The growing demand for large artificial intelligence model (LAIM) services is driving a paradigm shift from traditional cloud-based inference to edge-based inference for low-latency, privacy-preserving applications. In particular, edge-device co-inference, which jointly executes LAIM inference across edge devices and servers, has emerged as a promising strategy for resource-efficient LAIM execution in wireless networks. In this paper, we investigate a pruning-aware LAIM co-inference scheme, where a pre-trained LAIM is pruned and partitioned into on-device and on-server sub-models for deployment. For analysis, we first prove that the LAIM output distortion is upper bounded by its parameter distortion. Then, we derive a lower bound on the parameter distortion via rate-distortion theory, analytically capturing the relationship between pruning ratio and co-inference performance. Next, based on the analytical results, we formulate an LAIM co-inference distortion bound minimization problem by jointly optimizing the pruning ratio, split point, transmit power, and computation frequency under system latency, energy, and available resource constraints. Moreover, we propose an efficient algorithm to tackle the considered highly non-convex problem. Finally, extensive experimental results demonstrate the effectiveness of the proposed design. In particular, model parameter distortion is shown to provide a reliable bound on output distortion. Also, the proposed joint design achieves superior performance in balancing trade-offs among inference performance, system latency, and energy consumption compared with various benchmark schemes.

Deploying large language models (LLMs) in mobile and edge computing environments is constrained by limited on-device resources, scarce wireless bandwidth, and frequent model evolution. Although edge-cloud collaborative inference with speculative decoding (SD) can reduce end-to-end latency by executing a lightweight draft model at the edge and verifying it with a cloud-side target model, existing frameworks fundamentally rely on tight coupling between the two models. Consequently, repeated model synchronization introduces excessive communication overhead, increasing end-to-end latency, and ultimately limiting the scalability of SD in edge environments. To address these limitations, we propose FlexSpec, a communication-efficient collaborative inference framework tailored for evolving edge-cloud systems. The core design of FlexSpec is a shared-backbone architecture that allows a single and static edge-side draft model to remain compatible with a large family of evolving cloud-side target models. By decoupling edge deployment from cloud-side model updates, FlexSpec eliminates the need for edge-side retraining or repeated model downloads, substantially reducing communication and maintenance costs. Furthermore, to accommodate time-varying wireless conditions and heterogeneous device constraints, we develop a channel-aware adaptive speculation mechanism that dynamically adjusts the speculative draft length based on real-time channel state information and device energy budgets. Extensive experiments demonstrate that FlexSpec achieves superior performance compared to conventional SD approaches in terms of inference efficiency.

Channel state information (CSI) is a fundamental component in both wireless communication and sensing systems, enabling critical functions such as radio resource optimization and environmental perception. In wireless sensing, data scarcity and packet loss hinder efficient model training, while in wireless communication, high-dimensional CSI matrices and short coherent times caused by high mobility present challenges in CSI estimation. To address these issues, we propose a unified framework named CSI-BERT2 for CSI prediction and classification tasks, built on our previous work CSI-BERT, which adapts BERT to capture the complex relationships among CSI sequences through a bidirectional self-attention mechanism. We introduce a two-stage training method that first uses a mask language model (MLM) to enable the model to learn general feature extraction from scarce datasets in an unsupervised manner, followed by fine-tuning for specific downstream tasks. Specifically, we extend MLM into a mask prediction model (MPM), which efficiently addresses the CSI prediction task. To further enhance the representation capacity of CSI data, we modify the structure of the original CSI-BERT. We introduce an adaptive re-weighting layer (ARL) to enhance subcarrier representation and a multi-layer perceptron (MLP)-based temporal embedding module to mitigate temporal information loss problem inherent in the original Transformer. Extensive experiments on both real-world collected and simulated datasets demonstrate that CSI-BERT2 achieves state-of-the-art performance across all tasks. Our results further show that CSI-BERT2 generalizes effectively across varying sampling rates and robustly handles discontinuous CSI sequences caused by packet loss-challenges that conventional methods fail to address. The dataset and code are publicly available athttps://github.com/RS2002/CSI-BERT2.

