Heterogeneity-Aware Belief Synchronization for Semantic Communication in AI-Native 6G Networks

Haejoon Jung; Kapal Dev; Muhammad Abubakar Rashid; Muhammad Hannan Akram; Syed Ali Hassan; Wassi Haider Kabir

REVIEW 4 major objections 5 minor 14 references

Heterogeneity-Aware Belief Synchronization for Semantic Communication in AI-Native 6G Networks

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read MEC-hosted latent translators align heterogeneous agents' beliefs without joint training.

desk verdict The framework idea has merit, but the case study validates an oracle linear pseudoinverse, not the latent translation model central to the abstract's claim. read the letter →

arxiv 2608.13394 v1 pith:FHMCTNKQ submitted 2026-08-13 eess.SP cs.AIcs.MA

Muhammad Hannan Akram , Muhammad Abubakar Rashid , Wassi Haider Kabir , Haejoon Jung , Kapal Dev , Syed Ali Hassan This is my paper · ORCID

classification eess.SPcs.AIcs.MA

keywords semanticcommunicationbeliefsynchronizationheterogeneousAIagentslatenttranslationmodelmulti-accessedgecomputing6GnetworksAI-nativeknowledgedrift

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

6G networks will connect thousands of AI agents with very different model architectures and local observations, and semantic communication between them only works if their beliefs stay aligned. This paper claims that a latent translation model hosted on multi-access edge computing (MEC) servers can convert one agent's compact belief update into agent-specific updates for other agents, so heterogeneous agents can stay synchronized without joint training or identical architectures. The framework sends updates only when a change is detected and only to relevant agents, which keeps communication cost low and keeps raw data and full model weights off the channel. In a simplified linear case study, the proposed scheme keeps belief error at 0.123 versus 0.385 for homogeneous synchronization at equal traffic of 1.08e3 transmitted parameters, and its error stays relatively stable as heterogeneity grows.

What carries the argument

The central object is the latent translation model, a component hosted on MEC servers that maps a belief update $\Delta B$ from one agent's latent space into the latent space of each receiving agent. In the case study it is not a trained network but an algebraic operator: because the server knows the world vector $w$, the translation from agent $x$ to agent $y$ is computed as $Y X^{-1} x$, where $X$ is the sending agent's matrix, $X^{-1}$ is its Moore-Penrose inverse, and $Y$ is the receiving agent's matrix. This operator does the work of converting one agent's private representation change into a form that another agent can add to its own belief state. The framework wraps that operator in two rules: a change-detection rule that triggers propagation only when an agent's belief actually changes, and a relevance filter that sends translated updates only to connected agents for whom the information matters.

What would settle it

Train two neural-network agents with different architectures on overlapping but non-identical data, let each observe a changing world, and run the pseudoinverse-based translation with the server estimating the world vector from one agent's readout; if translated updates do not keep the readout error below the homogeneous baseline over time, the linear known-world assumption is load-bearing.

Watch

Extended reading notes

Core claim

The paper's central claim is that the obstacle to semantic communication among heterogeneous agents is not architecture difference itself but the absence of a shared mapping between latent spaces; once that mapping exists, belief updates can be exchanged directly. In the case study, each agent's latent vector is a linear function of a fixed world vector $w$, and the translation operator is computed as the Moore-Penrose inverse of the sending agent's matrix, so the update sent from agent $x$ to agent $y$ is $Y X^{-1} x$. The measured result is that this scheme reaches belief error 0.123, against 0.385 for homogeneous synchronization and 0.000 for full model transfer, while using the same number of transmitted parameters as the homogeneous baseline (1.08e3); that is a 68.1% error reduction at equal cost. The paper also reports that error rises with heterogeneity for both no-synchronization and homogeneous synchronization, while the proposed scheme stays relatively flat.

Load-bearing premise

The paper's case study assumes each agent's belief is a linear function of one known ground-truth vector $w$ and that the MEC server knows $w$ exactly, so translation reduces to a pseudoinverse; if real neural latent spaces are nonlinear or the server lacks ground truth, the reported comparison does not carry over automatically.

