Pith. sign in

REVIEW 2 major objections 1 minor 21 references

An autoencoder trained end-to-end on MIMO links with Rayleigh fading achieves lower bit error rates than conventional block-based designs.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.3

2026-06-29 20:55 UTC pith:V7PP2X6N

load-bearing objection Routine autoencoder application to MIMO under Rayleigh fading that asserts better BER but supplies no architecture, training details, or numbers to check the claim. the 2 major comments →

arxiv 2605.25458 v1 pith:V7PP2X6N submitted 2026-05-25 eess.SP

Deep Machine Learning in MIMO Communication Systems

Mohammad Reza Ghavidel Aghdam , Alireza Naghavi This is my paper
classification eess.SP
keywords MIMO communicationautoencodersdeep learningbit error rateRayleigh fadingend-to-end optimizationwireless systemssignal to noise ratio
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper aims to demonstrate that jointly optimizing the transmitter, receiver, and channel effects using an autoencoder framework in MIMO systems leads to better bit error rate performance under noise and fading. This matters because it suggests machine learning can replace separate hand-designed blocks with a single learned system that directly accounts for real-world impairments like Rayleigh fading. Simulations show the end-to-end approach outperforms traditional methods across different signal-to-noise ratios. A reader cares if this points to more reliable wireless links without complex separate processing stages.

Core claim

By incorporating the Rayleigh fading channel into the autoencoder, the communication system is trained to minimize bit error rate by simultaneously optimizing transmitter, receiver, and channel under noise and fading conditions, resulting in significantly lower BER than conventional block-based processing methods.

What carries the argument

The autoencoder framework that embeds the Rayleigh fading model to enable joint optimization of the entire MIMO communication chain.

Load-bearing premise

The Rayleigh fading channel model used in training accurately represents the real-world wireless conditions the system encounters.

What would settle it

Running the trained system over a measured wireless channel that deviates from Rayleigh fading and observing whether the BER advantage over block-based methods vanishes.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • The end-to-end system shows improved BER performance at various SNR levels.
  • Joint training handles channel impairments better than separate block designs.
  • Optimization tailored for deep learning-based MIMO reduces errors compared to traditional approaches.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the learned system generalizes, it could adapt to varying channel conditions without redesign.
  • Extensions might apply the same joint optimization to other modulation schemes or antenna configurations.
  • Real deployments could test whether the BER gains hold when the actual channel differs from the training model.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 1 minor

Summary. The paper proposes using autoencoders for end-to-end optimization of MIMO communication systems, jointly training the transmitter, receiver, and Rayleigh fading channel model to minimize BER, and reports that simulations demonstrate significantly lower BER than conventional block-based methods across SNR levels.

Significance. If the simulation results can be reproduced with full methodological disclosure, the work would add to the literature on learned physical-layer systems by extending autoencoder approaches to MIMO configurations under fading; however, the absence of any architecture, training, or comparison details prevents assessment of whether this constitutes a substantive advance.

major comments (2)
  1. [Abstract] Abstract: the central claim that the end-to-end system 'achieves significantly lower BER' rests entirely on unspecified simulation results; no neural-network architecture, loss function, optimizer, training procedure, MIMO antenna configuration, or quantitative baseline comparison is described, rendering the data-to-claim link unverifiable.
  2. [Abstract] Abstract (paragraph on channel incorporation): the assertion that incorporating the Rayleigh model allows the system to 'directly train ... to handle real world conditions' is presented without any analysis of model mismatch, generalization, or comparison to measured channels, leaving the scope of the claimed improvement unclear.
minor comments (1)
  1. [Abstract] The phrase 'novel optimization process tailored for deep learning-based MIMO communication' is introduced without any subsequent definition or pseudocode.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the comments highlighting the need for greater detail and clarity in the abstract. We will revise the manuscript accordingly to improve verifiability while preserving the core contribution.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that the end-to-end system 'achieves significantly lower BER' rests entirely on unspecified simulation results; no neural-network architecture, loss function, optimizer, training procedure, MIMO antenna configuration, or quantitative baseline comparison is described, rendering the data-to-claim link unverifiable.

    Authors: We agree that the abstract as written does not contain sufficient methodological information to allow independent verification of the BER claims. The current manuscript provides only high-level description. We will revise the abstract to include concise statements on the autoencoder architecture (e.g., layer counts and activation functions), the end-to-end loss, optimizer and training schedule, the specific MIMO antenna configuration used, and the conventional block-based baselines against which performance is compared. Full implementation details will also be added to the main text to support reproducibility. revision: yes

  2. Referee: [Abstract] Abstract (paragraph on channel incorporation): the assertion that incorporating the Rayleigh model allows the system to 'directly train ... to handle real world conditions' is presented without any analysis of model mismatch, generalization, or comparison to measured channels, leaving the scope of the claimed improvement unclear.

