Wideband MIMO-OFDM Systems

• Wideband MIMO-OFDM is a technology that combines multi-antenna transmission with OFDM to partition wideband channels into efficient narrowband subcarriers.
• Advanced detection and channel estimation techniques, including optimized coordinate descent and sparse methods, enhance spectral efficiency and reduce computational complexity.
• Hardware-efficient implementations, adaptive pilot scheduling, and integration with sensing and intelligent surfaces optimize performance for next-generation wireless networks.

Wideband MIMO-OFDM refers to the use of multi-input multi-output (MIMO) transmission over wideband channels employing orthogonal frequency-division multiplexing (OFDM). The combination of these technologies constitutes the backbone of current and emerging high-data-rate wireless systems, particularly in scenarios with dense multipath, high spatial multiplexing demands, and stringent hardware constraints.

1. Fundamentals of Wideband MIMO-OFDM Systems

A wideband MIMO-OFDM system consists of multiple transmit and receive antennas communicating over a highly frequency-selective channel, using OFDM to partition the wideband spectrum into narrowband subcarriers. Each OFDM subcarrier experiences flat-fading, whereas the overall channel remains frequency selective. Spatial multiplexing across antennas and subcarriers is key to high spectral efficiency and robust performance in rich-scattering or high-pathloss environments.

Channel Model: On subcarrier $n$, the received signal at the base station (BS) is

$$\mathbf{y}_n = \mathbf{H}_n \mathbf{x}_n + \mathbf{n}_n$$

where $\mathbf{H}_n \in \mathbb{C}^{B \times U}$, with $B \gg U$ in massive MIMO, and $\mathbf{x}_n \in \mathcal{O}^U$ denotes the $U$-user symbol vector drawn from a QAM constellation $\mathcal{O}$, typically with unit power $E[|s|^2]=1$. Noise $\mathbf{n}_n$ is modeled as $\mathcal{CN}(0, N_0 \mathbf{I}_B)$. OFDM parameters include the total number of subcarriers $N$, occupied data subcarriers $W$, and CP length sufficient for the channel delay spread (Wu et al., 2016).

2. Advanced Receiver and Detection Architectures

The computational complexity of optimal MIMO-OFDM data detection has driven research toward scalable, high-throughput linear (e.g., MMSE, ZF) and non-linear (e.g., box-constrained) equalizers.

2.1 Coordinate Descent Data Detection

Coordinate descent (CD) solves both MMSE and box-constrained detection by iteratively updating one symbol or coordinate at a time. For the MMSE case:

$$\mathbf{x}_n^{\mathrm{MMSE}} = \arg\min_{\mathbf{x} \in \mathbb{C}^U} \|\mathbf{y}_n - \mathbf{H}_n \mathbf{x}\|_2^2 + N_0 \|\mathbf{x}\|_2^2, \qquad \mathbf{x}_n^{\mathrm{MMSE}} = \mathbf{G}_n^{-1} \mathbf{z}_n$$

with $$\mathbf{G}_n = \mathbf{H}_n^H \mathbf{H}_n + N_0 \mathbf{I}_U$$ and $\mathbf{z}_n = \mathbf{H}_n^H \mathbf{y}_n$. For QAM, the box-constrained problem further enforces that each coordinate lies within the convex hull $\mathcal{C}$ of $\mathcal{O}$.

An optimized CD (OCD) algorithm avoids matrix inverses, relies on precomputed step sizes, and maintains stateful residuals to minimize recomputation. FPGA implementations achieve single-instance throughputs of 376 Mb/s (Xilinx Virtex-7, 128×8 MIMO, 64-QAM), scaling linearly when pipelined, and outperforming conventional Neumann, CG, or TASER-based designs in throughput per LUT by factors of 2.6–6× (Wu et al., 2016).