Editorial extensions

If this is right

New agents can join the network by training a small adapter on calibration examples instead of retraining all agents, so the network can expand without disrupting existing agents.
Because only compact belief updates cross the channel, not raw observations or full model weights, the framework exposes less private information than parameter-sharing schemes.
At equal communication traffic, the proposed scheme reduces belief error by roughly 68 percent compared with homogeneous synchronization.
As the number of distinct model architectures grows, the proposed scheme's belief error stays relatively stable, while no-synchronization and homogeneous-only baselines degrade.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The linear known-world-vector assumption is doing heavy lifting: for real deep models, whose latent spaces are nonlinear and entangled, the pseudoinverse rule would need to be replaced by a learned translator, turning the framework into a supervised cross-model alignment problem. The paper itself flags this as an open challenge.
The same server-side translation mechanism could serve as a hub for federated learning with heterogeneous client architectures, where client models are the agents and the MEC server coordinates belief alignment instead of standard parameter averaging; the paper does not draw this connection.
Change detection plus relevance filtering suggests a scheduling optimization: the value of propagating a belief update depends on downstream task error, so future work could derive update policies that trade synchronization freshness against bandwidth; this is a natural next test.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

The short version: the architecture idea is reasonable, and the writing is honest, but the central claim is not supported by the evidence. The abstract promises a "latent translation model" that translates belief updates across heterogeneous agents without joint training, but the case study replaces that model with a Moore-Penrose inverse computed from a known ground-truth world vector w. That is an oracle, not a learned translation model, and the paper itself concedes in Section V.B.2 that the translation model remains an open research direction. So the reported 0.123 error and 68.1% improvement over the homogeneous baseline are measurements of a simplified linear proxy, not of the actual proposed mechanism.

What is genuinely new is the conceptual combination: MEC-hosted latent translation, event-triggered receiver-specific propagation, and the treatment of belief alignment as a distinct knowledge layer. Table I is a useful comparison that frames the shortcomings of prior work. The paper is clearly written and commendably upfront about its limitations.

The soft spot is load-bearing. The case study assumes each agent's latent vector is a linear function of a single global world vector w, and that the MEC server knows w exactly. Real neural agents have nonlinear, entangled latent spaces, and servers do not have access to ground truth. The paper also leaves many simulation parameters undetermined—how the agent matrices are generated, what the readout r is, how update magnitudes are set—and no code or data are provided. I also expected to see comparisons with at least a knowledge-distillation style baseline or the DeepSC joint-training baseline, not just homogeneous synchronization. These omissions do not need to be fatal for a position paper, but they are fatal for the abstract's claim of validation.

If the paper were reframed as a vision paper with a clearly labeled toy linear case, it could serve as a useful resource for the semantic communication community. As is, the gap between claim and evidence is too large for an archival journal. I would send it to peer review only if the authors either implement a real translation model or strictly narrow the claim to the simplified setting. For the current version, the abstract overstates what is validated.

Referee Report

4 major / 5 minor

Summary. The paper proposes a belief synchronization framework for heterogeneous AI agents in 6G networks, centered on latent translation models deployed at MEC servers. The framework is described in four steps (observation, belief update generation, semantic translation, belief propagation), with the claim that it supports different model architectures without joint training, reduces communication overhead, preserves privacy, and limits knowledge drift. The paper reviews prior synchronization approaches, presents a case study using a linear matrix-vector model of agents, and reports quantitative results (Table II) comparing a "proposed scheme" with a homogeneous baseline. The conclusion acknowledges that the case study uses a simplified linear matrix-vector representation and lists several open challenges, including the statement that the latent translation model remains an open research direction.

Significance. If the proposed framework were actually realized, it could address a real and increasingly important problem: semantic communication among heterogeneous AI agents with different architectures and knowledge representations. The paper provides a useful high-level taxonomy of existing approaches and clearly identifies a gap in the literature. However, the central component of the proposed framework, the latent translation model, is not implemented or evaluated in the paper. The case study replaces it with a Moore-Penrose pseudoinverse computed under the assumption that the MEC server knows the ground-truth world vector, which is an oracle assumption. The manuscript itself states that the translation model is an open and promising research direction. Consequently, the current submission does not demonstrate the claimed contribution; its value is prospective rather than demonstrated, and the quantitative results do not validate the proposed mechanism.

major comments (4)

