PaperlayerView exact revision on arXiv

Source-linked AI summary

Reconfigurable Intelligent Surface (RIS) Aided Multi-User Networks: Interplay Between NOMA and RIS

Yuanwei Liu, Xidong Mu, Xiao Liu, Marco Di Renzo, Zhiguo Ding, Robert Schober

arXiv:2011.13336v2cs.ITeess.SP

TL;DR

The paper studies how RIS configurations interact with NOMA and OMA in multi-user wireless networks. It characterizes capacity regions, examines deployment strategies, and proposes joint active-passive beamformer designs. Dynamic RIS with NOMA is capacity-achieving, while asymmetric and symmetric deployments are preferable for NOMA and OMA, respectively.

  • Problem

    The paper addresses how RISs should be configured and combined with NOMA or OMA in multi-user networks, including capacity, deployment, and beamforming questions.

  • Method

    The paper analyzes static and dynamic RIS configurations, compares NOMA and OMA capacity or rate regions, studies deployment strategies, and proposes joint beamformer designs.

  • Results

    Dynamic RIS with NOMA is capacity-achieving, while RIS deployment is preferably asymmetric for NOMA and symmetric for OMA.

  • Takeaways & Limitations

    The interplay between RIS configuration and multiple access determines capacity behavior, deployment preferences, and beamformer design choices in RIS-aided networks.

Abstract

from arXiv · show

This article focuses on the exploitation of reconfigurable intelligent surfaces (RISs) in multi-user networks employing orthogonal multiple access (OMA) or non-orthogonal multiple access (NOMA), with an emphasis on investigating the interplay between NOMA and RIS. Depending on whether the RIS reflection coefficients can be adjusted only once or multiple times during one transmission, we distinguish between static and dynamic RIS configurations. In particular, the capacity region of RIS aided single-antenna NOMA networks is characterized and compared with the OMA rate region from an information-theoretic perspective, revealing that the dynamic RIS configuration is capacity-achieving. Then, the impact of the RIS deployment location on the performance of different multiple access schemes is investigated, which reveals that asymmetric and symmetric deployment strategies are preferable for NOMA and OMA, respectively. Furthermore, for RIS aided multiple-antenna NOMA networks, three novel joint active and passive beamformer designs are proposed based on both beamformer based and cluster based strategies. Finally, open research problems for RIS-NOMA networks are highlighted.

I. INTRODUCTION

RISs enable programmable wireless propagation with lower energy use than active relays, while multi-user operation creates interference-coordination challenges. The article examines NOMA and OMA with static and dynamic RIS configurations.

  • RISs control wireless propagation through configurable element phases and amplitudes, enhancing desired signals while mitigating interference.
  • Compared with active relaying, RISs consume less energy and cost less because they lack RF chains and high-power components.
  • Serving multiple users across time, frequency, or power domains requires coordinated interference management to exploit active and passive beamforming degrees of freedom.
  • Static RIS coefficients are adjusted once per transmission, whereas dynamic RIS coefficients are reconfigured N times with equal-duration configurations.
  • Dynamic RIS operation can create an artificial fading channel and support resource allocation, especially in static or quasi-static scenarios.
  • RIS reconfiguration takes 0.22 millisecond, shorter than typical channel coherence times of tens of milliseconds, facilitating dynamic operation.

B. Motivation for Applying NOMA in RIS-aided Multi-user Networks

The article investigates NOMA in RIS-aided multi-user networks because RISs can improve signal diversity, increase design flexibility, and relax some multiple-antenna constraints. It develops information-theoretic comparisons, deployment insights, and joint beamformer designs.

  • RIS reflection links provide additional signal diversity without requiring extra time slots or energy, potentially improving existing NOMA networks.
  • RISs increase NOMA design flexibility by supporting a transition from channel-condition-based NOMA toward QoS-based NOMA.
  • RISs can relax antenna-related constraints in multiple-antenna NOMA networks through additional passive array gains.
  • RIS-NOMA integration introduces new challenges that motivate studying how NOMA and RIS should be jointly designed in multi-user networks.
  • The article compares capacity and rate limits for NOMA and OMA, studies RIS deployment strategies, and proposes joint active-passive beamformer designs.
  • For single-antenna RIS-NOMA networks, the dynamic RIS configuration is shown to be capacity-achieving.

