pith. sign in

arxiv: 1907.05753 · v1 · pith:Z7RIOOH3new · submitted 2019-07-12 · 📡 eess.SP · cs.NI

Secrecy Analysis and Learning-based Optimization of Cooperative NOMA SWIPT Systems

Furqan Jameel , Wali Ullah Khan , Zheng Chang , Tapani Ristaniemi , Ju Liu This is my paper

Pith reviewed 2026-05-24 22:16 UTC · model grok-4.3

classification 📡 eess.SP cs.NI
keywords cooperative NOMASWIPTintercept probabilitydeep learning optimizationpower allocationsecrecyenergy harvestingeavesdropper
0
0 comments X

The pith

Deep learning finds better power allocation than iterative search for minimizing intercept probability in cooperative NOMA energy harvesting systems.

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

The paper models a cooperative NOMA setup in which the near user relays the source message to the far user using decode-and-forward while an eavesdropper listens in. All nodes harvest energy and decode information via power splitting. An exact analytical expression is derived for the intercept probability at the eavesdropper. A deep neural network is then trained to choose the power allocation factor that reduces this probability. Simulations indicate the learned allocation is more robust and yields lower intercept probability than repeated iterative search.

Core claim

The paper establishes a closed-form intercept probability for the decode-and-forward cooperative NOMA SWIPT link in the presence of an eavesdropper and demonstrates that a deep learning optimizer produces a superior power allocation factor compared with conventional iterative search.

What carries the argument

The closed-form intercept probability expression obtained from the power-splitting receiver model and the eavesdropper channel, together with the deep neural network trained to minimize that probability over the power allocation factor.

If this is right

  • The intercept probability can be evaluated directly from channel gains and the chosen power split without simulation.
  • The trained network supplies a power allocation that lowers intercept probability faster than repeated numerical search.
  • Secrecy performance remains stable when system parameters vary within the training distribution.
  • The same optimization framework applies when the number of users or the eavesdropper location changes, provided the model is retrained.

Where Pith is reading between the lines

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

  • The derived probability expression could support real-time secrecy monitoring if channel estimates are available at the source.
  • Replacing power splitting with time switching would require a new intercept-probability derivation but might admit the same deep-learning optimizer.
  • The approach could be tested in multi-eavesdropper scenarios by extending the intercept definition to the strongest eavesdropper.
  • Hardware experiments with actual power-splitting circuits would check whether the analytical expression remains accurate under non-ideal rectifier behavior.

Load-bearing premise

The near user always relays via decode-and-forward and every device splits received power between energy harvesting and information decoding.

What would settle it

Monte Carlo trials in which the deep learning optimizer returns a higher intercept probability than the iterative search for the same channel statistics and power constraints would falsify the superiority result.

Figures

Figures reproduced from arXiv: 1907.05753 by Furqan Jameel, Ju Liu, Tapani Ristaniemi, Wali Ullah Khan, Zheng Chang.

Figure 1
Figure 1. Figure 1: System model. √ αF P sF to UN and UF , where si and αi are the data symbol and power allocation coefficient of i-th destination and P denotes the total transmit power. It is assumed that hSUN > hSUF , hence the power allocation factor should satisfy αN < αF , where αN +αF = 1. The nodes UN , UF and E are assumed to use the power-splitting receiver architecture for ID and EH. According to power-splitting ar… view at source ↗
Figure 2
Figure 2. Figure 2: A typical deep neural network with input, output and m [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Intercept probability versus transmit SNR. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Performance comparison (a) achievable secrecy rate [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Performance comparison (a) achievable secrecy rate [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
read the original abstract

Non-orthogonal multiple access (NOMA) is considered to be one of the best candidates for future networks due to its ability to serve multiple users using the same resource block. Although early studies have focused on transmission reliability and energy efficiency, recent works are considering cooperation among the nodes. The cooperative NOMA techniques allow the user with a better channel (near user) to act as a relay between the source and the user experiencing poor channel (far user). This paper considers the link security aspect of energy harvesting cooperative NOMA users. In particular, the near user applies the decode-and-forward (DF) protocol for relaying the message of the source node to the far user in the presence of an eavesdropper. Moreover, we consider that all the devices use power-splitting architecture for energy harvesting and information decoding. We derive the analytical expression of intercept probability. Next, we employ deep learning based optimization to find the optimal power allocation factor. The results show the robustness and superiority of deep learning optimization over conventional iterative search algorithm.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript analyzes secrecy in a cooperative NOMA SWIPT system in which the near user relays the source message to the far user via decode-and-forward in the presence of an eavesdropper, with all nodes using power-splitting for energy harvesting and information decoding. It derives a closed-form expression for intercept probability and applies deep learning to optimize the power allocation factor, claiming that the DL approach is more robust and outperforms conventional iterative search.

