Pith. sign in

REVIEW 4 major objections 4 minor 41 references

The paper derives single-letter converse and achievability bounds for securely sending a two-part semantic source over a degraded wiretap channel, with separate fidelity and secrecy constraints on each component.

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 ↗

Single-letter converse and achievability bounds are derived for the rate-distortion-equivocation region of secure lossy joint source-channel coding of a two-component semantic source over a degraded wiretap channel, with separate fidelity and secrecy constraints per component.

T0 review reviewed 2026-08-04 challenge →

load-bearing objection New model and a sound converse, but the achievability proof as printed has an r-factor inconsistency and an unsupported semantic-leakage bound, so Theorem 5's inner region is not established. the 4 major comments →

arxiv 2509.12142 v3 pith:OHKVQVB7 submitted 2025-09-15 cs.IT math.IT

Secure Semantic Communication over Wiretap Channels: Rate-Distortion-Equivocation Tradeoff

Denis Kozlov , Mahtab Mirmohseni , Rahim Tafazolli This is my paper
classification cs.IT math.IT MSC 94A1594A2494A17
keywords semantic communicationwiretap channelrate-distortion-equivocation regionjoint source-channel codinginformation-theoretic securityequivocationsuperposition codingdegraded wiretap channel
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 reading

This paper aims to fix the fundamental limits of secure semantic communication. The model: a transmitter sends a two-part source—semantic data S and observed data U—over a degraded wiretap channel; the legitimate receiver must meet separate distortion targets for each part, the eavesdropper must face separate secrecy targets (equivocation, i.e., the eavesdropper's residual uncertainty) for each part, and an optional shared secret key of rate R_k may be used. The paper claims a converse region (Theorem 1), stated in single-letter quantities—one-symbol mutual informations and rate-distortion functions—in which each equivocation bound is the sum of three resources: the key rate, r times the wiretap channel's secrecy capacity, and the entropy lost by compressing that component (e.g., H(S) − R(D_s)), together with the joint bound R(D_s,D_u) ≤ r·I(X;Y). It claims an inner region (Theorem 5) via a four-layer stochastic superposition code with two private layers, valid when the key rate is zero. If the claims hold, these are the first such bounds for this model; they reduce to earlier results for secure source coding and secure joint source-channel coding as special cases, and the numerical evaluations quantify how much secrecy can be bought by protecting only the semantic part.

Core claim

The central claim is additive structure. Theorem 1 states that any achievable tuple must satisfy R(D_s,D_u) ≤ r·I(X;Y) and three equivocation bounds of the form Δ ≤ R_k + r·max_p I(W;Y|Z) − R(·) + H(·): key rate plus r times wiretap secrecy capacity plus compression entropy loss, applied separately to the semantic, observation, and joint components. Theorem 5 claims that for R_k = 0 a four-layer stochastic superposition code—public layers (A_0,Q_1) and (B_0,Q_2) carrying the common lossy description, private layers (A_1,W_1) and (B_1,W_2) carrying component-specific refinements with wiretap-random binning—attains the inner region given by inequalities (41)–(48). Specialized to Gaussian sourc

What carries the argument

Three tools carry the argument. (1) Semantic-source rate-distortion functions: the converse is written directly in terms of R_u(D_u), R_s(D_s), and the joint R(D_s,D_u); Case 1 uses the indirect RDF of Lemma 3 with its modified distortion metric. (2) Wiretap single-letterization [34, Lemma 17.12]: it turns multi-letter differences such as I(S^k;Z^n) − I(Ŝ^k;Y^n) into n·[I(W_1;Z|Q) − I(W_1;Y|Q)], producing the secrecy-capacity term max_p I(W;Y|Z). (3) The four-layer superposition code: public layers (A_0,Q_1) and (B_0,Q_2) carry the common lossy description; private layers (A_1,W_1) and (B_1,W_2) carry per-component refinement; random indices (M'_1,M'_2) supply the wiretap randomization; an i

Load-bearing premise

The achievability region stands on one asserted bound in the proof (Section IV-A7, item 5)—that the eavesdropper's observation, given the public channel indices, leaks no more about the semantic sequence than a single public index of rate R'_20—plus an imported identity for the Case-1 joint rate-distortion function; if either fails, the stated inner region is unsupported.