II. CAPACITY LIMITS: AN INFORMATION-THEORETIC PERSPECTIVE

The capacity analysis considers a RIS-aided broadcast channel with static and dynamic RIS configurations and compares NOMA against TDMA and FDMA. Dynamic RIS with NOMA achieves the capacity region, although exact finite-N characterization becomes exponentially complex.

  • The analysis compares NOMA capacity regions with TDMA and FDMA rate regions under both RIS configurations.
  • Dynamic RIS with NOMA is capacity-achieving for the considered RIS-aided multi-user broadcast channel.
  • For static RIS, fixed reflection coefficients reduce the system to a conventional broadcast channel, with capacity obtained by uniting NOMA rate tuples over coefficient choices.
  • The exact dynamic-RIS capacity region is prohibitively complex because the number of reflection-coefficient combinations grows exponentially with N.
  • As N tends to infinity, time sharing over RIS configurations yields the convex hull of the capacity regions from all static configurations.
  • The dynamic RIS capacity region contains the finite-N dynamic region, which contains the static RIS capacity region.

C. Discussions and Insights

The numerical example compares NOMA and TDMA/FDMA rate regions for static and dynamic RIS configurations, showing capacity gains and implementation trade-offs. RIS deployment benefits both access schemes, while NOMA gains more relative to OMA when an RIS is used.

  • Capacity-region comparison: The NOMA capacity region contains the TDMA/FDMA rate region under both static and dynamic RIS configurations.The comparison uses a random channel realization and evaluates the dynamic configuration in the limit N tending to infinity.
  • Capacity-complexity trade-off: Dynamic RIS phase adjustment provides a much larger capacity/rate gain for TDMA/FDMA than for NOMA.For NOMA, static RIS operation yields only a small gain relative to dynamic operation, whereas OMA requires real-time RIS control to fully exploit RIS benefits.
  • Capacity-complexity trade-off: NOMA trades additional user-side SIC complexity for reduced RIS implementation complexity, while OMA avoids SIC but needs more sophisticated RIS control.The paper identifies this capacity-versus-implementation-complexity trade-off as a topic for further investigation.
  • RIS deployment benefits: An RIS introduces significant capacity/rate gains for both NOMA and TDMA/FDMA compared with the corresponding systems without an RIS.The comparison includes both RIS configurations and the no-RIS case in Fig. 2.
  • RIS-NOMA interplay: The RIS-aided two-user BC shows a more pronounced NOMA-over-OMA capacity gain than the corresponding comparison without an RIS.This result underscores the benefit of combining RIS deployment with NOMA in the considered setting.
  • Open problems: Capacity limits and capacity-achieving strategies for RIS-aided networks with multiple-antenna transmitters and receivers remain largely unknown.The paper identifies these limits as an open topic for future research.

III. RIS DEPLOYMENT DESIGN FOR MULTIPLE ACCESS

The section examines how RIS deployment location interacts with multiple-access choices, focusing on QoS-aware NOMA and comparisons with OMA schemes. It also identifies joint deployment, resource-allocation, and mobility-related optimization challenges.

  • QoS-based NOMA: Consolidating the RIS near a strong user enlarges channel disparity and improves NOMA over OMA when that user has stricter QoS requirements.Reverse deployment near the weak user is appealing when the weak user has the higher QoS requirement.
  • RIS deployment and multiple access: RIS placement near users with poor channels or near spatially clustered users can jointly improve TDMA and FDMA performance.These schemes prefer deployment strategies that create more similar channel conditions across users.
  • RIS deployment and multiple access: Asymmetric RIS deployment is preferable for NOMA, whereas symmetric deployment provides better weighted sum-rate in FDMA and TDMA.Asymmetry creates stronger channel differences that benefit NOMA, while similar channel conditions favor FDMA and TDMA.
  • QoS-based NOMA: For nonuniform rate weights, the RIS is placed closest to the highest-weight user, increasing that user’s channel distinctiveness and benefiting NOMA.The reported four-user weighted-sum-rate study assumes blocked BS-user links and Rician fading on BS-RIS-user links.
  • Open problems: Future designs must jointly optimize RIS location, multiple-access choice, configuration, and wireless-resource allocation for moving users and time-varying QoS.The section suggests machine learning based on long-term performance metrics because conventional optimization may be ineffective for the resulting complicated problems.

