Adaptive Role Assignment and Controlled Feature Reconstruction for Federated Image Semantic Communication

Yuxuan Chen; Jiao Bie; Hang Qi; Jingjin Wu

cs.LGOct 11, 2026

Adaptive Role Assignment and Controlled Feature Reconstruction for Federated Image Semantic Communication

Authors: Yuxuan Chen, Jiao Bie, Hang Qi, Jingjin Wu

Organizations: Guangdong Prov./Zhuhai Key Lab of IRADS, Beijing Normal-Hong Kong Baptist University, Guangdong, P. R. China

Abstract

Hybrid federated semantic communication enables user equipment (UEs) to contribute model updates or semantic features under a shared training framework. However, heterogeneous role costs and the interaction between model aggregation and feature reconstruction (FR) remain key challenges in practical radio access network (RAN) deployment. To address these issues, we propose ARC-FedSC, a RAN-aware framework that coordinates UE role assignment and server-side controlled FR under a fixed communication budget. Within ARC-FedSC, RAN Aware Role-Bandit learns the historical contribution of UEs in different roles together with current uplink conditions, while Temporal Consistency Alignment (TCA) regulates FR according to recent model aggregation directions. Experiments on CIFAR-10 with time-varying RAN channels show that ARC-FedSC outperforms baselines across all five tested role configurations in the later training stage, with the best configurations achieving about 24.17 dB average PSNR over the last 20 rounds. Component level results further show that Role-Bandit maintains an average rolling UE coverage of 85.3%, compared with about 65.7% for baseline, while TCA reduces the probability of harmful FR events from 0.245 to 0.093, corresponding to a 62.0% reduction. Ablation and sensitivity studies further support the effectiveness and robustness of the proposed design.

Figures & tables

Fig. 1 : Overall architecture of the proposed ARC-FedSC framework, including the hybrid FL and FR architecture, Temporal Consistency Alignment, and Role-Bandit based UE selection and role allocation.

Fig. 2 : Test PSNR under different model-role and feature-role configurations of ARC-FedSC, comparing with baselines.

Fig. 3 : 20-round rolling UE coverage. Solid/dashed lines denote the mean over three random seeds, and shaded regions indicate ±1 standard deviation.

Fig. 4 : Distribution of per-round FR gain for the original FR and TCA controlled FR across training rounds. The dashed line at

\Delta_{t}=-0.05

dB explicitly marks the threshold used to identify harmful FR events.

Method	Last-20 PSNR (dB)	Round to 23.5 dB
FedAvg	$22.540\pm 0.176$	$>200$
FedSFR	$23.231\pm 0.096$	$>200$
Role-Bandit only	$23.683\pm 0.125$	$139.0$
TCA only	$23.736\pm 0.217$	$118.7$
ARC-FedSC	$\mathbf{24.030\pm 0.186}$	$\mathbf{98.7}$

TABLE I : Ablation results under

(K_{m},K_{o})=(10,10)

averaged over three random seeds.

Variant	Last-20 PSNR (dB)	$\Delta$ PSNR (dB)
w/o Communication	$23.587\pm 0.217$	$-0.096$
w/o Exploration	$23.608\pm 0.192$	$-0.075$
w/o Staleness	$23.613\pm 0.168$	$-0.070$
Full Role-Bandit	$\mathbf{23.683\pm 0.125}$	–

TABLE II : Component ablation of Role-Bandit under

(K_{m},K_{o})=(10,10)

averaged over three random seeds.

Role-Bandit UE selection	Parameter Value	Last-20 PSNR (dB)
Discount factor $\lambda$	0.90	23.636
	0.95	23.637
	0.97	23.683
	0.99	23.657
	1.00	23.630
Exploration factor $\xi_{\exp}$	0.50	23.664

TABLE III : Parameter sensitivity of Role-Bandit UE selection and TCA FR under

(K_{m},K_{o})=(10,10)

. Default parameter settings are shown in bold.

Semantic Communication

Semantic communication sends meaning- or task-relevant information so receivers can act without reconstructing every transmitted bit or signal.

See all 100 papers

FedGenSC: Federated Generative Semantic Communication with Channel-Aware Adaptation
Sep 8, 2026Rita Abou Fares, Razan Al Kakoun, Maher Nouiehed +1Semantic Communication Generative Adversarial Networks
Department of Electrical and Computer Engineering, American University of Beirut, Beirut, Lebanon · Department of Industrial Engineering, American University of Beirut, Beirut, Lebanon
Non-Coherent Over-the-Air Federated Learning: Protocol, Convergence, and Device Scheduling
Sep 8, 2026Haifeng Wen, Nicolò Michelusi, Osvaldo Simeone +2Federated Learning Aggregation Communication-Efficient Distributed Training
IoT Thrust, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, 511453, China · School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ 85281 USA · Institute for Intelligent Networked Systems, Northeastern University London, London, E1 8PH, UK +2
FedCritic-MIMO: Communication-Efficient Serverless Federated Critic Learning for Massive-MIMO Resource Control in Open and Disaggregated 6G RANs
Aug 4, 2026Amin Farajzadeh, Melike Erol-KantarciFederated RL 6G Networks
School of Electrical Engineering and Computer Science, University of Ottawa, Ottawa, ON K1N 6N5, Canada · with the NETCORE Lab

Sep 8, 2026J/K move · Enter open · S save

FedGenSC: Federated Generative Semantic Communication with Channel-Aware Adaptation

Rita Abou Fares, Razan Al Kakoun, Maher Nouiehed, Hadi Sarieddeen

Department of Electrical and Computer Engineering, American University of Beirut, Beirut, Lebanon · Department of Industrial Engineering, American University of Beirut, Beirut, Lebanon

Semantic Communication Generative Adversarial Networks

Open paper AskPDF arXiv

Integrating generative adversarial networks (GANs) into federated semantic communication (SemCom) is a natural progression, as generative priors can recover semantic fidelity under channel distortion that discriminative decoders cannot. However, naive GAN federation introduces three failure modes that prior work has, to the best of our knowledge, neither identified nor resolved: discriminator aggregation instability under non-independent and identically distributed (non-IID) data, semantic drift caused by divergent local embedding spaces, and channel-agnostic generation that cannot adapt to heterogeneous link conditions. We propose federated generative semantic communication (FedGenSC), which mitigates all three by employing a global generator with local-only discriminators, providing cross-client semantic information through a semantic prototype bank, and conditioning generation on the instantaneous signal-to-noise ratio (SNR). Experiments on the Europarl dataset over Rayleigh fading channels (K=10 clients, Dirichlet α=0.5) show that FedGenSC under non-IID data outperforms the FedDeepSC baseline across the tested SNR range, achieving up to a 58.2% relative improvement in bilingual evaluation understudy (BLEU)-1 at 18 dB. Ablation studies confirm the independent contribution of each component.

Abstract

Semantic Communication

Explore similar work

FedGenSC: Federated Generative Semantic Communication with Channel-Aware Adaptation

Non-Coherent Over-the-Air Federated Learning: Protocol, Convergence, and Device Scheduling

FedCritic-MIMO: Communication-Efficient Serverless Federated Critic Learning for Massive-MIMO Resource Control in Open and Disaggregated 6G RANs