What would settle it

Take any concrete instance of the scheme, e.g., the Gaussian model of Section V-A, compute the right-hand side of inequality (46), and measure the eavesdropper's true equivocation (1/k)H(S^k | Z^n): a strict exceedance falsifies the achievability claim, and the specific assertion to probe is I(Z^n; S^k | I', J') ≤ H(L') ≤ k·R'_20. A cheaper check: numerically sample the inner region (41)–(48) and verify it never violates the converse inequalities (12)–(15); any crossing is an immediate contradiction.

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

If this is right

  • Each equivocation bound is additive: one bit of shared key, r times one bit of wiretap secrecy capacity, or one bit of compression loss each buy the same one bit of equivocation for that component, so the three secrecy resources can be traded against each other.
  • Protecting only the semantic part is quantifiably cheaper than protecting everything: the Gaussian and binary evaluations show the semantic-only-secrecy region strictly contains the full-secrecy region.
  • Letting the encoder see the semantic samples (Case 2) strictly enlarges the region over observation-only encoding (Case 1); in Case 1 a minimum semantic distortion exists below which no tuple is achievable.
  • The converse is directly evaluable because it is expressed through rate-distortion functions and secrecy capacity terms: closed forms hold for Gaussian sources over Gaussian wiretap channels (Corollary 1) and Bernoulli sources over binary-symmetric channels (Corollary 2), where equivocation is seen to saturate at finite distortion.
  • The general region specializes to known problems: discarding the channel recovers the secure semantic source-coding regions of [27] and [13], and collapsing to one source component recovers the secure JSCC region of [19].

Where Pith is reading between the lines

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

  • A likely next step the paper leaves open is full separation: if the achievability scheme can be extended to positive key rate by appending a one-time pad to the source code, the inner region may meet the converse at all R_k, making the additive decomposition of Theorem 1 exact.
  • The saturation of equivocation at finite distortion in the binary numerics suggests a design rule: compress the semantic part only until its secrecy saturates, then spend remaining rate on the observed part or on key—extra semantic compression is wasted.
  • The Case-1 converse imports the identity R_i(D_s,D_u) = max{R_u(D_u), R_s,i(D_s)} from an external result; testing that identity beyond Gaussian/Bernoulli sources, such as heavy-tailed or mixture sources, would delimit where the converse region holds.
  • The crux inequality I(Z^n; S^k | I', J') ≤ H(L') is testable in simulation: search over strongly correlated (S,U) distributions and code instances; a violation would force a stronger inner bound, while slack would suggest the inner bound can be tightened toward the outer.
Share X Bluesky LinkedIn Reddit HN

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

4 major / 4 minor

Summary. The paper studies lossy joint source-channel coding of a two-component 'semantic' source (S,U) over a degraded wiretap channel, with separate distortion constraints at the legitimate receiver and separate equivocation constraints at the eavesdropper. Two encoder cases are considered: Case 1 observes only U, and Case 2 observes both S and U. Theorem 1 gives a single-letter converse region expressed in terms of rate-distortion functions and wiretap secrecy capacities. Theorem 5 proposes a four-layer superposition coding scheme and claims an inner region for the case R_k = 0. The general results are specialized to Gaussian and Bernoulli sources/channels, and numerical comparisons of Case 1 and Case 2 are provided. Reductions to prior work of Yamamoto [19], Guo et al. [13], and Yamamoto [27] are also discussed.

