 5
Abstract
from arXiv · showhide
Integrated sensing and communication (ISAC) has been envisioned as a promising technique to alleviate the spectrum congestion problem. Inspired by the applications of reconfigurable intelligent surface (RIS) in dynamically manipulating wireless propagation environment, in this paper, we investigate to deploy a RIS in an ISAC system to pursue performance improvement. Particularly, we consider a RIS-assisted ISAC system where a multi-antenna base station (BS) performs multi-target detection and multi-user communication with the assistance of a RIS. Our goal is maximizing the weighted summation of target detection signal-to-noise ratios (SNRs) by jointly optimizing the transmit beamforming and the RIS reflection coefficients, while satisfying the communication quality-of-service (QoS) requirement, the total transmit power budget, and the restriction of RIS phase-shift. An efficient alternating optimization algorithm combining the majorization-minimization (MM), penalty-based, and manifold optimization methods is developed to solve the resulting complicated non-convex optimization problem. Simulation results illustrate the advantages of deploying RIS in ISAC systems and the effectiveness of our proposed algorithm.
I. INTRODUCTION
ISAC shares spectrum and hardware for sensing and communication, but channel degradation can limit performance. The paper motivates RIS-assisted ISAC and jointly designs beamforming and RIS elements for a broader multi-target, multi-user setting.
I. INTRODUCTION
- ISAC shares spectrum resources between radar and communication systems to support both functions on a fully shared platform.The paper presents this as promoting spectrum and hardware efficiencies.Source: §I · p. 1
- Channel degradation in practical environments can leave MIMO-based ISAC performance unsatisfactory despite its spatial degrees of freedom.The paper identifies RIS as a promising technology for addressing this problem.Source: §I · p. 1
- Prior RIS-ISAC studies considered settings that omitted either multiple targets, multiple users, or a direct BS-to-target link.The paper specifically notes that and assume the direct target link is blocked, while the general multi-target, multi-user scenario was not considered.Source: §I · p. 1
- The paper jointly optimizes transmit beamforming and RIS elements to maximize weighted target-detection SNRs subject to communication QoS, transmit-power, and RIS phase-shift constraints.It proposes an alternating algorithm combining MM, penalty-based, and manifold optimization methods.Source: §I · p. 1
II. SYSTEM MODEL AND PROBLEM FORMULATION
The system combines multi-target radar sensing and multi-user communication at a multi-antenna BS assisted by an RIS. Its formulation maximizes weighted radar sum-SNR while enforcing communication SINR, transmit-power, and RIS phase constraints.
II. SYSTEM MODEL AND PROBLEM FORMULATION
- A multi-antenna BS detects T point-like targets and serves K single-antenna users with an N-element RIS in a narrowband system.Targets are detected successively in time division during the channel coherence time.Source: §II · pp. 1–2
- The BS transmit signal combines communication and radar probing components in an overall beamforming matrix W.W comprises communication and radar beamforming matrices, and the probing signal is statistically independent of the communication symbols.Source: §II · p. 2
- Radar echoes reach each target through direct and RIS-reflected paths, and return to the BS over propagation links.The model neglects propagation-delay differences between paths.Source: §II · p. 2
- Communication performance is constrained through each user's SINR requirement, with the model assuming perfect CSI at the BS.The user channels include direct BS-user and RIS-user links.Source: §II · p. 2
- The design maximizes weighted radar sum-SNR while satisfying user SINR thresholds, a total transmit-power budget, and unit-modulus RIS coefficients.The weights represent target priorities, and each user's threshold is predefined.Source: §II · p. 2
- The formulation is non-convex because of its objective, unit-modulus constraint, and coupling between beamforming and RIS variables.The paper therefore uses an alternating-optimization strategy.Source: §II · p. 2
III. JOINT BEAMFORMING AND RIS REFLECTION DESIGN FOR RIS-ASSISTED ISAC SYSTEMS
The design section makes the original problem more tractable by splitting it into two subproblems solved iteratively.
III. JOINT BEAMFORMING AND RIS REFLECTION DESIGN FOR RIS-ASSISTED ISAC SYSTEMS
- The method converts the joint design into two more tractable subproblems and develops algorithms to solve them iteratively.The section presents this as its approach to problem (7).Source: §III · p. 2
A. Optimize the beamforming matrix W
With the RIS coefficients fixed, the beamforming subproblem is expressed after algebraic transformations.
A. Optimize the beamforming matrix W
- For fixed RIS coefficients φ, the method algebraically reformulates the optimization problem over beamforming matrix W.This is the W-update subproblem in the alternating design.Source: A. Optimize the beamforming matrix W · p. 2
W Tr(WHC1W) (8a)
The beamforming subproblem is a non-convex QCQP; although SDR can convert it to an SDP, it may not yield a rank-one solution. The authors instead use MM surrogates and SOCP to avoid SDR’s high complexity.