2.2 Data Detection with 1-Bit ADCs

Advances in ADC technology have enabled direct RF-sampling architectures with 1-bit quantization. Signal reconstruction is performed after sign quantization using Bussgang decompositions to analyze quantization effects and devise ZF-combiners:

$$\mathrm{EVM} = 100\%\, \sqrt{ \frac{1}{SU E_s} \sum_{k \in \mathcal{S}} \operatorname{tr} [\mathbf{W}_k^\mathrm{ZF}(N_0\mathbf{G}^2 + \mathbf{C}_{\widehat e_k})\mathbf{W}_k^{\mathrm{ZF},H} ] }$$

This supports 16-QAM and 64-QAM at moderate SNRs and provides concrete hardware simplification recommendations (Jacobsson et al., 2019).

3. Channel Estimation and Acquisition in Wideband MIMO-OFDM

Efficient channel estimation in wideband, high-mobility environments with massive arrays must exploit spatial-frequency structure and sparsity.

3.1 Angle–Delay Domain Sparsity and APSPs

Massive MIMO-OFDM channels are sparse in the joint angle-delay domain. Adjustable phase-shift pilots (APSPs) leverage this sparsity to cyclically embed each user’s "angle–delay channel power matrix" (ADCPM) into non-overlapping frequency slots. Given the ADCPM support $\Omega_k$ for user $k$, optimal pilot scheduling seeks assignments $\{\phi_k\}$ so that cyclic shifts $\Omega_k^{(\phi_k)}$ are non-overlapping for all users, minimizing multiuser interference during pilot-aided MMSE estimation. APSPs can cut pilot overhead by 33–66% over conventional phase-shift orthogonal pilot (PSOP) schemes and preserve error-rate and spectral efficiency performance (You et al., 2015).

3.2 Sparse Channel Estimation and Beam Squint

At mmWave and THz, "beam squint" causes the array response to vary across the OFDM subcarriers due to large inter-element delays relative to the OFDM symbol period. True-time-delay (TTD) elements per virtual subarray mitigate squint, while generalized sparse OMP (GSOMP) channel estimation jointly recovers support and gain across subcarriers, achieving near-CRLB NMSE with as little as 0.8$N_B$ pilots. Hybrid analog–digital combining using TTD can reach within 1–2 dB of fully-digital beamforming rates, even across 40 GHz (Dovelos et al., 2021) and is amenable to GNN-based low-complexity inference (Li et al., 27 Jan 2025).

3.3 Block-Sparse and Multidimensional Approaches

Channel models with block-sparsity in the delay–angle domain (as in Saleh–Valenzuela) enable parametric multidimensional (4D/3D/2D-ESPRIT) prediction and acquisition. Joint exploitation of block-sparsity and codebook design yields improved estimation NMSE and BER at reduced pilot loads (Wang et al., 2022, Adeogun et al., 2014).

4. Hardware-Efficient Architectures and Practical Implementation

4.1 Pipelined and Parameterizable VLSI

Optimal hardware efficiency for MIMO-OFDM data detection is achieved via deeply pipelined and parameterizable FPGA architectures. Fully shared arithmetic units (e.g., parallel inner product/adder trees, reciprocal units), fixed-point path optimizations, and configurable levels of pipeline interleaving (e.g., 24-way across subcarriers) enable clock rates upwards of 258 MHz and per-pipeline throughputs exceeding 370 Mb/s for full-rate high-order QAM (Wu et al., 2016).

4.2 Wideband Joint Beamforming and SI Cancellation

Emerging wideband full-duplex (FD) MIMO-OFDM systems require joint analog and digital cancellation of self-interference (SI) spanning the dispersive channel. Multi-tap wideband analog cancellers use programmable MUX/DEMUX taps, and adaptive digital cancellation is realized via truncated SVD (TSVD) regularized least-squares. Hardware reuse and TSVD mode reduction sharply decrease complexity and training required for SI suppression (up to 95 dB) over prior approaches (Islam et al., 2021).

5. Extensions: ISAC, Sensing, and Multifunctional Wideband MIMO-OFDM

Wideband MIMO-OFDM is foundational for integrated sensing and communication (ISAC), enabling simultaneous high-resolution positioning and information delivery.