Significance. If the results are correct, the converse is a useful compact single-letter characterization that generalizes several existing secure source/channel coding results, and the explicit presence of rate-distortion functions makes it numerically tractable. The proposed achievability scheme is an interesting attempt to control the equivocation of two source components separately. However, as printed, the achievability proof has load-bearing gaps: the semantic-equivocation analysis relies on an unproved leakage bound, the observation-equivocation analysis contains an unjustified conditional-independence step, and at least one displayed condition is algebraically wrong (missing a factor r). The converse and the comparison sections appear sound, but the claimed single-letter characterization is not yet established. With a substantial revision of the achievability proof the paper could be a significant contribution; in its present form it requires major reworking.

major comments (4)
  1. [Section IV-A7, eqs. (71)-(75)] The bound I(Z^n;S^k|I',J') ≤ H(L') ≤ kR'_20 is asserted in item 5 with a one-sentence justification. This is not a consequence of the code construction: conditioned on (I',J'), the transmitted codewords include the layers indexed by L' and P', and P' is selected from U^k (and in Case 2 from S^k), so P' can carry information about S^k beyond L'. No Markov chain S^k-(I',J',L')-Z^n is proved, and no wiretap-secrecy or resolvability bound on the W2 layer is invoked. Since (73) feeds directly into (75) and hence (46), the semantic-equivocation claim of Theorem 5 is unsupported without a further argument.
  2. [Section IV-A7, observation item 5(e), eqs. (78)-(80)] The step I(Z^n;U^k|I',L',P',M2') = 0 is justified by saying that the eavesdropper can reconstruct a lossy version of U^k. Knowing the indices and codebooks determines the reconstruction \tilde U, but not the actual source sequence U^k, which remains random. The source is only jointly typical with the chosen codewords, so the conditional mutual information need not vanish. This invalidates the observation-equivocation lower bound (78)-(80) and therefore inequality (47) of Theorem 5.
  3. [Theorem 5, eq. (43); Section IV-A8, eq. (88)] The factor r is missing. Combining the source covering condition (53), R10 > I(A0;V), the mapping condition (59), R10 ≤ R'_10, and the channel decoding condition (62), R'_10 < (r+ε)I(Q1;Y), gives I(A0;V) < rI(Q1;Y), not I(A0;V) < I(Q1;Y). The statement and the FM-elimination summary repeat the same omission. As written, (43) is dimensionally inconsistent and weaker than what the proof actually establishes; this must be corrected.
  4. [Theorem 5, eq. (46) vs. Section IV-A8, eq. (91)] The final simplification introduces [H(S)-I(A;V)]^+ and [I(B;V|A0)-rI(W2;Y|Q)]^+ without deriving them from (91), where the first term has no positive part. If H(S)-I(A;V) is negative, replacing it by zero can increase the right-hand side and thus overstate the achievable equivocation threshold. The paper should either justify the positive-part operations or state the bound with a single max(0,·) applied to the entire lower bound derived in the proof.
minor comments (4)
  1. [Section IV-A3] The rate notation is garbled: 'R'_10, R'_11, R'_10, and R'_11' should presumably be R'_10, R'_11, R'_20, R'_21. Also, the codebook variables q2 and w1 conflict with the Q2/W1 notation of Theorem 5 and Fig. 3; the naming should be harmonized.
  2. [Lemmas 5 and 7] The Case 1 joint rate-distortion formulas R_i(Ds,Du)=max{R_u(Du),R_{s,i}(Ds)} are quoted from [38, Proposition 3] and [9] without stating the conditions under which the max formula holds. Since these formulas are used for the numerical converse bounds, the relevant hypotheses should be reproduced or a proof sketched.
  3. [Section III-A1] In the reduction to [27], the text says that in Case 1 'the term H(S|\hat U) can be substituted with H(U|\hat U)'. This seems to mix up the secrecy constraint of [27]; please re-check and clarify.
  4. [Abstract] Typo: 'underlaying' should be 'underlying'.

