MIMO Assisted Networks Relying on Intelligent Reflective Surfaces

I. INTRODUCTION

• IRSs can assist coverage-hole users when direct BS-user links are blocked, while reflected signals are phase-aligned to enhance reception or mitigate interference.The surfaces use many reflective elements whose phase shifts and possibly amplitudes can be adjusted.Source: §I · pp. 2–3
• The study targets the limited analysis of user-location effects in IRS-aided MIMO networks by modeling spatial randomness with stochastic geometry.The proposed framework provides a mathematical model for multiple-user spatial randomness.Source: §I · p. 3, A. Motivations and Contributions · p. 4

A. Motivations and Contributions

• The proposed design jointly uses BS active beamforming, user detection vectors, and IRS passive beamforming to serve multiple users and suppress intra-cell interference.The design is intended to work in both IRS and non-IRS scenarios.Source: A. Motivations and Contributions · p. 4
• The paper derives high-SNR channel-statistics distributions, closed-form outage probabilities, ergodic rates, diversity orders, SEs, and EEs for the MIMO-IRS network.The analytical development includes both analytical and asymptotic outage results and high-SNR ergodic-rate slopes.Source: A. Motivations and Contributions · p. 4
• Simulation results indicate that spatial gain improves performance, IRS-user fading has modest impact, and increasing BS antennas or IRS elements significantly improves SE and EE.The reported conclusions also include agreement between approximations and simulations for the low-complexity analytical method.Source: A. Motivations and Contributions · p. 5

B. Organization and Notations

• The paper proceeds from the system model to analytical results, numerical verification, and conclusions, with Table II listing critical notation.Source: B. Organization and Notations · p. 5
• The notation section defines matrix transpose, conjugate transpose, rank, trace, probability, expectation, distributions, Gamma functions, and the Meijer-G function.Source: B. Organization and Notations · p. 5

II. SYSTEM MODEL

• The system model equips the BS and users with M transmit and K receive antennas, respectively, while N co-located IRS elements serve M users under N ≥ K ≥ M.The IRS elements are positioned at the center of the disc.Source: §II · p. 5

TABLE OF NOTATIONS

• The paper states that inter-cell interference significantly affects performance, but multi-cell analysis is outside the scope of the treatise.Source: TABLE OF NOTATIONS · p. 6

A. System Description

• Users are randomly distributed on a radius-R disc according to a binomial point process, with blocked direct BS-user links and product-distance path loss.Distances are simplified as exceeding 1 meter, while r0 avoids a small-distance singularity.Source: A. System Description · p. 7
• The BS-IRS and IRS-user channels are represented by Nakagami-fading matrices, with fading parameters describing the corresponding links.The global CSI is assumed perfectly known at the IRSs and users.Source: A. System Description · p. 7
• The IRS reflection matrix contains controllable phase shifts and amplitude coefficients, while the active beamforming matrix is set to the identity for IRS and non-IRS access scenarios.The amplitude coefficients satisfy βn ∈ (0, 1].Source: A. System Description · p. 8

B. Passive Beamforming Design

• The passive-beamforming design stacks effective BS-IRS-user channels and selects IRS coefficients to maximize channel gain while satisfying phase and amplitude constraints.Solutions require sufficient IRS elements, with no passive-beamforming solution when N < MK and infinitely many homogeneous solutions when N > MK.Source: B. Passive Beamforming Design · pp. 8–10
• User detection vectors are designed alongside passive beamforming to mitigate the strong intra-cell interference caused by superimposed signals.The resulting effective channel gains are expressed through the detection vector, user channel matrix, and IRS reflection matrix.Source: B. Passive Beamforming Design · pp. 9–10
• The normalized amplitude-coefficient construction may not represent the optimized solution because the normalization factor can satisfy βmax ≥ 1.Source: B. Passive Beamforming Design · p. 10

C. Detection Vector Design

• The resulting detection vectors can mitigate intra-cell interference and operate in both IRS and non-IRS scenarios.Source: C. Detection Vector Design · p. 11
• Zero-forcing-based detection removes the m-th effective-channel column and constrains the detection vector to the null space of the remaining users’ channels.The reduced matrix contains K × (M − 1) elements.Source: C. Detection Vector Design · p. 11
• The design applies maximal-ratio combining after obtaining the detection vector from the left singular vectors of the effective channel.Source: C. Detection Vector Design · p. 11
• The left-singular-vector solution requires K ≥ M; otherwise, the proposed detection-vector solution does not exist.More powerful but more complex active beamformers could reduce the required number of receive antennas, but this is outside the paper’s scope.Source: C. Detection Vector Design · p. 11
• Using the designed active beamforming and downlink user-detection vectors, the BS transmit signal obeys the maximum transmit power constraint, and user m’s post-detection SNR can be written.The SNR expression includes σ², denoting the AWGN power.Source: C. Detection Vector Design · p. 12

III. PERFORMANCE ANALYSIS OF MIMO-IRS NETWORK

• The performance analysis derives high-SNR effective-channel statistics, outage probabilities, ergodic rates, spectral efficiency, and energy efficiency for the proposed network.Source: §III · p. 12

A. Channel Statistics

• The high-SNR effective-channel-gain PDF is derived for a network where N IRS elements simultaneously serve M users with N ≥ K ≥ M.The derivation assumes i.i.d. channel elements with distinct fading parameters t1 and t2.Source: A. Channel Statistics · p. 13