[Section IV, case study setup] The case study validates an oracle linear pseudoinverse, not the latent translation model that is the paper's central claim. The text states that 'the value of w is known in the system' and the translation operator is computed as the Moore-Penrose inverse of the agent matrix X, with updates propagated as Y X^+ x. This requires exact knowledge of the ground-truth world state and assumes each agent's latent vector is a linear function of w. Real agents have nonlinear, architecture-specific latent spaces, and a MEC server would not have access to the true world state. Therefore, the results in Table II (0.123 vs. 0.385 error) measure the performance of an oracle pseudoinverse, not the performance of the proposed latent translation framework.
[Sections V.A.2 and V.B.2, future directions] The manuscript explicitly concedes that the central mechanism does not yet exist. Section V.A.2 states that the latent translation model 'cannot be performed using simple MLPs' and that a unique architecture must be developed, and Section V.B.2 states that 'the latent translation model proposed in this study remains an open and promising research direction.' This directly contradicts the abstract's claim that the framework 'uses latent translation models' and that the case study validates the framework. The paper's central advertised contribution is therefore not supported by the presented evidence.
[Section IV, simulation parameters and reproducibility] Key simulation parameters are missing, preventing reproduction and interpretation of the quantitative results. The paper specifies n=1024 but does not specify the number of agents, the set of latent dimensions m used to realize the heterogeneity index, the world-change magnitude delta, the readout vectors r, the time horizon, or the synchronization triggering schedule. Without these values, Figs. 3 and 4 and Table II cannot be independently reproduced or properly assessed, which is a serious issue for a quantitative validation of the central claim.
[Section IV, baseline comparison fairness] The comparison against the homogeneous baseline is not controlled for the information available to each method. The proposed scheme is given exact access to the ground-truth world vector w, while the no-synchronization and homogeneous baselines are not. The reported 68.1% error reduction over the homogeneous baseline could therefore be due to this oracle access rather than to any property of latent translation. A fair comparison would either provide the same world-state information to all methods or remove the oracle assumption entirely.

minor comments (5)

[Section IV, notation] The notation 'X^{-1}_{n x m}' is confusing because the inverse of a non-square matrix does not exist; the subsequent use of the Moore-Penrose inverse should be defined explicitly and consistently throughout the section.
[Section IV, figure references] Figures 3 and 4 are referenced in the text and discussed in detail, but the actual figures are not present in the manuscript text provided; please ensure the submitted file includes all figure graphics.
[Section IV, Table II] Table II reports 'Full Model Transfer' with error 0.000±0.001, which is surprising if 'full model transfer' means transmitting all model parameters; the mechanism by which this achieves near-perfect alignment and the distinction from the proposed scheme should be clarified.
[Section IV, baseline homogeneous scheme] The description of the homogeneous baseline states that synchronization occurs only between models of matching architectures, but the paper does not specify how an agent is matched or how the communication traffic remains identical to the proposed scheme; this should be explained for the comparison in Table II to be meaningful.
[Section V.A.4, catastrophic forgetting] The discussion of catastrophic forgetting states that belief synchronization updates 'weights and biases' of all agents, but the earlier case study uses linear matrix representations without any neural network weights; the connection between the case study and this challenge should be made explicit.

Circularity Check

1 steps flagged · score 3.0 of 10

The case study's low error is built into the known-w oracle setup: the translation operator is computed from the same matrices and ground truth that define the ideal, so Table II validates a linear pseudoinverse, not the proposed latent translation model.

self definitional [Section IV, Case Study (system model and error metric); cf. Section V.B.2]
"Since the value of w is known in the system, the latent translation model to convert belief from the latent space of agent x to the latent space of model y is approximated via the matrix X^{-1}_{n×m}. ... we use the Moore-Penrose inverse of X to approximate the latent translation model. The change in belief ΔB from agent x to y is obtained via Y X^{-1} x where x is the latent vector of model x. ... the error is |r x − r Rx| where r is the fixed readout defined for agent x, x is the latent vector, and Rx is the latent vector of agent x obtained by directly observing the world state w."