Circularity Check

0 steps flagged

No circular derivation: the central converse and achievability bounds are derived from standard single-letterization and coding arguments; the only self-citation is the authors' earlier conference paper and is not load-bearing.

full rationale

The derivation chain is self-contained in the required sense. The converse (Theorem 1) is obtained by single-letterizing standard multi-letter information inequalities (Eqs. (23)-(28)) and by applying the Csiszár-Körner Lemma 17.12 and support lemma; the equivocation bounds (40), etc., follow from entropy manipulations, not from the statement being proved. The achievability (Theorem 5) is constructive: source and channel codebooks are generated from auxiliary variables, covering/packing lemmas give the rate conditions (49)-(65), and the final region (86)-(93) is obtained by Fourier-Motzkin elimination. No parameter is fitted to data and then called a prediction. The reductions in Section III-A to [19], [13], [27] are consistency checks against independent external results; they are consequences, not premises. The Gaussian and binary corollaries import RDF closed forms from external papers [30], [38]-[41] by different author groups, which is independent support. The only self-citation is [1] (the authors' earlier conference version), used only to say this article extends it; no theorem in this paper is justified by [1], so the self-citation is not load-bearing. For completeness, the Section IV-A7 item-5 bound I(Z^n;S^k|I',J') <= H(L') is asserted with a one-sentence justification and is a genuine proof gap for inequality (46); however, that is a missing-support/correctness issue rather than a circular reduction, because it is not an equation that is equal to the target by construction. Hence the circularity score is low (2), reflecting the minor non-load-bearing self-citation, with no circular steps in the derivation.

Axiom & Free-Parameter Ledger

0 free parameters · 6 axioms · 0 invented entities

No data fitting is involved: the β_1, β_2, γ_1, γ_2 parameters in the Gaussian and binary converse bounds are slack variables that can be set to 1 (or 0) for a valid bound, not fitted quantities; the Monte Carlo covariance matrices are random draws for evaluation, not optimization against data. The derivation relies on standard information-theoretic machinery (typical sequences, covering and packing lemmas, the Csiszar-Korner single-letterization Lemma 17.12) and on external RDF results. No invented entities (particles, mediators, forces, dimensions) are introduced; the auxiliary random variables A, B, Q, W are proof devices within the coding scheme.

axioms (6)
  • domain assumption Degraded wiretap channel: p_{Y,Z|X} = p_{Y|X}·p_{Z|Y}
    Invoked in Theorem 1 and throughout; the secrecy-capacity form I(W;Y|Z) = I(W;Y) - I(W;Z) and the Gaussian and binary specializations rely on degradedness; the results do not claim to cover general wiretap channels.
  • domain assumption Memorylessness: i.i.d. source blocks (S^k,U^k) and DMC channel p_{Y,Z|X}
    Section II-A; enables the single-letterization steps (23)-(26) via [35, Theorem 6.1] and the typicality-based proofs.
  • standard math Csiszar-Korner Lemma 17.12 single-letterization
    Used at (37)-(38) to bound I(S^k;Z^n) - I(S^k;Y^n) through auxiliary r.v.s W_1, W_2, W; the cardinality bound |W| ≤ |X|+1 is referenced to [34, Thm 17.13].
  • standard math Covering and packing lemmas and the support lemma (El Gamal-Kim [30])
    Used in the achievability proof: rate conditions (53)-(54) (covering), (62)-(65) (packing), and the cardinality arguments.
  • standard math Gaussian wiretap secrecy-capacity bound with power split, Bloch-Barros [37, Thm 5.1]
    Corollary 1, inequality (111); bounds I(W;Y) - I(W;Z) by the β-split Gaussian capacity; imported without derivation, its applicability to the semantic code's W_1/W_2 power split is assumed.
  • standard math Closed-form RDFs: [38, Prop. 3] (Case 1 Gaussian joint RDF), [39, Thm 6] (Case 2 Gaussian joint RDF), [41, Thm 2] (Case 2 binary RDF)
    Lemmas 4, 5, 6, 7; load-bearing for the numerical evaluations; the Case 1 formula max{R_u(D_u), R_s,i(D_s)} in particular is stated without derivation in this paper.