    Authors: We acknowledge that the abstract statement is not accompanied by supporting analysis of model mismatch or generalization beyond the Rayleigh assumption. In the revision we will qualify the language to indicate that the system is trained under the Rayleigh fading model and will add a short discussion (or simulation results) addressing the limitations of this model and the expected behavior under channel mismatch or measured channels. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical simulation claim stands on its own

full rationale

The paper reports simulation results from training an autoencoder-based end-to-end MIMO system that incorporates the Rayleigh channel model directly into the training loop to minimize BER, then compares the achieved BER against conventional methods across SNR values. No derivation chain, equations, or first-principles predictions are present that could reduce to fitted inputs or self-citations by construction. The performance claim is scoped explicitly to the described simulations, which is a standard, externally verifiable empirical procedure rather than a mathematical reduction. No load-bearing self-citation or ansatz smuggling is indicated in the provided text.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The paper rests on standard assumptions of deep learning training and the suitability of Rayleigh fading as a channel model; no new entities are introduced and no free parameters beyond network weights are named.

axioms (2)
  • domain assumption Gradient-based optimization of an autoencoder can jointly learn transmitter, channel, and receiver mappings that minimize BER
    Invoked when the abstract states the system is trained end-to-end to minimize BER under noise and fading
  • domain assumption Rayleigh fading plus noise is a sufficient model for the impairments the learned system must handle
    Explicitly stated in the abstract as the channel incorporated into the framework

pith-pipeline@v0.9.1-grok · 5659 in / 1151 out tokens · 21508 ms · 2026-06-29T20:55:06.425337+00:00 · methodology

0 comments
read the original abstract

This paper presents an innovative approach to enhancing machine learning based communication systems, specifically focusing on multiple-input multiple-output (MIMO) configurations using autoencoders. We optimize the transmitter, receiver, and channel simultaneously under conditions of noise and channel fading, aiming to minimize the bit error rate (BER). By incorporating the Rayleigh fading channel a widely recognized model for wireless channel impairments into the autoencoder framework, we directly train the communication system to handle real world conditions. We introduce a novel optimization process tailored for deep learning-based MIMO communication, and thoroughly analyze the resulting BER performance across various signal to noise ratio (SNR) levels. Our simulation results reveal that the proposed end-to-end wireless communication system achieves significantly lower BER compared to conventional block-based processing methods, highlighting its potential for more efficient and reliable wireless communication.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

21 extracted references · 2 canonical work pages · 2 internal anchors

  1. [1]

    Machine Learning Paradigms for Next-Generation Wireless Networks,

    C. Jiang, H. Zhang, Y. Ren, Z. Han, K.-C. Chen, and L. Hanzo, “Machine Learning Paradigms for Next-Generation Wireless Networks,” IEEE Wireless Communications, vol. 24, no. 2, pp. 98–105, 2016

    2016

  2. [2]

    Evolving deep gated recurrent unit using improved marine predator algorithm for profit prediction based on financial accounting information system,

    X. Li, M. Khishe, and L. Qian, “Evolving deep gated recurrent unit using improved marine predator algorithm for profit prediction based on financial accounting information system,” Complex & Intelligent Systems, pp. 1–17, 2023

    2023

  3. [3]

    Employing categorical boosting (catboost) and meta-heuristic algorithms for predicting the urban gas consumption,

    L. Qian, Z. Chen, Y. Huang, and R. J. Stanford, “Employing categorical boosting (catboost) and meta-heuristic algorithms for predicting the urban gas consumption,” Urban Climate, vol. 51, p. 101647, 2023

    2023

  4. [4]

    Breast cancer diagnosis using evolving deep convolutional neural network based on hybrid extreme learning machine technique and improved chimp optimization algorithm,

    L. Qian, J. Bai, Y. Huang, D. Q. Zeebaree, A. Saffari, and D. A. Zebari, “Breast cancer diagnosis using evolving deep convolutional neural network based on hybrid extreme learning machine technique and improved chimp optimization algorithm,” Biomedical Signal Processing and Control, vol. 87, p. 105492, 2024

    2024

  5. [5]

    Seb-choa: an improved chimp optimization algorithm using spiral exploitation behavior,

    L. Qian, M. Khishe, Y. Huang, and S. Mirjalili, “Seb-choa: an improved chimp optimization algorithm using spiral exploitation behavior,” Neural Computing and Applications, pp. 1–24, 2023

    2023

  6. [6]

    Solar photovoltaic model parameter identification using robust niching chimp optimization,

    Q. Bo, W. Cheng, M. Khishe, M. Mohammadi, and A. H. Mohammed, “Solar photovoltaic model parameter identification using robust niching chimp optimization,” Solar Energy, vol. 239, pp. 179–197, 2022

    2022

  7. [7]