The translation operator is constructed algebraically from the same agent matrices and the same known world vector w that define the ideal target Rx. Since the MEC server knows w, the translated belief is, up to the Moore-Penrose residual, equal to the quantity the error metric treats as ground truth. The reported 0.123 error and 68.1% improvement are therefore residuals of a known linear projection, not evidence that a learned latent translation model can synchronize heterogeneous neural agents. The paper itself concedes the actual translation task is non-trivial (V.A.1), cannot be performed by MLPs (V.A.2), and that 'the latent translation model proposed in this study remains an open and promising research direction' (V.B.2).

full rationale

No fitted parameters are hidden in the reported errors, and the comparison against homogeneous synchronization is a genuine simulation with defined baselines, so this is not a case of 'fitted input called prediction' in the usual sense. Nor are there load-bearing self-citations: all references are external. However, the central claim — that a MEC-hosted latent translation model enables belief synchronization across heterogeneous agents — is not what the case study exercises. The case study replaces the proposed model with the Moore-Penrose inverse of known linear agent matrices, and it lets the MEC server know the true world vector w that the agents are trying to estimate. Because the ideal latent vector Rx is defined as the vector obtained by directly observing that same w, the translation operator is effectively given the answer before the error is measured. The low error in Table II is therefore a mathematical consequence of the linear oracle setup rather than an empirical demonstration of the proposed framework. This is a partial, construction-level circularity in the validation: score 3, not higher, because the paper does not fit constants to the target errors and the simulation itself is internally consistent.

Assumptions & free parameters 6 free parameters · 5 assumptions · 1 invented entities

The central claim rests on a linear, ground-truth-visible simulation rather than on the actual proposed mechanism. All simulation constants are chosen by hand and several are unspecified. The latent translation model, the paper's main invented entity, has no architecture or training and is explicitly left as future work.

free parameters (6)

World state dimension n = 1024
Set to 1024 in Section IV; the value is arbitrary and affects matrix dimensions and simulation scale.
World change magnitude delta = not stated
The text says w is changed by a fixed magnitude delta but never gives its value; error growth rates depend on it.
Latent dimension set {m} = not stated
Each agent is assigned an m value determining latent-space width; the heterogeneity index counts distinct m values, but the actual values and number of agents per layer are not specified.
Readout vector r = not stated
The error metric |r x - r x_hat| depends on a fixed readout r defined for each agent; its construction is not specified.
Agent latent matrices X and Y = not stated
Agent latent matrices are random, but their distribution and generation are unspecified; the pseudoinverse translation and results depend on them.
Number of agents = not stated
The network composition, including how many agents exist at each heterogeneity index, is not given.

assumptions (5)

domain assumption Agents' latent spaces are linear functions of a single shared ground-truth world vector w.
Section IV models an agent as an n x m matrix and latent vectors as derived from w; real neural agents have nonlinear, architecture-specific representations.
ad hoc to paper The MEC server knows the true world state w.
Section IV states "the value of w is known in the system", which removes the estimation and learning problem and trivializes cross-agent translation.
ad hoc to paper Latent translation via the Moore-Penrose inverse preserves belief semantics.
The paper assumes Y X^{-1} x is a valid belief update for the receiver; this is an assumption not tested on real models.
domain assumption The world state changes by a fixed magnitude delta at discrete update events, with one agent observing and propagating per event.
This is the simulation mechanics in Section IV and is not justified for real continuous-time operations.
standard math The Moore-Penrose inverse is a valid standard mathematical tool for the linear least-squares approximation.
Used in Section IV as the linear translation operator; this is standard background mathematics.

invented entities (1)

Latent translation model
purpose: Translate a belief update from one agent's latent space to another agent's latent space on a MEC server, enabling heterogeneous agents to synchronize without joint training.
No architecture, training procedure, or learned weights are provided. Section V-B states this remains an open research direction, and the case study substitutes a Moore-Penrose inverse, so the entity has no independent empirical support.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Heterogeneity-Aware Belief Synchronization for Semantic Communication in AI-Native 6G Networks." pith.science (2026). https://pith.science/paper/FHMCTNKQ

@misc{pith2026260813394,
  author       = {Pith},
  title        = {Pith review of: Heterogeneity-Aware Belief Synchronization for Semantic Communication in AI-Native 6G Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FHMCTNKQ}},
  note         = {Machine review of arXiv:2608.13394}
}