IV. RIS-AIDED MULTIPLE-ANTENNA NOMA NETWORKS: JOINT ACTIVE AND PASSIVE BEAMFORMER DESIGN

The paper studies RIS-aided multiple-antenna NOMA through two broad strategy classes: beamformer-based and cluster-based transmission. For each class, it considers joint active beamforming at transmitters and passive beamforming at RISs.

  • System framework: RIS-aided multiple-antenna NOMA uses joint active beamforming and passive RIS beamforming to enhance spectral efficiency.The considered networks equip both transmitters and receivers with multiple antennas.
  • Strategy classification: Multiple-antenna NOMA strategies are classified as beamformer-based or cluster-based according to whether one beamformer serves one user or multiple users.The paper discusses joint active and passive beamformer designs for both classes.

A. Beamformer-based RIS-NOMA: A Novel Equivalent Reconfigurable Channel Inspired Design

The beamformer-based strategy constructs an equivalent reconfigurable channel by combining direct and RIS-assisted links, enabling conventional NOMA tools and channel-condition-based passive design. Initial results indicate that RIS reconfiguration can reduce regions where NOMA underperforms DPC, while joint optimization remains challenging.

  • Equivalent reconfigurable channel: The proposed equivalent reconfigurable channel combines direct and RIS-assisted links to jointly optimize BS active and RIS passive beamformers.The cascaded BS-RIS-user channels are reconfigured by the RIS and combined with direct BS-user channels.
  • Joint design challenge: RIS-NOMA decoding and SIC depend jointly on the BS active beamformer and RIS passive beamformer, complicating optimal decoding-order and beamformer design.For a two-user pair, the strong user decodes the weak user before decoding its own signal, subject to the SIC rate condition.
  • Equivalent reconfigurable channel: For a fixed RIS passive beamformer, the equivalent channel maps the RIS-NOMA network into a conventional network where known NOMA active-beamformer solutions can be used.The mapping provides a basis for the beamformer-based RIS-NOMA strategy.
  • Passive beamformer design: Passive beamformer design can target special channel conditions such as quasi-degradation, where NOMA can achieve performance comparable to dirty paper coding.These conditions support low-complexity or closed-form passive-beamformer designs.
  • Initial numerical results: RIS reconfiguration can significantly reduce the non-quasi-degradation region where NOMA performs worse than DPC.The comparison considers two-user downlink MISO systems with and without RIS for a fixed channel vector of user 1.

B. Cluster-based RIS-NOMA: Centralized and Distributed Design

The cluster-based strategy partitions users into clusters served by common beamformers, with centralized RIS-enabled and distributed RIS-enhanced designs. RIS passive array gains can relax antenna-to-user constraints and facilitate inter-cluster interference cancellation and intra-cluster enhancement.

  • Cluster-based strategy: Cluster-based RIS-NOMA partitions users into clusters and serves users within each cluster using one common beamformer.The strategy builds on decomposing MIMO NOMA into multiple SISO NOMA channels when suitable precoding and zero-forcing detection are available.
  • Cluster-based strategy: The conventional decomposition requires specific relationships between BS antennas and users, limiting the practicality of cluster-based NOMA.This requirement is identified as a constraint on conventional cluster-based transmission.
  • RIS-enabled clustering: RISs can relax antenna-to-user requirements by supplying passive array gains for inter-cluster interference cancellation and intra-cluster signal enhancement.This motivates RIS-assisted cluster-based designs.
  • Centralized and distributed designs: The paper proposes centralized RIS-enabled and distributed RIS-enhanced cluster-based designs and compares their advantages and disadvantages.The two designs are illustrated in Fig. 5, alongside initial numerical results for the centralized strategy.