reviewed 2026-08-04 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Secure Semantic Communication over Wiretap Channels: Rate-Distortion-Equivocation Tradeoff." pith.science (2026). https://pith.science/paper/OHKVQVB7

@misc{pith2026250912142,
  author       = {Pith},
  title        = {Pith review of: Secure Semantic Communication over Wiretap Channels: Rate-Distortion-Equivocation Tradeoff},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OHKVQVB7}},
  note         = {Machine review of arXiv:2509.12142}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

This paper investigates an information-theoretic model of secure semantic-aware communication. For this purpose, we consider the lossy joint source-channel coding (JSCC) of a memoryless semantic source transmitted over a memoryless wiretap channel. The source consists of two correlated parts that represent semantic and observed aspects of the information. Our model assumes separate fidelity and secrecy constraints on each source component and, in addition, encompasses two cases for the source output, in order to evaluate the performance gains if the encoder has an extended access to the source. Specifically, in Case 1, the encoder has direct access only to the samples from a single (observed) source component, while in Case 2 it has additional direct access to the samples of the underlying semantic information. We derive single-letter converse and achievability bounds on the rate-distortion-equivocation region. The converse bound explicitly contains rate-distortion functions, making it easy to evaluate, especially for some common distributions. The proposed achievability coding scheme involves novel stochastic superposition coding with two private parts to enable analysis of the equivocation for each source component, separately. Our results generalise some of the previously established source and source-channel coding problems. The general results are further specialised to Gaussian and Bernoulli sources transmitted over Gaussian and binary wiretap channels, respectively. The numerical evaluations illustrate the derived bounds for these distributions.

Figures

Figures reproduced from arXiv: 2509.12142 by Denis Kozlov, Mahtab Mirmohseni, Rahim Tafazolli.

Figure 1
Figure 1. Figure 1: Wiretap channel model for semantic communication with two encoder configurations: Case 1 (observation-only input) [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Codebook design. Indices m′ 1 and m′ 2 are picked at random from a uniform distribution. a k 0 (i) b k 0 (i, l) b k 1 (i, l, p) a k 1 (i, j) (a) Source Encoding Stage q n 1 (i ′ ) w n 1 (i ′ , l′ ) w n 2 (i ′ , l′ , p′ , m′ 2 ) q n 2 (i ′ , j′ , m′ 1 ) (b) Channel Encoding Stage [PITH_FULL_IMAGE:figures/full_fig_p017_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Superposition coding layouts rate conditions must be satisfied, which follows from the property for the size of the typical set. R10 < H(A0) − δ, R11 < H(A1|A0) − δ (49) R20 < H(B0|A0) − δ, R21 < H(B1|A0, B0) − δ (50) where δ → 0 with k → ∞. 2) Source Encoding Process: The encoding process depends on the encoder case as described in Section II. Let v k .= (s k , uk ) and V .= (S, U) in Case 2 (encoder has … view at source ↗
Figure 4
Figure 4. Figure 4: Relation between source and channel coding. [PITH_FULL_IMAGE:figures/full_fig_p019_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Converse bounds the Gaussian and binary system models. Grey edges: Case 1 encoder. Black edges: Case 2 encoder. [PITH_FULL_IMAGE:figures/full_fig_p028_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Inner and outer bounds for the Gaussian system model. Dotted curves correspond to the no secrecy case, i.e, [PITH_FULL_IMAGE:figures/full_fig_p029_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Outer bounds for binary system model. Two scenarios are considered, without encryption, [PITH_FULL_IMAGE:figures/full_fig_p032_7.png] view at source ↗

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

41 extracted references · 3 linked inside Pith

  1. [1]

    Secure semantic communication over wiretap channel,

    D. Kozlov, M. Mirmohseni, and R. Tafazolli, “Secure semantic communication over wiretap channel,” in2024 IEEE Information Theory Workshop (ITW), 2024, pp. 37–42