    Machine Learning Methods for Optical Communication Systems,

    F. N. Khan, C. Lu, and A. P. T. Lau, “Machine Learning Methods for Optical Communication Systems,” in Signal Processing in Photonic Communications. Optical Society of America, 2017, pp. SpW2F–3

    2017

  8. [8]

    Physical Layer Security with its Applications in 5G Networks: A Review,

    L. Sun and Q. Du, “Physical Layer Security with its Applications in 5G Networks: A Review,” China Communications, vol. 14, no. 12, pp. 1–14, 2017. Deep ML−based SISO over Rayleigh Conventional SISO over Rayleigh Deep ML−based MIMO over Rayleigh Conventional MIMO over Rayleigh BER BER 7

    2017

  9. [9]

    Optimization of constraint engineering problems using robust universal learning chimp optimization,

    L. Liu, M. Khishe, M. Mohammadi, and A. H. Mohammed, “Optimization of constraint engineering problems using robust universal learning chimp optimization,” Advanced Engineering Informatics, vol. 53, p. 101636, 2022

    2022

  10. [10]

    Applications of Neural Networks to Digital Communications-A Survey,

    M. Ibnkahla, “Applications of Neural Networks to Digital Communications-A Survey,” Signal processing, vol. 80, no. 7, pp. 1185–1215, 2000

    2000

  11. [11]

    Communication Algorithms via Deep Learning

    H. Kim, Y. Jiang, R. Rana, S. Kannan, S. Oh, and P. Viswanath, “Communication Algorithms via Deep Learning,” arXiv preprint arXiv:1805.09317, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  12. [13]

    End-to-end deep learning of optical fiber communications,

    B. Karanov, M. Chagnon, F. Thouin, T. A. Eriksson, H. Bu¨low, D. Lavery, P. Bayvel, and L. Schmalen, “End-to-end deep learning of optical fiber communications,” Journal of Lightwave Technology, vol. 36, no. 20, pp. 4843–4855, 2018

    2018

  13. [14]

    Deep Learning Based Communication Over the Air,

    S. Do¨rner, S. Cammerer, J. Hoydis, and S. ten Brink, “Deep Learning Based Communication Over the Air,” IEEE Journal of Selected Topics in Signal Processing, vol. 12, no. 1, pp. 132–143, 2017

    2017

  14. [15]

    An Introduction to Deep Learning for the Physical Layer,

    T. O’Shea and J. Hoydis, “An Introduction to Deep Learning for the Physical Layer,” IEEE Transactions on Cognitive Communications and Networking, vol. 3, no. 4, pp. 563–575, 2017

    2017

  15. [16]

    Application of End-to-End Deep Learning in Wireless Communications Systems

    W. Lee, O. Jo, and M. Kim, “Application of End-to-End Deep Learning in Wireless Communications Systems,” arXiv preprint arXiv:1808.02394, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  16. [17]

    Energy-Efficient Joint Resource Allocation and Power Control for D2D Communications,

    Y. Jiang, Q. Liu, F. Zheng, X. Gao, and X. You, “Energy-Efficient Joint Resource Allocation and Power Control for D2D Communications,” IEEE Transactions on Vehicular Technology, vol. 65, no. 8, pp. 6119–6127, 2016

    2016

  17. [18]

    Learning to Optimize: Training Deep Neural Networks for Wireless Resource Management,

    H. Sun, X. Chen, Q. Shi, M. Hong, X. Fu, and N. D. Sidiropoulos, “Learning to Optimize: Training Deep Neural Networks for Wireless Resource Management,” pp. 1–6, 2017

    2017

  18. [19]

    Deep Power Control: Transmit Power Control Scheme Based on Convolutional Neural Network,

    W. Lee, M. Kim, and D.-H. Cho, “Deep Power Control: Transmit Power Control Scheme Based on Convolutional Neural Network,” IEEE Communications Letters, vol. 22, no. 6, pp. 1276–1279, 2018

    2018

  19. [20]

    Power Control for D2D Underlaid Cellular Networks: Modeling, Algorithms, and Analysis

    N. Lee, X. Lin, J. G. Andrews, and R. W. Heath Jr, “Power Control for D2D Underlaid Cellular Networks: Modeling, Algorithms, and Analysis.” IEEE Journal on Selected Areas in Communications, vol. 33, no. 1, pp. 1–13, 2015

    2015

  20. [21]

    Linear Pre-Coding Performance in Measured Very-Large MIMO Channels

    X. Gao, O. Edfors, F. Rusek, and F. Tufvesson, “Linear Pre-Coding Performance in Measured Very-Large MIMO Channels.” pp. 1–5, 2011

    2011

  21. [22]

    Robust Decentralized Joint Precoding Using Team Deep Neural Network,

    P. de Kerret and D. Gesbert, “Robust Decentralized Joint Precoding Using Team Deep Neural Network,” pp. 1–5, 2018

    2018