B. Outage Probability

• A closed-form outage-probability expression is obtained for user m when the IRS array is centered in the disc and serves M users.The expression includes Gauss hypergeometric functions and depends on the target rate Rm.Source: B. Outage Probability · pp. 13–14
• For sufficiently many IRS elements, the paper derives an upper bound on the approximated outage probability and a high-SNR asymptotic expression.The hypergeometric form makes direct engineering interpretation challenging, motivating the asymptotic analysis.Source: B. Outage Probability · p. 14
• The diversity order is affected by both BS-IRS and IRS-user fading links, while increasing the number of IRS elements reduces outage probability.Source: B. Outage Probability · p. 15
• For M = 1, the minimal diversity order is 2tsN; when M = K = t2 = 1 and t1 →∞, the diversity order is N.When t1 = N = M = K = 1 and t2 > 1, the minimum diversity order is 2. The diversity order approaches dm = 2N.Source: B. Outage Probability · pp. 15–16

C. Ergodic Rate

• A classic Gamma distribution is assumed to obtain an approximated channel-gain distribution for the m-th user.The approximation uses the parameter th = (1 + t1 + Qt2) / (t1t2).Source: C. Ergodic Rate · p. 16
• The paper derives a lower bound on the achievable ergodic rate using exact expressions represented with Meijer-G functions.Source: C. Ergodic Rate · p. 16
• The high-SNR slope of the ergodic rate is one and is unaffected by the number of IRS elements, antennas, or fading environments.Source: C. Ergodic Rate · p. 17

D. SE and EE

• A tractable spectral-efficiency expression is derived from the ergodic rates of the proposed MIMO-IRS network.Source: D. SE and EE · p. 17
• The energy-efficiency expression combines the derived spectral efficiency with modeled total power consumption.The model includes BS transmit-power efficiency, user power consumption, and IRS-element power consumption.Source: D. SE and EE · p. 18

Validation

• Analytical and simulation results agree closely in the high-SNR regime, although gaps are large outside that regime because the analysis uses high-SNR approximations.The evaluations use Monte Carlo simulations and compare analytical with asymptotic results.Source: §IV · pp. 18–19

Outage probability

• The outage probability improves as IRS elements increase, with diversity orders of 4 for N = 2 and 6 for N = 3.The curves' slopes also increase with the number of IRS elements, consistent with the diversity-order analysis.Source: §IV · p. 18
• Stronger LoS environments typically decrease outage probability, while the BS-IRS fading parameter mainly governs outage behavior and curve slope.The IRS-user fading parameter has almost no effect on outage probability in the evaluated setting.Source: §IV · pp. 19–20

Ergodic rate

• The ergodic-rate high-SNR slope is one, and adding IRS elements significantly increases ergodic rate without changing that slope.LoS links between the BS and IRS improve ergodic rate, whereas IRS-user LoS links have nearly no effect.Source: §IV · p. 20

Comparison setup

• The MIMO-IRS network is evaluated against half-duplex AF- and DF-relay setups under identical total transmit power and single-antenna relay configurations.The relay comparisons use rate expressions derived for equal-duration two-phase transmission.Source: §IV · pp. 20–22

Spectral efficiency

• For N = 15, the proposed MIMO-IRS network outperforms the AF-relay and DF-relay setups, becoming more competitive as IRS elements increase.The rate difference between the three setups becomes smaller with more IRS elements, under equal total transmit power.Source: §IV · p. 22

Throughput and spectral efficiency

• Network throughput increases monotonically with the numbers of TAs and IRS elements, but spectral efficiency is zero when N < MK because passive beamforming weights have no solution.More TAs allow the BS to serve more users simultaneously.Source: §IV · p. 22

Energy efficiency

• Energy efficiency improves as the numbers of RAs, TAs, and IRS elements increase, and its slope can likewise be improved by increasing these quantities.The reported explanation is increased spatial diversity gain.Source: §IV · p. 22

• Joint detection vectors and IRS passive beamforming are designed for the proposed MIMO-IRS network.Source: §V · p. 22
• Closed-form results cover channel characteristics, outage probabilities, ergodic rates, spectral efficiency, and energy efficiency.Source: §V · p. 22
• Multiple IRS arrays distributed across different building facades are proposed as a future research direction.Source: §V · p. 22
• Modeling users within a restricted geographic area is identified as another future direction for IRS stochastic-geometry research.Source: §V · p. 22

APPENDIX A: PROOF OF LEMMA 1

• The high-SNR effective-channel PDF and CDF are obtained by transforming products of Nakagami variables with Laplace methods.Source: Appendix A · p. 24

APPENDIX B: PROOF OF THEOREM 1

• The outage probability is derived from the channel statistics and user-distance distribution, yielding a closed-form expression in (32).Source: Appendix B · pp. 24–25

APPENDIX C: PROOF OF COROLLARY 1

• An asymptotic outage result is obtained by expanding the hypergeometric function for bR^α < 1.Source: Appendix C · p. 25

APPENDIX D: PROOF OF LEMMA 2

• Zero-forcing detection vectors are used to derive the mean and variance of the effective channel gain.Source: Appendix D · p. 26

APPENDIX E: PROOF OF THEOREM 3

• The ergodic-rate derivation begins from the rate definition and substitutes the channel CDF into the resulting integral.Source: Appendix E · pp. 26–27
• The upper incomplete Gamma function is transformed into a Meijer-G function to obtain a tractable rate expression.Source: Appendix E · p. 27
• The resulting expression gives the ergodic rate of the m-th user in (40).Source: Appendix E · p. 27