1) Centralized RIS-enabled Design:

The centralized RIS-enabled cluster-based NOMA design uses one centrally deployed RIS to support user clustering, particularly when direct BS-user links are absent or highly correlated. It outperforms ZF- and OMA-based baselines, while requiring many RIS elements and accurate network-wide instantaneous CSI.

  • 1) Centralized RIS-enabled Design:: The design partitions users into clusters enabled by the passive beamformer at one centrally deployed RIS.
  • 1) Centralized RIS-enabled Design:: Centralized RIS-enabled clustering is appealing when direct BS-user links are absent or highly correlated, preventing active-beamformer-based clustering at the BS.
  • 1) Centralized RIS-enabled Design:: The centralized RIS-enabled cluster-based NOMA scheme outperforms the baseline ZF-based and OMA-based schemes.
  • 1) Centralized RIS-enabled Design:: As the number of RIS elements increases, RIS gains are more pronounced for NOMA than for ZF and OMA.
  • 1) Centralized RIS-enabled Design:: A high passive array gain for user clustering may require extremely many RIS elements and accurate instantaneous CSI for the entire network.
  • 1) Centralized RIS-enabled Design:: Robust joint active and passive beamformer design with imperfect CSI remains a promising research direction.

2) Distributed RIS-enhanced Design:

The distributed RIS-enhanced cluster-based NOMA design assigns distributed RISs to enhance separate user clusters and is suitable for widely distributed users. Its localized impact reduces CSI dependence, while distributed coordination introduces challenges.

  • 2) Distributed RIS-enhanced Design:: Distributed RIS deployment introduces coordination challenges, including simultaneous information exchange at the BS and additional challenges under dynamic RIS configurations.
  • 2) Distributed RIS-enhanced Design:: Distributed RIS-enhanced cluster-based NOMA is suitable when users are widely distributed, with each RIS enhancing one specific cluster within local coverage.
  • 2) Distributed RIS-enhanced Design:: The distributed arrangement acts as an add-on performance enabler for existing conventional NOMA networks.
  • 2) Distributed RIS-enhanced Design:: Each distributed RIS has little impact on unintended clusters because of relatively large distances, so its passive beamformer mainly depends on target-cluster CSI.
  • 2) Distributed RIS-enhanced Design:: The distributed design can serve more users and expand BS coverage by pairing cell-edge users with cell-center users in NOMA clusters.

V. CONCLUSIONS AND RESEARCH OPPORTUNITIES

The paper summarizes RIS-aided NOMA results across capacity, deployment, and joint beamforming, then identifies open challenges in CSI, multi-objective learning, and hardware validation. Dynamic RIS operation is necessary for capacity maximization, while deployment preferences differ between NOMA and OMA.

  • V. CONCLUSIONS AND RESEARCH OPPORTUNITIES: Dynamic RIS configuration is necessary to maximize capacity in RIS-aided single-antenna NOMA networks.
  • V. CONCLUSIONS AND RESEARCH OPPORTUNITIES: Asymmetric RIS deployment is preferable for NOMA, whereas symmetric deployment is preferable for OMA.
  • V. CONCLUSIONS AND RESEARCH OPPORTUNITIES: The paper proposes joint active and passive beamformer designs for multiple-antenna NOMA using beamformer-based and cluster-based strategies.
  • V. CONCLUSIONS AND RESEARCH OPPORTUNITIES: For cluster-based strategies, centralized RIS-enabled and distributed RIS-enhanced designs are developed with their respective advantages and disadvantages.
  • V. CONCLUSIONS AND RESEARCH OPPORTUNITIES: Future work includes efficient CSI estimation and robust design because CSI affects both NOMA user clustering and SIC decoding order.
  • V. CONCLUSIONS AND RESEARCH OPPORTUNITIES: Further research also requires multi-objective ML optimization, hardware experiments, real-time RIS control, and mitigation of hardware impairments and SIC error propagation.
Loading 2011.13336v2…