This paper exploits the networked integrated sensing and communications (ISAC) to support low-altitude economy (LAE), in which a set of networked ground base stations (GBSs) cooperatively transmit joint information and sensing signals to communicate with multiple authorized uncrewed aerial vehicles (UAVs) and concurrently detect unauthorized objects over the interested region in the three-dimensional (3D) space. We assume that each GBS is equipped with uniform linear array (ULA) antennas, which are deployed either horizontally or vertically to the ground. We also consider two types of UAV receivers, which have and do not have the capability of canceling the interference caused by dedicated sensing signals, respectively. Under each setup, we jointly design the coordinated transmit beamforming at multiple GBSs together with the authorized UAVs’ trajectory control and their GBS associations, for enhancing the authorized UAVs’ communication performance while ensuring the sensing requirements. In particular, we aim to maximize the average sum rate of authorized UAVs over a given flight period, subject to the minimum illumination power constraints toward the interested 3D sensing region, the maximum transmit power constraints at individual GBSs, and the flight constraints of UAVs. These problems are highly non-convex and challenging to solve, due to the involvement of binary UAV-GBS association variables as well as the coupling of beamforming and trajectory variables. To solve these non-convex problems, we propose efficient algorithms by using the techniques of alternating optimization, successive convex approximation, and semi-definite relaxation. Numerical results show that the proposed joint coordinated transmit beamforming and UAV trajectory designs efficiently balance the sensing-communication performance tradeoffs and significantly outperform various benchmarks. It is also shown that the horizontally placed antennas lead to enhanced performance compared with their vertical counterparts due to the more flexible multi-beam design, and the sensing interference cancellation ability at UAV receivers is advantageous for further enhancing ISAC performance.

The map matching of cellular data reconstructs real trajectories of users by exploiting the sequential connections between mobile devices and cell towers. The difficulty in obtaining paired cellular-GPS data and the cellular variation compromise the accuracy and reliability of existing map matching approaches. In this paper, we propose a novelunsupervisedconvolutional network for cellularmapmatching (UCMM) to address these challenges. UCMM employs a dual encoder-decoder network to capture a shared representation from both the cellular and GPS domains in an unsupervised manner. It leverages a dedicated convolutional architecture to tackle the varying lengths of output sequential data. An attention mechanism is specially introduced to deal with the cellular variation. The effectiveness of UCMM is demonstrated through comprehensive evaluations, which show that UCMM achieves a substantial improvement in matching accuracy and deduction of training time compared with the best-known prior works. These improvements make UCMM a significant advancement in the field of map matching.

In some applications, edge learning is experiencing a shift in focus from conventional learning from scratch to two-stage learning combining pre-training and task-specific fine-tuning. This paper considers the problem of joint communication and computation resource management in a two-stage edge learning system. In this system, model pre-training is first conducted at an edge server via centralized learning on local pre-stored general data, and then task-specific fine-tuning is performed at edge devices based on the pre-trained model via federated edge learning. For the two-stage learning model, we first analyze the convergence behavior (in terms of the average squared gradient norm bound), which characterizes the impacts of various system parameters, such as the number of learning rounds and batch sizes in the two stages, on the convergence rate. Based on our analytical results, we then propose a joint communication and computation resource management design to minimize an average squared gradient norm bound, subject to constraints on the transmit power, overall system energy consumption, and training delay. The decision variables include the number of learning rounds, batch sizes, clock frequencies, and transmit power control for both pre-training and fine-tuning stages. Finally, numerical results are provided to evaluate the effectiveness of our proposed design. It is shown that the proposed joint resource management over the pre-training and fine-tuning stages well balances the system performance trade-off among the training accuracy, delay, and energy consumption. The proposed design is also shown to effectively leverage the inherent trade-off between pre-training and fine-tuning, which arises from the differences in data distribution between pre-stored general data versus real-time task-specific data, thus efficiently optimizing overall system performance.

Reconfigurable intelligent surfaces (RIS) play a crucial role in enhancing the security of integrated sensing and communication (ISAC) systems. In this paper, RIS is explored to assist the secure transmission of user data in ISAC system. Through the joint design of the transmit beamforming and RIS discrete phase shifter, we aim to maximize user's secure rates while ensuring target sensing performance. Due to the coupling of optimization variables, conventional optimization methods are hard to address this formulated problem. Therefore, a deep reinforcement learning (DRL) scheme by utilizing the soft actor-critic (SAC) and alternating optimization (AO) algorithms is employed to design the transmit beamforming and the RIS discrete phase shifter, respectively. Simulation results indicate that the problem scheme could obtain a significant improvement in enhancing user secure rates compared to other benching scheme.