    2024

  2. [2]

    Aazhang, P

    B. Aazhang, P. Ahokangas, H. Alves, M.-S. Alouini, J. Beek, H. Benn, M. Bennis, J. Belfiore, E. Strinati, F. Chen, K. Chang, F. Clazzer, S. Dizit, D. Kwon, M. Giordiani, W. Haselmayr, J. Haapola, E. Hardouin, E. Harjula, and P. Zhu, Key drivers and research challenges for 6G ubiquitous wireless intelligence (white paper). University of Oulu, 09 2019

    2019

  3. [3]

    Semantic communications for future internet: Fundamentals, applications, and challenges,

    W. Yang, H. Du, Z. Q. Liew, W. Y . B. Lim, Z. Xiong, D. Niyato, X. Chi, X. Shen, and C. Miao, “Semantic communications for future internet: Fundamentals, applications, and challenges,”IEEE Communications Surveys & Tutorials, vol. 25, no. 1, pp. 213–250, 2023

    2023

  4. [5]

    Beyond transmitting bits: Context, semantics, and task-oriented communications,

    D. G ¨und¨uz, Z. Qin, I. E. Aguerri, H. S. Dhillon, Z. Yang, A. Yener, K. K. Wong, and C.-B. Chae, “Beyond transmitting bits: Context, semantics, and task-oriented communications,”IEEE Journal on Selected Areas in Communications, vol. 41, no. 1, pp. 5–41, 2023

    2023

  5. [6]

    Joint source–channel coding: Fundamentals and recent progress in practical designs,

    D. G ¨und¨uz, M. A. Wigger, T.-Y . Tung, P. Zhang, and Y . Xiao, “Joint source–channel coding: Fundamentals and recent progress in practical designs,”Proceedings of the IEEE, pp. 1–32, 2024

    2024

  6. [7]

    Deep joint source-channel coding for wireless image transmission,

    E. Bourtsoulatze, D. B. Kurka, and D. G ¨und¨uz, “Deep joint source-channel coding for wireless image transmission,” in ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2019, pp. 4774–4778. September 25, 2025 DRAFT 34

    2019

  7. [8]

    Deepwive: Deep-learning-aided wireless video transmission,

    T.-Y . Tung and D. G ¨und¨uz, “Deepwive: Deep-learning-aided wireless video transmission,”IEEE Journal on Selected Areas in Communications, vol. 40, no. 9, pp. 2570–2583, 2022

    2022

  8. [9]

    The role of fidelity in goal-oriented semantic communication: A rate distortion approach,

    P. A. Stavrou and M. Kountouris, “The role of fidelity in goal-oriented semantic communication: A rate distortion approach,” IEEE Transactions on Communications, vol. 71, no. 7, pp. 3918–3931, 2023

    2023

  9. [10]

    Goal-oriented single-letter codes for lossy joint source-channel coding,

    ——, “Goal-oriented single-letter codes for lossy joint source-channel coding,” in2023 IEEE Information Theory Workshop (ITW), 2023, pp. 64–69

    2023

  10. [12]

    An indirect rate-distortion characterization for semantic sources: General model and the case of Gaussian observation,

    J. Liu, S. Shao, W. Zhang, and H. V . Poor, “An indirect rate-distortion characterization for semantic sources: General model and the case of Gaussian observation,”IEEE Transactions on Communications, vol. 70, no. 9, pp. 5946–5959, 2022

    2022

  11. [13]

    Protecting semantic information using an efficient secret key,

    T. Guo, J. Han, H. Wu, Y . Wang, B. Bai, and W. Han, “Protecting semantic information using an efficient secret key,” in 2022 IEEE International Symposium on Information Theory (ISIT), 2022, pp. 2660–2665

    2022

  12. [14]

    Semantic compression with side information: A rate-distortion perspective,

    T. Guo, Y . Wang, J. Han, H. Wu, B. Bai, and W. Han, “Semantic compression with side information: A rate-distortion perspective,” 2022. [Online]. Available: https://arxiv.org/abs/2208.06094

