PaperlayerView exact revision on arXiv

Source-linked AI summary

IRS-Enhanced OFDMA: Joint Resource Allocation and Passive Beamforming Optimization

Yifei Yang, Shuowen Zhang, Rui Zhang

arXiv:1912.01228v1cs.ITmath.IT

TL;DR

IRS-aided multiuser OFDMA must use common reflection coefficients across frequency sub-bands, limiting how well they can adapt to users’ channels. This paper varies the coefficients across time slots and jointly optimizes them with resource allocation; numerical results show higher common rates than fixed beamforming, random phases, or no IRS.

  • Problem

    An IRS cannot set separate reflection coefficients for different frequency sub-bands, so one set must adapt to many users’ channels at once.

    Source: §II · p. 2

  • Method

    The scheme dynamically varies IRS reflection coefficients across time slots and jointly optimizes them with OFDMA resource-block assignments and power.

    Source: §I · p. 1, §II · p. 2

  • Results

    Dynamic passive beamforming yields a higher common rate than fixed passive beamforming and random-phase schemes, while IRS schemes outperform no IRS as the number of reflecting elements increases.

    Source: §V · p. 5

  • Takeaways & Limitations

    Dynamic reflection can more flexibly balance multiuser diversity and IRS passive-beamforming gains than fixed reflection.

    Source: §V · p. 5

Abstract

from arXiv · show

Intelligent reflecting surface (IRS) is an emerging technique to enhance the wireless communication spectral efficiency with low hardware and energy cost. In this letter, we consider the integration of IRS to an orthogonal frequency division multiple access (OFDMA) based multiuser downlink communication system, and study the pertinent joint optimization of the IRS reflection coefficients and OFDMA time-frequency resource block as well as power allocations to maximize the users' common (minimum) rate. Specifically, due to the lack of frequency-selective passive beamforming capability at the IRS, only one set of reflection coefficients can be designed for adapting to a large number of channels of multiple users over different frequency sub-bands. To tackle this difficulty, we propose a novel dynamic passive beamforming scheme where the IRS reflection coefficients are dynamically adjusted over different time slots within each channel coherence block to create artificial time-varying channels and select only a subset of the users to be simultaneously served in each time slot, thus achieving a higher passive beamforming gain. Although the formulated optimization problem is non-convex, we propose an efficient algorithm to obtain a suboptimal solution to it. Numerical results show that the proposed scheme significantly improves the system common rate over the setup without IRS and that with random IRS reflection coefficients. Moreover, our proposed dynamic passive beamforming outperforms the fixed passive beamforming which employs a common set of reflection coefficients in each channel coherence block, by more flexibly balancing between passive beamforming and multiuser diversity gains.

I. INTRODUCTION

IRS can alter wireless propagation through passive reflection, but extending IRS beamforming to multiuser broadband OFDMA requires jointly designing reflections and resource allocation. The paper proposes time-varying reflection coefficients and evaluates their common-rate performance.

I. INTRODUCTION

  • Prior IRS passive-beamforming work mainly addressed narrowband frequency-flat channels, while broadband frequency-selective studies considered single-user setups.The paper identifies multiuser broadband communication as not previously investigated to the authors’ knowledge.Source: §I · p. 1, Figure 1 · p. 1
  • IRS uses tuned passive elements to create favorable signal paths without signal generation or amplification, reducing hardware and energy costs relative to active beamforming and relaying.Earlier broadband frequency-selective studies cited by the paper address single-user setups [6]–[8].Source: §I · p. 1
  • The proposed dynamic design changes IRS coefficients across time slots and schedules fewer simultaneous users to improve passive beamforming.The paper jointly optimizes reflection coefficients, OFDMA time-frequency resource blocks, and power to maximize the minimum user rate; its numerical results report gains over no IRS, random coefficients, and fixed beamforming.Source: §I · p. 1

II. SYSTEM MODEL

The system is a multiuser IRS-aided OFDMA downlink with frequency-selective channels and time-varying IRS reflection coefficients. The base station allocates time-frequency resources and power within a channel coherence block.

II. SYSTEM MODEL

  • The model includes a single-antenna base station, K single-antenna users, and an IRS whose elements reflect signals from the base station to users.Signals reach users through direct and IRS-reflected links.Source: §II · pp. 1–2
  • The channels follow quasistatic block fading and remain approximately constant within a channel coherence block.Users receive N equal sub-bands and Q equal-sized time slots; channels are frequency-flat within each sub-band but may vary across sub-bands.Source: §II · p. 2
  • Because IRS coefficients cannot vary by sub-band, one common coefficient set must accommodate multiple users and frequency channels in each time slot.The proposed dynamic design varies coefficients across time slots, enabling artificial time-varying channels and generally serving fewer users simultaneously to improve passive beamforming.Source: §II · p. 2
  • The base station assigns each time-slot/sub-band resource block to at most one user and limits total transmit power across sub-bands in each time slot.The channel state information is used for sub-band and power allocation and dynamic IRS coefficient design.Source: §II · pp. 2–3
  • The achievable user rate is modeled over the assigned time-frequency resources, with a practical modulation-and-coding gap and independent Gaussian receiver noise.The exposition ignores rate loss from cyclic-prefix insertion.Source: §II · p. 3

III. PROBLEM FORMULATION

The problem jointly chooses IRS reflection coefficients, OFDMA resource-block assignments, and transmit powers to maximize the minimum user rate. Its coupled, non-convex structure makes the optimum generally difficult to obtain.