W Tr(WHC1W) (8a)
- SDR converts the non-convex QCQP to an SDP, but may not guarantee a rank-one solution and can require costly randomization without ensuring overall convergence.These limitations motivate avoiding SDR’s prohibitively high complexity.Source: W Tr(WHC1W) (8a) · p. 3
- MM constructs tractable lower-bound surrogate functions for the non-convex objective at the current beamforming iterate.The resulting beamforming problem is reformulated after substituting the surrogate and dropping a constant term.Source: W Tr(WHC1W) (8a) · p. 3
- The remaining non-convex constraint is converted to a second-order cone form and solved with SOCP.This constraint transformation is cited as a way to solve the reformulated beamforming problem.Source: W Tr(WHC1W) (8a) · p. 3
B. Optimize the reflection coefficients φ
The reflection design is difficult because its objective is quartic in the RIS coefficients and it includes coupled variables and unit-modulus constraints. The authors use a penalty-based alternating scheme with MM and manifold optimization, then update the auxiliary variables by SOCP.
B. Optimize the reflection coefficients φ
- The reflection-design objective is quartic in φ, while fractional constraints and the unit-modulus constraint further hinder optimization.This follows because the target-channel matrix is quadratic with respect to φ.Source: B. Optimize the reflection… · p. 3
- Auxiliary variables relax the fractional constraint into a penalty term, and the coupled reflection and auxiliary-variable subproblems are optimized alternately.With the penalty coefficient fixed, the method separates the joint problem into updates for φ and the auxiliary variables.Source: B. Optimize the reflection… · p. 3
- MM bounds the difficult reflection objective, yielding a smooth objective on the complex-circle manifold that is solved with Riemannian conjugate gradient.The Riemannian gradient is derived from the Euclidean gradient.Source: B. Optimize the reflection… · pp. 3–4
- The auxiliary-variable update is solved by SOCP, while the penalty coefficient is increased to tighten the penalty term.The initial penalty coefficient is set relatively small.Source: B. Optimize the reflection… · p. 4
- Figure 2 presents convergence performance.Source: Figure 2 · p. 4
C. Summary
The algorithm alternates updates of beamforming, RIS reflection, auxiliary variables, and the penalty coefficient. It stops when radar SNR converges and the stopping indicator falls below its threshold.
C. Summary
- The algorithm iteratively updates the beamforming matrix, reflection coefficients, auxiliary variables, and penalty coefficient.The updates are given by equations (14), (28), (29), and (30), respectively.Source: C. Summary · p. 4
- The iteration continues until the achievable radar SNR converges and stopping indicator ζ falls below threshold ǫ.The stopping indicator is defined in the paper.Source: C. Summary · p. 4
- The updates use appropriate initial values for the beamforming matrix, reflection coefficients, auxiliary variables, and penalty coefficient.The algorithm summary identifies these as the iteratively updated quantities.Source: C. Summary · p. 4
IV. SIMULATION RESULTS
Simulations evaluate convergence and radar sum-SNR under varying transmit power, RIS size, and communication QoS. The proposed RIS-assisted ISAC scheme improves radar performance, while tighter communication QoS reduces radar sum-SNR.
IV. SIMULATION RESULTS
- The simulations use a 16-antenna BS, three targets, four single-antenna users, and an RIS-assisted system.The RIS element count varies in the reported sweep; all users have the same SINR requirement and target weights are set to one.Source: §IV · p. 4
- All schemes converge within about 30 iterations.The convergence performance is presented in Fig. 2.Source: §IV · p. 4
- About 6dB performance gain over “ISAC, w/o RIS” is achieved by the proposed algorithm with N = 60 RIS elements.Radar sum-SNR increases with RIS element count.Source: Figure 4 · p. 5, §IV · p. 5
- About 4dB performance improvement over other schemes is achieved by “ISAC, proposed” in the considered ISAC system.The figure plots radar sum-SNR versus transmit power with N = 36 and Γ = 5dB.Source: Figure 3 · p. 5, §IV · p. 5
- As the communication QoS requirement increases, radar sum-SNR decreases because more resources are allocated to guarantee QoS.This reflects a trade-off between radar detection and communication performance.Source: §IV · p. 5
V. CONCLUSION
The paper jointly designs transmit beamforming and RIS reflection for multi-target detection and multi-user communication in an ISAC system. Numerical results demonstrate RIS advantages and a radar–communication trade-off; clutter and fairness remain future extensions.
V. CONCLUSION
- The paper jointly designs transmit beamforming and RIS reflection to assist multi-target detection and multi-user communication, maximizing radar sum-SNR under communication SINR, power, and RIS-coefficient constraints.An alternating optimization algorithm addresses the resulting non-convex problem.Source: §V · p. 5
- The authors identify cluttered scenarios and fairness as topics for extending the work to more practical settings.They conclude that numerical results show RIS advantages in ISAC and reveal a performance trade-off.Source: §V · p. 5