5.1 Ultra-Wideband Sensing and Ambiguity Resolution

At THz and mmWave, the large available bandwidth raises range resolution to the cm scale but introduces delay ambiguity due to limited subcarrier count and increased subcarrier spacing. A two-stage ML estimation (coarse–fine) over carefully allocated contiguous and uniform subcarriers resolves ambiguity and secures mm-level accuracy even at low SNR (Bacchielli et al., 2024). OFDM radar beampattern and detection performance can be jointly optimized with communication sum-rate and PAPR constraints via SDR (Hu et al., 2021).

5.2 Intelligent Surfaces and Metasurfaces

RIS and stacked intelligent metasurfaces (SIM) have been integrated into wideband MIMO-OFDM to achieve physically-diagonalized multicarrier channels through programmable, cascaded phase layers, drastically reducing RF-chain usage and hardware complexity. With 7-layer SIMs, channel capacity can exceed single-frequency metasurface designs by over 300%, with power scaling laws allowing for aggressive reductions in TX power per user as antenna/RIS element count grows (Li et al., 1 Mar 2025, Chen et al., 2024, Li et al., 26 Sep 2025).

6. Robustness to Hardware Impairments and Security Extensions

6.1 Low-Resolution Quantization

Wideband MIMO-OFDM with low-resolution ADCs (1–2 bits) is now practically realizable at the BS, with careful signal processing (Bussgang decompositions, dithering) achieving error-vector-magnitude (EVM) compliant with high-order QAM needs within expected SNR ranges (Jacobsson et al., 2019, Chen et al., 2024).

6.2 Security and Joint Optimization

Cooperative double-IRS extends physical-layer security in wideband MIMO-OFDM by jointly optimizing transmit precoding and frequency-flat IRS phases via product Riemannian gradient descent. This approach addresses multi-subcarrier coupling and delivers up to 32%–60% secrecy sum-rate gains over single/distributed IRS and AO/Euclidean GD benchmarks even with finite channel state information error (Xiong et al., 27 Jan 2026).

7. Performance Benchmarks and Future Directions

• Properly designed wideband MIMO-OFDM systems can attain multi-Gb/s throughputs with FPGA hardware, while spectral efficiency and error rates approach theoretical bounds across massive MIMO, hybrid, or metasurface-based regimes.
• Significant future work includes the joint design of wideband metasurface hardware with frequency-selective control, practical deployment of data-driven GNN-based hybrid beamforming, and the integration of ultra-reliable, low-latency MIMO-OFDM frameworks in ISAC and 6G systems.

---

References:

• (Wu et al., 2016): High-Throughput Data Detection for Massive MU-MIMO-OFDM using Coordinate Descent
• (Jacobsson et al., 2019): Massive MU-MIMO-OFDM Uplink with Direct RF-Sampling and 1-Bit ADCs
• (You et al., 2015): Channel Acquisition for Massive MIMO-OFDM with Adjustable Phase Shift Pilots
• (Dovelos et al., 2021): Channel Estimation and Hybrid Combining for Wideband Terahertz Massive MIMO Systems
• (Bacchielli et al., 2024): Bistatic Sensing at THz Frequencies via a Two-Stage Ultra-Wideband MIMO-OFDM System
• (Islam et al., 2021): Joint Analog and Digital Transceiver Design for Wideband Full Duplex MIMO Systems
• (Li et al., 1 Mar 2025): Stacked Intelligent Metasurfaces-Enhanced MIMO OFDM Wideband Communication Systems
• (Chen et al., 2024): Performance Analysis on RIS-Aided Wideband Massive MIMO OFDM Systems with Low-Resolution ADCs
• (Li et al., 26 Sep 2025): Stacked Intelligent Metasurface-Enhanced Wideband Multiuser MIMO OFDM-IM Communications
• (Xiong et al., 27 Jan 2026): Cooperative Double IRS aided Secure Communication for MIMO-OFDM Systems