III. PROBLEM FORMULATION

  • The objective is to maximize the common rate, defined as the minimum achievable rate among all users..Source: §III · p. 3
  • The optimization jointly selects dynamic IRS coefficients, binary resource-block assignments, and power allocations at the base station.The resource and power variables are constrained by the formulation’s allocation and power requirements.Source: §III · p. 3
  • The problem is non-convex and combinatorial because assignment variables are binary, rates are nonconcave in IRS coefficients, and the variables are coupled.Consequently, obtaining an optimal solution is generally difficult.Source: §III · p. 3

IV. PROPOSED SOLUTION

The proposed solution uses alternating optimization to seek a suboptimal solution by updating the IRS coefficients and OFDMA allocation variables in turn.

IV. PROPOSED SOLUTION

  • An alternating optimization algorithm is used to find a suboptimal solution to the formulated problem.Source: §IV · p. 3
  • The algorithm divides the variables into IRS reflection coefficients and OFDMA resource allocation, including assignments and powers.Source: §IV · p. 3
  • At each iteration, it optimizes one variable set while holding the other fixed.Source: §IV · p. 3

1) OFDMA Resource Allocation Optimization Given IRS Coefficients:

With the IRS reflection coefficients fixed, the effective CNRs are fixed, reducing the joint problem to an OFDMA optimization over the remaining variables and constraints.

1) OFDMA Resource Allocation Optimization Given IRS Coefficients:

2) IRS Coefficients Optimization Given OFDMA Resource Allocation:

Given OFDMA resource allocation, the IRS coefficients are optimized using successive convex approximations within an alternating algorithm. The procedure uses multiple initializations, converges, and can generate artificial time-varying channels.

2) IRS Coefficients Optimization Given OFDMA Resource Allocation:

  • The IRS coefficients for each time slot are optimized with OFDMA resource allocation fixed, targeting the minimum user rate over a coherence block.Source: 2) IRS Coefficients Optimization Given… · p. 3
  • Because the IRS-coefficient subproblem remains non-convex, the method introduces auxiliary variables and applies successive convex approximation.The reformulation exploits the independence of OFDMA transmission across resource blocks.Source: 2) IRS Coefficients Optimization Given… · p. 3
  • The convex subproblem uses a first-order affine lower bound and is solved repeatedly to update the auxiliary variables until convergence.The approximation is tight, with matching gradients, at the current auxiliary-variable values.Source: 2) IRS Coefficients Optimization Given… · p. 4
  • The overall algorithm selects the highest common-rate solution among multiple random initializations, with non-decreasing objective values and guaranteed convergence to at least a locally optimal solution.Source: 2) IRS Coefficients Optimization Given… · p. 4
  • Slot-wise IRS coefficients can create optimized artificial time-varying channels, enhancing multiuser diversity and allowing more flexible OFDMA resource allocation.Source: 2) IRS Coefficients Optimization Given… · p. 4

V. NUMERICAL RESULTS

In simulations, dynamic passive beamforming achieves a higher common rate than fixed passive beamforming and random-phase schemes. It also produces distinct resource-allocation patterns by time slot.

V. NUMERICAL RESULTS

  • The simulation setup uses Q = 6 time slots, N = 16 sub-bands, and K = 3 users, and averages results over 100 independent channel realizations.Users are placed on a semi-circle around the IRS at distance d = 2 m, with D = 100 m.Source: §V · p. 4
  • The comparisons include fixed passive beamforming, two random-phase schemes, and a system without IRS.Fixed beamforming uses one IRS coefficient set across all Q time slots; the random-phase schemes differ in whether phase shifts are generated per coherence block or per time slot.Source: §V · p. 4
  • With dynamic passive beamforming, resource blocks are assigned to fewer users per time slot, sometimes to a single user, unlike the nearly invariant patterns with fixed beamforming or no IRS.The passage attributes this pattern to the difficulty of using non-frequency-selective IRS coefficients to serve all user channels at once.Source: §V · p. 5
  • Dynamic passive beamforming yields a higher common rate than fixed passive beamforming, while both channel-based designs outperform random-phase schemes as the number of reflecting elements increases.Random Phase II also outperforms Random Phase I.Source: §V · p. 5

VI. CONCLUSION

The conclusion reports improved multiuser common rate with IRS and describes dynamic passive beamforming as a way to induce artificial channel fading. It also notes that an alternating optimization algorithm finds a locally optimal solution.

VI. CONCLUSION

  • The proposed dynamic passive beamforming scheme induces artificial channel fading to improve passive-beamforming performance.The paper jointly optimizes IRS passive beamforming and OFDMA resource allocation to maximize multiuser common rate.Source: §VI · p. 5
  • An alternating optimization algorithm is proposed to find a locally optimal solution to the formulated problem.The conclusion says numerical results validated the algorithm’s effectiveness.Source: §VI · p. 5
  • IRS improves multiuser common rate, and dynamic beamforming uses more varied RB allocation with fewer users per slot to balance diversity and passive-beamforming gains.Fig. 5 plots common rate versus the number of reflecting elements M.Source: Figure 5 · p. 5
Loading 1912.01228v1…