With the recent advancements in edge artificial intelligence (AI), future sixth-generation (6G) networks need to support new AI tasks such as classification and clustering apart from data recovery. Motivated by the success of deep learning, the semantic-aware and task-oriented communications with deep joint source and channel coding (JSCC) have emerged as new paradigm shifts in 6G from the conventional data-oriented communications with separate source and channel coding (SSCC). However, most existing works focused on the deep JSCC designs for one task of data recovery or AI task execution independently, which cannot be transferred to other unintended tasks. Differently, this paper investigates the JSCC semantic communications to support multi-task services, by performing the image data recovery and classification task execution simultaneously. First, we propose a new end-to-end deep JSCC framework by unifying the coding rate reduction maximization and the mean square error (MSE) minimization in the loss function. Here, the coding rate reduction maximization facilitates the learning of discriminative features for enabling to perform classification tasks directly in the feature space, and the MSE minimization helps the learning of informative features for high-quality image data recovery. Next, to further improve the robustness against variational wireless channels, we propose a new gated deep JSCC design, in which a gated net is incorporated for adaptively pruning the output features to adjust their dimensions based on channel conditions. Finally, we present extensive numerical experiments to validate the performance of our proposed deep JSCC designs as compared to various benchmark schemes. It is shown that our proposed designs simultaneously provide efficient multi-task services, and the proposed gated deep JSCC framework efficiently reduces the communication overhead with only marginal performance loss. It is also shown that performing the classification task on the feature space via coding rate reduction maximization is able to better defend the label corruption than the traditional label-fitting methods.

This paper studies the unmanned aerial vehicle (UAV)-enabled integrated sensing and communication (ISAC), in which UAVs are dispatched as aerial dual-functional access points (APs) that can exploit the UAV maneuver control and strong line-of-sight (LoS) air-to-ground (A2G) links for efficient communication and sensing. In particular, we consider that one UAV-AP, equipped with a vertically placed uniform linear array (ULA), sends combined information and sensing signals to communicate with multiple users and at the same time sense potential targets at interested areas on the ground. Under this setup, we consider two scenarios with quasi-stationary and fully mobile UAVs, in which the UAV is deployed at an optimizable location over the whole ISAC mission period and can fly over different locations during the ISAC mission period, respectively. For the two scenarios, our objective is to jointly design the UAV maneuver (deployment location or flight trajectory) and the transmit beamforming, for maximizing the weighted sum-rate throughput of communication users, while ensuring the sensing beampattern gain requirements, subject to the transmit power and flight constraints. However, due to the ULA consideration at the UAV, the two formulated problems are highly non-convex and very difficult to be optimally solved, as the UAV’s location/trajectory variables are involved on the exponent parts of each entry in the steering vectors, and are closely coupled with the transmit beamforming vectors. To tackle this issue, we propose efficient algorithms to find their suboptimal but high-quality solutions, by using various techniques from convex and non-convex optimization. Finally, numerical results are provided to validate the superiority of our proposed designs as compared to various benchmark schemes with heuristic maneuver designs. It is shown that the joint maneuver and transmit beamforming design efficiently balances the inherent tradeoff between sensing and communication with regards to different beampattern gain thresholds.

This paper studies the unmanned aerial vehicle (UAV)-enabled integrated sensing and communication (ISAC), in which UAVs are dispatched as aerial dual-functional access points (APs) that can exploit the UAV maneuver control and strong line- of-sight (LoS) aerial-to-ground (A2G) links for efficient ISAC. Particularly, we consider a scenario with one UAV-AP equipped with a vertically placed uniform linear array (ULA), which sends combined information and sensing signals to communicate with multiple users and at the same time sense potential targets on the ground. Our objective is to jointly design the UAV trajectory and transmit beamforming to maximize the average weighted sum-rate throughput of communication users over the whole period, subject to the sensing beampattern gain requirements and transmit power constraints over different time slots, as well as practical flight constraints. While the above problem is challenging to solve, we propose an efficient algorithm by adopting the alternating optimization together with the successive convex approximation (SCA) and semidefinite relaxation (SDR). Numerical results are provided to validate the superiority of our proposed designs as compared to various benchmark schemes with heuristic trajectory designs.