Significance. A correct closed-form intercept probability would supply useful analytical insight into secrecy performance under cooperative NOMA with SWIPT. If the DL optimizer were shown to generalize beyond the training ensemble, the combination of analysis and learning-based optimization could be practically relevant for secure resource allocation. No machine-checked proofs or parameter-free derivations are present.

major comments (1)
  1. [Numerical Results] Numerical Results section: the superiority and robustness claims for deep learning optimization rest on comparisons performed exclusively on the Rayleigh-fading, fixed-SNR ensemble used to generate training data; no out-of-distribution tests (different fading parameters, SNR ranges, or path-loss exponents) are reported, leaving open the possibility that reported gains are artifacts of in-distribution fitting rather than a genuine algorithmic advantage.
minor comments (1)
  1. [Abstract] The abstract states that all devices use power-splitting but does not indicate the assumed fading model (e.g., Rayleigh) used for the intercept-probability derivation; this should be stated explicitly in Section II.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We provide a point-by-point response to the major comment below.

read point-by-point responses

  1. Referee: [Numerical Results] Numerical Results section: the superiority and robustness claims for deep learning optimization rest on comparisons performed exclusively on the Rayleigh-fading, fixed-SNR ensemble used to generate training data; no out-of-distribution tests (different fading parameters, SNR ranges, or path-loss exponents) are reported, leaving open the possibility that reported gains are artifacts of in-distribution fitting rather than a genuine algorithmic advantage.

    Authors: The deep learning model is trained and tested on data drawn from the Rayleigh fading distribution and SNR ranges that exactly match the system model and closed-form intercept probability analysis presented in the manuscript. The iterative search baseline is evaluated under identical conditions, providing a direct and fair comparison within the considered scenario. The reported superiority and robustness therefore apply specifically to this standard ensemble for secrecy analysis of cooperative NOMA SWIPT systems. While additional out-of-distribution tests would be valuable for assessing broader generalization, they fall outside the scope of the current work and are not required to support the claims made for the analyzed setting. revision: no

Circularity Check

0 steps flagged

No significant circularity in derivation or optimization chain

full rationale

The paper derives an analytical closed-form expression for intercept probability directly from the system model (cooperative NOMA with DF relaying, power-splitting SWIPT, and eavesdropper) using standard information-theoretic steps on Rayleigh fading channels. This derivation stands independently of the subsequent numerical optimization. Deep learning is then applied as an external numerical solver to optimize the power allocation factor, benchmarked against conventional iterative search; no equations reduce the DL output to the input parameters by construction, no self-citation load-bearing premises are invoked for uniqueness or ansatz, and no fitted inputs are relabeled as predictions. The analysis is self-contained against external benchmarks and contains no enumerated circular patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review yields no identifiable free parameters, axioms, or invented entities beyond the standard wireless channel and protocol assumptions stated in the system description.

pith-pipeline@v0.9.0 · 5724 in / 1104 out tokens · 24389 ms · 2026-05-24T22:16:17.926168+00:00 · methodology

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

18 extracted references · 18 canonical work pages

  1. [1]

    Efficien t power allocation in downlink multi-cell multi-user noma ne tworks,

    W. U. Khan, Z. Y u, S. Y u, G. A. S. Sidhu, and J. Liu, “Efficien t power allocation in downlink multi-cell multi-user noma ne tworks,” IET Communications, vol. 13, no. 4, pp. 396–402, 2019

    work page 2019

  2. [2]

    Outage Analysis of Relay-Aided Non-Orthogonal Multiple Access wi th Partial Relay Selection,

    F. Jameel, S. Wyne, S. J. Nawaz, Z. Chang, and T. Ristaniem i, “Outage Analysis of Relay-Aided Non-Orthogonal Multiple Access wi th Partial Relay Selection,” in 2018 IEEE Globecom W orkshops (GC Wkshps) , Dec 2018, pp. 1–6

    work page 2018

  3. [3]

    Cooperative Non-Orthog onal Multiple Access in 5G Systems,

    Z. Ding, M. Peng, and H. V . Poor, “Cooperative Non-Orthog onal Multiple Access in 5G Systems,” IEEE Communications Letters, vol. 19, no. 8, pp. 1462–1465, Aug 2015

    work page 2015

  4. [4]

    Joint downlink/uplink design for wireles s powered networks with interference,

    P . D. Diamantoulakis, K. N. Pappi, G. K. Karagiannidis, H . Xing, and A. Nallanathan, “Joint downlink/uplink design for wireles s powered networks with interference,” IEEE Access, vol. 5, pp. 1534–1547, 2017

    work page 2017

  5. [5]