read the original abstract

6G networks will not be serving as communication infrastructures only; rather, they are expected to evolve into intelligent systems, where thousands of autonomous artificial intelligence (AI) agents are interconnected. The agents are deployed across a wide range of platforms including low Earth orbit (LEO) satellites, high-altitude platforms (HAPs), unmanned aerial vehicles (UAVs), edge servers, and terrestrial devices. These agents continuously observe their environment and exchange information. Semantic communication provides an efficient mechanism for exchanging meaningful information instead of raw data. However, its effectiveness depends on the communicating agents having sufficiently aligned beliefs to correctly interpret and decode the transmitted messages. This assumption becomes difficult to satisfy in the 6G network where heterogeneous AI models operate under diverse computational constraints and continuously acquire different knowledge from their local environments. This article presents a heterogeneity-aware belief synchronization framework for 6G AI-native networks. It uses latent translation models deployed on multi-access edge computing (MEC) servers. These models translate belief updates from one agent to agent-specific knowledge without requiring joint training and a homogeneous architecture of models. By exchanging compact belief updates through a latent translation model only when necessary, the framework preserves privacy, reduces synchronization cost, and minimizes local knowledge drift. We validate the framework through a case study on a multi-layered terrestrial/non-terrestrial network. Results demonstrate that it maintains low synchronization cost, measured by the number of parameters transmitted, and low belief alignment error across the heterogeneous agents in the case study.

Figures

Figures reproduced from arXiv: 2608.13394 by the authors.

**Figure 1.** Multi-layer 6G network architecture with heterogeneous AI agents and proposed belief synchronization framework. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

**Figure 2.** The proposed four-step belief synchronization workflow. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** System error trend with increasing heterogeneity [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 9 canonical work pages

[1]

Distributed coordination for heterogeneous non-terrestrial networks,

J. Deng, H. Zhou, and M.-S. Alouini, “Distributed coordination for heterogeneous non-terrestrial networks,”IEEE Commun. Mag., vol. 64, no. 6, pp. 74-80, Jun. 2026

work page 2026
[2]

The roadmap to 6G: AI empowered wireless networks,

K. B. Letaief, W. Chen, Y . Shi, J. Zhang, and Y .-J. A. Zhang, “The roadmap to 6G: AI empowered wireless networks,”IEEE Commun. Mag., vol. 57, no. 8, pp. 84–90, Aug. 2019

2019
[3]

Synchronizing LLM-based semantic knowledge bases via secure federated fine-tuning in semantic communication,

L. Li, Y . He, R. Xu, B. Chen, B. Han, Y . Zhao, and J. Li, “Synchronizing LLM-based semantic knowledge bases via secure federated fine-tuning in semantic communication,”Front. Artif. Intell., vol. 8, p. 1690950, 2025

work page 2025
[4]

Deep learning enabled semantic communication systems,

H. Xie, Z. Qin, G. Y . Li, and B.-H. Juang, “Deep learning enabled semantic communication systems,”IEEE Trans. Signal Process., vol. 69, pp. 2663–2675, Apr. 2021

2021
[5]

Collaborative learning for task-oriented semantic communications: Overcoming data mismatch between transceivers,

Y . Wang, R. Li, C. Wang, J. Ye, C. Feng, and S. Guo, “Collaborative learning for task-oriented semantic communications: Overcoming data mismatch between transceivers,”IEEE Open J. Commun. Soc., vol. 6, pp. 5778–5794, 2025

work page 2025
[6]

Efficient knowledge base synchro- nization in semantic communication network: A federated distillation approach,

X. Lu, K. Zhu, J. Li, and Y . Zhang, “Efficient knowledge base synchro- nization in semantic communication network: A federated distillation approach,” inProc. IEEE Wireless Commun. Netw. Conf. (WCNC), Dubai, UAE, Apr. 2024, pp. 1–6

work page 2024
[7]

Performance optimization of semantic communications with heterogeneous knowl- edge: An adversarial reinforcement learning approach,

J. Zhang, Y . Zhu, T. Luo, Z. Zhang, Y . Hu, and M. Chen, “Performance optimization of semantic communications with heterogeneous knowl- edge: An adversarial reinforcement learning approach,”IEEE Trans. Commun., vol. 73, no. 11, pp. 11123–11137, Nov. 2025

work page 2025
[8]

Semantic communication: A survey on research landscape, challenges, and future directions,

T. M. Getu, G. Kaddoum, and M. Bennis, “Semantic communication: A survey on research landscape, challenges, and future directions,”Proc. IEEE, vol. 112, no. 11, pp. 1649–1685, Nov. 2024

2024

Show all 14 references

[9]

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 Commun. Surveys Tuts., vol. 25, no. 1, pp. 213–250, 2023

2023
[10]

Beyond transmitting bits: Context, seman- tics, and task-oriented communications,

