Source-linked AI summary
Spectral-Efficient MIMO-OFDM: Low-Complexity Solution based on Random Multiplexing
Jie Yang, Wanchen Hu, Yi Song, Shuangyang Li, Burak Çakmak, Lei Liu, Xin Wang, Giuseppe Caire
TL;DR
MIMO-OFDM needs higher spectral efficiency without sacrificing practical complexity or compatibility with existing processing. The paper adds randomized frequency-domain compression and OAMP detection to conventional MIMO-OFDM, and reports improved mutual information and BER in compressed operation, including β = 0.75.
Problem
Achieving high spectral efficiency together with low-complexity, high-performance detection is challenging under practical MIMO-OFDM constraints.
Method
The scheme applies randomized frequency-domain linear compression before conventional MIMO-OFDM and uses OAMP to recover the information symbols.
Results
For β = 0.75, the proposed scheme approaches Gaussian-input capacity before saturation and improves BER compared with conventional MIMO-OFDM.
Takeaways & Limitations
The modular add-on preserves standard 5G procedures, including SVD-based precoding and per-subcarrier equalization, while supporting low-complexity signal recovery.
Abstract
from arXiv · showhide
This paper presents a low-complexity precoded MIMO-OFDM system for achieving improved spectral efficiency (SE) via intentionally compressing information symbols among subcarriers. Particularly, the proposed scheme leverages the powerful random multiplexing mechanism for precoding, and adopts the linear-complexity orthogonal approximate message passing (OAMP) estimator for symbol detection, where the compatibility with the existing fifth generation (5G) architectures is fully preserved. We further provide the theoretical analysis based on the replica-symmetric (RS) formula. This analysis confirms the advantages of the proposed system with respect to the adopted compression ratios, where an interesting phase transition behavior is verified. Numerical results coincide with our analysis and demonstrate significant improvements in terms of achievable rates and bit error rate (BER) compared to conventional MIMO-OFDM counterpart, making the proposed scheme a promising solution to 6G and beyond wireless networks.
I. Introduction
The paper targets higher spectral efficiency in MIMO-OFDM under practical bandwidth, antenna, and complexity constraints. It proposes randomized frequency-domain compression with OAMP detection while preserving compatibility with conventional architectures.
- I. Introduction: Higher spectral efficiency remains difficult to achieve alongside low-complexity, high-performance detection under practical MIMO-OFDM constraints.The stated constraints include limited bandwidth, finite antennas, and hardware complexity.
- I. Introduction: Existing AMP-family analyses generally assume i.i.d. or right-unitarily invariant channels, assumptions that practical channel distributions may violate.The related work identifies possible performance degradation when these assumptions do not hold.
- I. Introduction: The proposed SE-MIMO-OFDM introduces randomized linear transformation and compression in the frequency domain before conventional MIMO-OFDM transmission.This reshapes the effective channel toward a statistically isotropic form and removes dominant signal-space directions.
- I. Introduction: The transformed detection problem decouples into equivalent scalar Gaussian channels with identical effective SNR, enabling replica-based fixed-point analysis.The analysis uses self-consistent equations derived through the replica method.
- I. Introduction: An OAMP estimator refines frequency-domain outputs while preserving standard MIMO-OFDM signaling, SVD-based precoding, and existing processing techniques.The scheme is presented as a modular add-on compatible with current 5G and emerging 6G systems.
A. Conventional MIMO-OFDM System
Conventional MIMO-OFDM maps spatially precoded symbols across subcarriers and processes each subcarrier through parallel spatial layers. After equalization, each resource element has a layer- and subcarrier-dependent effective SNR.
- A. Conventional MIMO-OFDM System: A standard MIMO-OFDM system uses Nf subcarriers and maps transmitted data onto Ns ≤ min{Nr, Nt} spatial layers.The frequency-domain symbol matrix is X ∈ C^Nf×Ns, and spatial and frequency dimensions are processed separately.
- A. Conventional MIMO-OFDM System: SVD-based spatial precoding decomposes each subcarrier channel as Hk = ΓkΛkVk^H and uses the first Ns columns of Vk for precoding.This is the conventional spatial-precoding workflow described for each subcarrier.
- A. Conventional MIMO-OFDM System: After modulation, cyclic-prefix insertion, removal, and FFT demodulation, the received signal follows ȳk = HkPkx̄k + z̄k.The noise is modeled as AWGN with covariance N0I_Ns.
- A. Conventional MIMO-OFDM System: Receiver processing transforms the MIMO channel into Ns parallel spatial layers, with each layer represented by yk,i = λk,ixk,i + zk,i.Here λk,i is the corresponding singular value and zk,i is effective noise.
- A. Conventional MIMO-OFDM System: The effective SNR is defined as 1/σk,i^2/N0 and fluctuates across subcarriers and spatial layers.Frequency-selective fading and spatial-eigenmode power variations create this heterogeneity.
B. Proposed SE-MIMO-OFDM Model
SE-MIMO-OFDM adds frequency-domain compression and OAMP detection to conventional MIMO-OFDM. A channel-agnostic semi-unitary random matrix maps information symbols onto fewer subcarriers while preserving compatibility with existing architectures.
- The scheme introduces two additional modules into standard MIMO-OFDM to enhance spectral efficiency.
- A frequency-domain compression layer maps N information symbols onto Nf subcarriers, with Nf ≤ N.
- The compression matrix U is Haar-distributed, semi-unitary, and channel-agnostic, requiring no channel state information.
- The effective channel A ≜ ΛU combines frequency-selective channel effects with compression, after which OAMP recovers the information vector s.