    Pith/arXiv arXiv 2022

  13. [15]

    Rate-distortion theory for strategic semantic communication,

    Y . Xiao, X. Zhang, Y . Li, G. Shi, and T. Bas ¸ar, “Rate-distortion theory for strategic semantic communication,” in2022 IEEE Information Theory Workshop (ITW), 2022, pp. 279–284

    2022

  14. [16]

    Rate-distortion-perception theory for semantic communication,

    J. Chai, Y . Xiao, G. Shi, and W. Saad, “Rate-distortion-perception theory for semantic communication,” in2023 IEEE 31st International Conference on Network Protocols (ICNP), 2023, pp. 1–6

    2023

  15. [17]

    The wire-tap channel,

    A. D. Wyner, “The wire-tap channel,”The Bell System Technical Journal, vol. 54, no. 8, pp. 1355–1387, 1975

    1975

  16. [18]

    Broadcast channels with confidential messages,

    I. Csiszar and J. Korner, “Broadcast channels with confidential messages,”IEEE Transactions on Information Theory, vol. 24, no. 3, pp. 339–348, 1978

    1978

  17. [19]

    Rate-distortion theory for the shannon cipher system,

    H. Yamamoto, “Rate-distortion theory for the shannon cipher system,”IEEE Transactions on Information Theory, vol. 43, no. 3, pp. 827–835, 1997

    1997

  18. [20]

    Shannon’s secrecy system with informed receivers and its application to systematic coding for wiretapped channels,

    N. Merhav, “Shannon’s secrecy system with informed receivers and its application to systematic coding for wiretapped channels,” in2007 IEEE International Symposium on Information Theory, 2007, pp. 2481–2485

    2007

  19. [21]

    Secure transmission of sources over noisy channels with side information at the receivers,

    J. Villard, P. Piantanida, and S. Shamai, “Secure transmission of sources over noisy channels with side information at the receivers,”IEEE Transactions on Information Theory, vol. 60, no. 1, pp. 713–739, 2014

    2014

  20. [22]

    Task-oriented source coding: Secure transmission, detection,

    J. Villard, “Task-oriented source coding: Secure transmission, detection,” Ph.D. dissertation, SUPELEC, 2011

    2011

  21. [23]

    The rate-distortion function for source coding with side information at the decoder,

    A. Wyner and J. Ziv, “The rate-distortion function for source coding with side information at the decoder,”IEEE Transactions on Information Theory, vol. 22, no. 1, pp. 1–10, 1976

    1976

  22. [24]

    To code or not to code,

    M. Gastpar, B. Rimoldi, and M. Vetterli, “To code or not to code,” in2000 IEEE International Symposium on Information Theory (Cat. No.00CH37060), 2000, pp. 236–

    2000

  23. [25]

    Fundamental limitation of semantic communications: Neural estimation for rate-distortion,

    D. Li, J. Huang, C. Huang, X. Qin, H. Zhang, and P. Zhang, “Fundamental limitation of semantic communications: Neural estimation for rate-distortion,”Journal of Communications and Information Networks, vol. 8, no. 4, pp. 303–318, 2023

    2023

  24. [26]

    Coding theorems for shannon’s cipher system with correlated source outputs, and common information,

    H. Yamamoto, “Coding theorems for shannon’s cipher system with correlated source outputs, and common information,” IEEE Transactions on Information Theory, vol. 40, no. 1, pp. 85–95, 1994

    1994

  25. [27]

    A source coding problem for sources with additional outputs to keep secret from the receiver or wiretappers (corresp.),

    ——, “A source coding problem for sources with additional outputs to keep secret from the receiver or wiretappers (corresp.),”IEEE Transactions on Information Theory, vol. 29, no. 6, pp. 918–923, 1983

    1983

  26. [28]