D. Gündüz, Z. Qin, I. E. Aguerri, H. S. Dhillon, Z. Yang, A. Yener, K. K. Wong, and C.-B. Chae, “Beyond transmitting bits: Context, seman- tics, and task-oriented communications,”IEEE J. Sel. Areas Commun., vol. 41, no. 1, pp. 5–41, Jan. 2023

2023
[11]

C. E. Shannon and W. Weaver,The Mathematical Theory of Communi- cation. Urbana, IL, USA: Univ. Illinois Press, 1949

1949
[12]

A survey on semantic communication networks: Architecture, security, and privacy,

S. Guo, Y . Wang, N. Zhang, Z. Su, T. H. Luan, Z. Tian, and X. Shen, “A survey on semantic communication networks: Architecture, security, and privacy,”IEEE Commun. Surveys Tuts., vol. 27, no. 5, pp. 2860–2894, Oct. 2025

2025
[13]

A survey on goal oriented semantic communication: Techniques, challenges, and future directions,

T. M. Getu, G. Kaddoum, and M. Bennis, “A survey on goal oriented semantic communication: Techniques, challenges, and future directions,” IEEE Access, vol. 12, pp. 51223–51274, Dec. 2024

2024
[14]

6G networks: Beyond Shannon towards semantic and goal-oriented communications,

E. C. Strinati and S. Barbarossa, “6G networks: Beyond Shannon towards semantic and goal-oriented communications,”Comput. Netw., vol. 190, p. 107930, May 2021

2021

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Distributed coordination for heterogeneous non-terrestrial networks,

J. Deng, H. Zhou, and M.-S. Alouini, “Distributed coordination for heterogeneous non-terrestrial networks,”IEEE Commun. Mag., vol. 64, no. 6, pp. 74-80, Jun. 2026

work page 2026

[2] [2]

The roadmap to 6G: AI empowered wireless networks,

K. B. Letaief, W. Chen, Y . Shi, J. Zhang, and Y .-J. A. Zhang, “The roadmap to 6G: AI empowered wireless networks,”IEEE Commun. Mag., vol. 57, no. 8, pp. 84–90, Aug. 2019

2019

[3] [3]

Synchronizing LLM-based semantic knowledge bases via secure federated fine-tuning in semantic communication,

L. Li, Y . He, R. Xu, B. Chen, B. Han, Y . Zhao, and J. Li, “Synchronizing LLM-based semantic knowledge bases via secure federated fine-tuning in semantic communication,”Front. Artif. Intell., vol. 8, p. 1690950, 2025

work page 2025

[4] [4]

Deep learning enabled semantic communication systems,

H. Xie, Z. Qin, G. Y . Li, and B.-H. Juang, “Deep learning enabled semantic communication systems,”IEEE Trans. Signal Process., vol. 69, pp. 2663–2675, Apr. 2021

2021

[5] [5]

Collaborative learning for task-oriented semantic communications: Overcoming data mismatch between transceivers,

Y . Wang, R. Li, C. Wang, J. Ye, C. Feng, and S. Guo, “Collaborative learning for task-oriented semantic communications: Overcoming data mismatch between transceivers,”IEEE Open J. Commun. Soc., vol. 6, pp. 5778–5794, 2025

work page 2025

[6] [6]

Efficient knowledge base synchro- nization in semantic communication network: A federated distillation approach,

X. Lu, K. Zhu, J. Li, and Y . Zhang, “Efficient knowledge base synchro- nization in semantic communication network: A federated distillation approach,” inProc. IEEE Wireless Commun. Netw. Conf. (WCNC), Dubai, UAE, Apr. 2024, pp. 1–6

work page 2024

[7] [7]

Performance optimization of semantic communications with heterogeneous knowl- edge: An adversarial reinforcement learning approach,

J. Zhang, Y . Zhu, T. Luo, Z. Zhang, Y . Hu, and M. Chen, “Performance optimization of semantic communications with heterogeneous knowl- edge: An adversarial reinforcement learning approach,”IEEE Trans. Commun., vol. 73, no. 11, pp. 11123–11137, Nov. 2025

work page 2025

[8] [8]

Semantic communication: A survey on research landscape, challenges, and future directions,

T. M. Getu, G. Kaddoum, and M. Bennis, “Semantic communication: A survey on research landscape, challenges, and future directions,”Proc. IEEE, vol. 112, no. 11, pp. 1649–1685, Nov. 2024

2024