- The framework preserves existing physical-layer architectures and MIMO processing through modular compression and OAMP add-on modules.
- The compression ratio β provides a throughput–reliability trade-off, while heterogeneous eigen-subchannel SNRs motivate adaptive β and per-layer modulation and coding.
III. Replica Prediction of Mutual Information
The paper analyzes asymptotic mutual information using replica-symmetric formulas and interprets the resulting effective SNR through an equivalent scalar Gaussian channel. Numerical evaluation reveals a compression-dependent phase transition that is mitigated by higher input SNR.
- The asymptotic normalized input-output mutual information is evaluated in the large-system limit N, Nf →∞.
- Replica-symmetric analysis expresses the mutual information through an optimization involving the relevant fixed-point quantities.
- The global-minimum parameter ρ⋆ represents the effective SNR of an asymptotically equivalent scalar Gaussian channel.
- The fixed-point iteration computes the mutual information in the algorithmic phase with a unique RS solution and coincides with OAMP/VAMP state evolution.
- When β falls below a threshold, effective SNR ρ degrades sharply, producing significant signal-recovery loss.
- Increasing input SNR alleviates the phase transition, enabling reliable recovery at lower compression ratios.
A. OAMP Algorithm
OAMP recovers compressed high-dimensional signals through alternating linear estimation and nonlinear constellation-aware denoising. Its structured effective channel enables low-complexity implementation while maintaining Gaussian error decoupling across iterations.
- OAMP iteratively recovers the signal s from compressed observations y using linear estimation and nonlinear estimation stages.
- The linear-estimation step decouples mixed signals and produces an extrinsic estimate for nonlinear estimation.
- The linear stage uses LMMSE filtering to suppress additive noise and inter-symbol interference, then forms a de-biased AWGN-corrupted observation.
- The nonlinear-estimation step exploits the discrete constellation to denoise observations and project posterior means onto the valid signal space.
- Extrinsic updates remove self-feedback and refresh the prior estimate and variance for the next iteration.
B. Complexity Analysis
The proposed SE-MIMO-OFDM structure reduces OAMP detection complexity from cubic to linear scaling by exploiting semi-unitary compression and diagonalized noise covariance.
- Complexity bottleneck: OAMP's conventional LMMSE inversion requires O(N_f^3) complexity, with matrix-vector multiplications adding O(N_fN) per iteration.This cubic inversion can become prohibitive for large-scale systems.
- Structural simplification: The semi-unitary property AAH = Λ^2 and diagonalized noise covariance Σ simplify the inversion in the proposed scheme.These structures enable a closed-form linear-estimation step.
- Resulting complexity: The proposed inversion complexity collapses from O(N_f^3) to O(N_f), while the remaining linear transformation is efficiently implementable.The nonlinear estimator contributes O(N|S|) complexity because it operates independently symbol by symbol.
- Resulting complexity: Overall, the proposed SE-MIMO-OFDM achieves per-iteration complexity that scales linearly with system dimension.The reduction is compared with conventional OAMP implementations and is presented as suitable for practical wireless systems.
V. Numerical Results
Numerical evaluations compare SE-MIMO-OFDM with conventional MIMO-OFDM under QPSK transmission across multiple MIMO settings. The proposed scheme improves mutual information and BER, including under compression, while transmitting more information bits.
- Experimental setup: The numerical study uses QPSK transmission and conventional MIMO-OFDM as the baseline under a Saleh-Valenzuela mmWave channel with N_f = 1024.The channel uses L = 10 multipaths and parameters generated according to 3GPP 38.901.
- Mutual information: For 2 × 16 MIMO, β = 0.75 produces a mutual-information gain at low-to-moderate SNR and approaches Gaussian-input capacity before saturating at log2 |S|/β.Conventional OFDM saturates at log2 |S| bits per symbol, causing shaping loss at high SNR.
- Mutual information: The semi-unitary transformation couples subcarriers and redistributes excess capacity from strong subcarriers to weaker ones, reducing shaping loss at moderate SNR.The mapping uses a Haar-distributed matrix U to map N symbols onto N_f < N subcarriers.
- BER performance: In 1 × 16 SIMO and 4 × 16 MIMO, OAMP-based SE-MIMO-OFDM significantly outperforms the conventional LMMSE baseline even when β = 1.The figures evaluate β ∈ {0.75, 1}, with N_s = 1 and N_s = 4 data layers, respectively.
- Detection mechanism: The Haar structure makes A = ΛU right-rotationally invariant, enabling OAMP to decouple detection into equivalent scalar Gaussian channels and exploit constellation priors.The linear-estimation and nonlinear-denoising steps alternate to achieve the Bayes-optimal performance predicted by the replica method.
- BER performance: At β = 0.75, the proposed scheme improves BER over conventional MIMO-OFDM while transmitting more information bits.The conventional system faces difficult information recovery in the compressed case because of low-rank measurements.
VI. Conclusion
The paper presents a compatible add-on framework that improves conventional MIMO-OFDM spectral-efficiency limits without overhauling the physical-layer architecture. It combines frequency-domain linear transformation with low-complexity OAMP recovery, supported by asymptotic analysis and numerical results.
- Conclusion: The framework embeds supplementary data into available subcarriers through a frequency-domain linear transformation while preserving standard 5G and future 6G procedures.The preserved procedures include SVD-based precoding and per-subcarrier equalization.
- Conclusion: OAMP enables reliable signal recovery with low computational complexity by exploiting the structural properties of the proposed precoding.The paper presents this balance between spectral-efficiency gains and engineering feasibility as a practical pathway for future wireless networks.