    Relay selection for coope rative NOMA,

    Z. Ding, H. Dai, and H. V . Poor, “Relay selection for coope rative NOMA,” IEEE Wireless Communications Letters, vol. 5, no. 4, pp. 416– 419, 2016

    work page 2016

  6. [6]

    Secure Communications i n Three- Step Two-Way Energy Harvesting DF Relaying,

    F. Jameel, S. Wyne, and Z. Ding, “Secure Communications i n Three- Step Two-Way Energy Harvesting DF Relaying,” IEEE Communications Letters, vol. 22, no. 2, pp. 308–311, Feb 2018

    work page 2018

  7. [7]

    Cooperativ e Non- orthogonal Multiple Access With Simultaneous Wireless Inf ormation and Power Transfer,

    Y . Liu, Z. Ding, M. Elkashlan, and H. V . Poor, “Cooperativ e Non- orthogonal Multiple Access With Simultaneous Wireless Inf ormation and Power Transfer,” IEEE Journal on Selected Areas in Communica- tions, vol. 34, no. 4, pp. 938–953, 2016

    work page 2016

  8. [8]

    Improving t he performance of cell-edge users in MISO-NOMA systems using T AS and SWIPT-based cooperative transmissions,

    T. N. Do, D. B. da Costa, T. Q. Duong, and B. An, “Improving t he performance of cell-edge users in MISO-NOMA systems using T AS and SWIPT-based cooperative transmissions,” IEEE Transactions on Green Communications and Networking , vol. 2, no. 1, pp. 49–62, 2018

    work page 2018

  9. [9]

    Secrecy sum rate maximization in NOMA systems with wireless informa tion and power transfer,

    G. He, L. Li, X. Li, W. Chen, L. L. Y ang, and Z. Han, “Secrecy sum rate maximization in NOMA systems with wireless informa tion and power transfer,” in 2017 9th International Conference on Wireless Communications and Signal Processing (WCSP) , Oct 2017, pp. 1–6

    work page 2017

  10. [10]

    Artificial Noise Aided Secure Cognitive Beamforming for Cooperative MISO-N OMA Using SWIPT,

    F. Zhou, Z. Chu, H. Sun, R. Q. Hu, and L. Hanzo, “Artificial Noise Aided Secure Cognitive Beamforming for Cooperative MISO-N OMA Using SWIPT,” IEEE Journal on Selected Areas in Communications , pp. 1–1, 2018

    work page 2018

  11. [11]

    Deep learning,

    Y . LeCun, Y . Bengio, and G. Hinton, “Deep learning,” nature, vol. 521, no. 7553, p. 436, 2015

    work page 2015

  12. [12]

    Rectified linear units improve restricted boltz- mann machines,

    V . Nair and G. E. Hinton, “Rectified linear units improve restricted boltz- mann machines,” in Proceedings of the 27th international conference on machine learning (ICML-10) , 2010, pp. 807–814

    work page 2010

  13. [13]

    A fast learnin g algorithm for deep belief nets,

    G. E. Hinton, S. Osindero, and Y .-W. Teh, “A fast learnin g algorithm for deep belief nets,” Neural computation , vol. 18, no. 7, pp. 1527–1554, 2006

    work page 2006

  14. [14]

    On d eep learning- based channel decoding,

    T. Gruber, S. Cammerer, J. Hoydis, and S. ten Brink, “On d eep learning- based channel decoding,” in Information Sciences and Systems (CISS), 2017 51st Annual Conference on . IEEE, 2017, pp. 1–6

    work page 2017

  15. [15]

    Deep learning for wireless physical layer: Opportunities and ch allenges,

    T. Wang, C.-K. Wen, H. Wang, F. Gao, T. Jiang, and S. Jin, “ Deep learning for wireless physical layer: Opportunities and ch allenges,” China Communications , vol. 14, no. 11, pp. 92–111, 2017

    work page 2017

  16. [16]

    Deep Learning- Aided SCMA,

    M. Kim, N.-I. Kim, W. Lee, and D.-H. Cho, “Deep Learning- Aided SCMA,” IEEE Communications Letters , vol. 22, no. 4, pp. 720–723, 2018

    work page 2018

  17. [17]

    Power of deep learning f or channel estimation and signal detection in OFDM systems,

    H. Y e, G. Y . Li, and B.-H. Juang, “Power of deep learning f or channel estimation and signal detection in OFDM systems,” IEEE Wireless Communications Letters , vol. 7, no. 1, pp. 114–117, 2018

    work page 2018

  18. [18]

    Deep learning for an effective nonorthogonal multiple access scheme,

    G. Gui, H. Huang, Y . Song, and H. Sari, “Deep learning for an effective nonorthogonal multiple access scheme,” IEEE Transactions on V ehicular Technology, vol. 67, no. 9, pp. 8440–8450, 2018

    work page 2018