    Secure lossy helper and gray-wyner problems,

    M. Benammar and A. Zaidi, “Secure lossy helper and gray-wyner problems,” in2016 IEEE International Symposium on Information Theory (ISIT), 2016, pp. 2454–2458. September 25, 2025 DRAFT 35

    2016

  27. [29]

    Secure lossy source coding for some classes of helper and gray-wyner models,

    ——, “Secure lossy source coding for some classes of helper and gray-wyner models,” 2017. [Online]. Available: https://arxiv.org/abs/1701.03621

    Pith/arXiv arXiv 2017

  28. [30]

    El Gamal and Y .-H

    A. El Gamal and Y .-H. Kim,Network Information Theory. Cambridge University Press, 2011

    2011

  29. [31]

    Achievable rates for multiple descriptions,

    A. Gamal and T. Cover, “Achievable rates for multiple descriptions,”IEEE Transactions on Information Theory, vol. 28, no. 6, pp. 851–857, 1982

    1982

  30. [32]

    Information transmission with additional noise,

    R. Dobrushin and B. Tsybakov, “Information transmission with additional noise,”IRE Transactions on Information Theory, vol. 8, no. 5, pp. 293–304, 1962

    1962

  31. [33]

    Indirect rate distortion problems,

    H. Witsenhausen, “Indirect rate distortion problems,”IEEE Transactions on Information Theory, vol. 26, no. 5, pp. 518–521, 1980

    1980

  32. [34]

    Csisz ´ar and J

    I. Csisz ´ar and J. K ¨orner,Information Theory: Coding Theorems for Discrete Memoryless Systems, 2nd ed. Cambridge University Press, 2011

    2011

  33. [35]

    Polyanskiy and Y

    Y . Polyanskiy and Y . Wu,Information Theory: From Coding to Learning. Cambridge University Press, 2025

    2025

  34. [36]

    Information theoretic security,

    Y . Liang, H. V . Poor, and S. S. (Shitz), “Information theoretic security,”Foundations and Trends® in Communications and Information Theory, vol. 5, no. 4–5, pp. 355–580, 2009. [Online]. Available: http://dx.doi.org/10.1561/0100000036

    doi:10.1561/0100000036 2009

  35. [37]

    Bloch and J

    M. Bloch and J. Barros,Physical-Layer Security: From Information Theory to Security Engineering. Cambridge University Press, 2011

    2011

  36. [38]

    A rate-distortion framework for characterizing semantic information,

    J. Liu, W. Zhang, and H. V . Poor, “A rate-distortion framework for characterizing semantic information,” in2021 IEEE International Symposium on Information Theory (ISIT), 2021, pp. 2894–2899

    2021

  37. [39]

    Compression of correlated Gaussian sources under individual distortion criteria,

    J. Xiao and Z.-Q. Luo, “Compression of correlated Gaussian sources under individual distortion criteria,” in43rd Annual Allerton Conference on Communication, Control and Computing 2005, 2005, pp. 438–447

    2005

  38. [40]

    Berger,Rate-Distortion Theory

    T. Berger,Rate-Distortion Theory. John Wiley and Sons, Ltd, 2003

    2003

  39. [41]

    Successive refinement of vector sources under individual distortion criteria,

    J. Nayak, E. Tuncel, D. Gunduz, and E. Erkip, “Successive refinement of vector sources under individual distortion criteria,” in2007 IEEE Information Theory Workshop, 2007, pp. 319–324

    2007

  40. [42]

    An automated theorem proving framework for information-theoretic results,

    C. T. Li, “An automated theorem proving framework for information-theoretic results,” in2021 IEEE International Symposium on Information Theory (ISIT), 2021, pp. 2750–2755. September 25, 2025 DRAFT

    2021

  41. [2022]

    Available: https://arxiv.org/abs/2201.01389

    [Online]. Available: https://arxiv.org/abs/2201.01389

    Pith/arXiv arXiv

This paper was first reviewed by deepseek-v4-flash on August 4, 2026.