Caili Guo

Physical layer key generation (PLKG) enables wireless devices to establish shared secret keys by exploiting channel reciprocity. However, most existing PLKG methods rely on single-modality features, which provide limited shared randomness and are sensitive to channel non-reciprocity in frequency division duplexing (FDD) systems, resulting in high key disagreement rates (KDR). To address this challenge, this work proposes a hierarchical multimodal feature fusion (HiMFF) scheme that jointly leverages time, frequency, and spatial modality features to capture complementary channel characteristics, and thereby provides rich and consistent shared randomness for reliable key generation under the non-reciprocal conditions of FDD systems. First, a preprocessing module is employed to transform raw multimodal channel observations into structured feature representations. Then, a multimodal fusion module is developed to align and integrate multimodal features across various scales effectively. Finally, a large language model is adapted through fine-tuning to produce reciprocal channel features, ensuring reliable key generation. Simulation results show that the proposed scheme outperforms current baselines, reducing KDR by up to 9.2%.

Large-scale models (LSMs) can be an effective framework for semantic representation and understanding, thereby providing a suitable tool for designing semantic communication (SC) systems. However, their direct deployment is often hindered by high computational complexity and resource requirements. In this paper, a novel robust knowledge distillation based semantic communication (RKD-SC) framework is proposed to enable efficient and channel-noise-robust LSM-powered SC. The framework addresses two key challenges: determining optimal compact model architectures and effectively transferring knowledge while maintaining robustness against channel noise. First, a knowledge distillation-based lightweight differentiable architecture search (KDL-DARTS) algorithm is proposed. This algorithm integrates knowledge distillation loss and a complexity penalty into the neural architecture search process to identify high-performance, lightweight semantic encoder architectures. Second, a novel two-stage robust knowledge distillation (RKD) algorithm is developed to transfer semantic capabilities from an LSM (teacher) to a compact encoder (student) and subsequently enhance system robustness. To further improve resilience to channel impairments, a channel-aware transformer (CAT) block is introduced as the channel codec, trained under diverse channel conditions with variable-length outputs. Extensive simulations on image classification tasks demonstrate that the RKD-SC framework significantly reduces model parameters while preserving a high degree of the teacher model’s performance and exhibiting superior robustness compared to existing methods.

U-shaped Split Federated Learning (U-SFL) is a promising paradigm for distributed image coding, offering parallel training capabilities and privacy preservation while mitigating computational burdens on edge devices. However, the frequent bidirectional transmission of intermediate features between dual-split points incurs substantial communication overhead. To mitigate this issue, we propose a compact-feature U-shaped split federated learning framework (CoF U-SFL), which reduces communication overhead and improves training efficiency while maintaining low image distortion. We introduce a feature entropy estimation network to model the distribution of split-layer features, enabling effective compression during transmission. Furthermore, we formulate a joint optimization objective incorporating entropy constraints to guide the end-to-end training. Experimental results demonstrate that CoF U-SFL reduces communication overhead by 104.6 times while maintaining reconstruction performance.

Remote estimation is vital in Internet of Things (IoT) networks. However, in multi-cell Fog Radio Access Networks (F-RAN), it faces significant challenges due to limited spectrum resources and constrained fronthaul link capacity, leading to distortion and degraded system performance. To address these issues, we propose a novel vector quantization and information bottleneck-inspired type-based multiple access (VQ-IB-TBMA) protocol tailored for the F-RAN scenario, aiming to enhance remote estimation accuracy and improve resource utilization. Specifically, we leverage information bottleneck (IB) theory to enhance type-based multiple access (TBMA) techniques and incorporate vector quantization (VQ) for signal quantization at the edge nodes (ENs). To realize these improvements, we design a tailored loss function and a two-phase training strategy, optimizing the semantic knowledge base within the proposed protocol. Numerical experiments demonstrate that, compared to edge-based remote estimation methods, the VQ-IB-TBMA protocol significantly improves system performance in terms of probability of error and spectrum resource utilization.

While deep learning has considerably advanced semantic communication, image transmission techniques continue to face persistent challenges, particularly in capturing spatial structure and preserving semantic integrity against channel noise. To address these issues, we propose a pose-guided end-to-end semantic communication system with a semantic knowledge base (Pose-VAE). Specifically, we integrate human pose information to guide semantic feature extraction, enabling the network to model spatial structures. In addition, we design a novel semantic knowledge base (SKB) that encodes clustered pose categories as structural priors to enhance robustness against channel noise. Extensive experiments demonstrate that Pose-VAE significantly improves image reconstruction quality, particularly under the low signal-to-noise ratio (SNR) conditions.

U-shaped split federated learning (U-SFL) has been widely applied in image coding, as it can effectively balance parallel training, model privacy, and local computational cost. The performance of U-SFL largely depends on the selection of split points and the aggregation frequency, and thus, model splitting (MS) and model aggregation (MA) strategies are critical. In uncrewed aerial vehicle scenarios, fluctuating communication links and computational resources can significantly impact U-SFL performance. To address this challenge, we propose a resource-adaptive U-SFL (AU-SFL) framework, which adaptively selects optimal MS and MA strategies based on the available computational and communication resources. Specifically, we first conduct a theoretical convergence analysis that systematically quantifies the individual and joint impacts of MS and MA strategies on convergence behavior. Then, we formulate an optimization problem aimed at minimizing training latency, grounded in the theoretical convergence analysis. Subsequently, an alternating MS–MA algorithm is proposed to solve the problem, which is decomposed into two subproblems and solved alternatively. Extensive experiments on multiple benchmark datasets demonstrate that AU-SFL achieves a $1.61\times $ reduction in convergence time while improving multiscale structural similarity index (MS-SSIM) by 4.4%, conclusively validating our adaptive optimization strategy’s effectiveness.