As a new paradigm focusing on transmitting the meaning of information, semantic communications (SCs) have been revealed significant potential in alleviating network congestion and improving energy efficiency. By extracting a small-size semantic feature from the large-size raw-data, the communication loads can be reduced at a price of more computing power, i.e., the computational resource used in semantic extraction. In this paper, we explore the computation dimension to improve the communication performance in an uplink SC system. The SC-oriented user first compresses its original data via local computing during other users’ transmission time and then transmits it to the base station within the assigned time. To achieve a balanced tradeoff between communications and computing, we formulate an optimization problem to minimize the energy consumption of all users by jointly considering the compression ratio and time allocation. We first propose an efficient general algorithm that can be applied to different SC models. To gain more insights, we then derive the closed-form solutions for two special cases: equal-time allocation and two-user transmission. The obtained analytical results reveal a pronounced energy saving of adopting SC, especially for large transmitted data size, high transmission energy coefficient, scarce time resources, and poor channel conditions. Simulation results reveal that our proposed SC-based scheme achieves excellent performance in terms of saving energy consumption compared to the conventional communication.

In this paper, we focus on computation-aware offloading for executing deep neural network (DNN) inference tasks in a mobile edge computing (MEC) system. To cope with the challenges of insufficient wireless resources during task offloading, we resort to semantic communications (SCs), through which the users can offload the compressed task data to the edge server for remote execution. Specifically, we establish the relationship between the compression ratio and computation ratio for different DNN tasks. To achieve energy-efficient offloading, we formulate an optimization problem to minimize the energy consumption of all users by jointly optimizing the compression ratio, computation allocation, uploading time, and DNN layer selection. We first consider a special case with the preconfigured time scheduling and derive closed-form solutions to computation allocation and offloading time, which yield a threshold-based structure determined by users’ channel conditions and local computation consumption. Inspired by the characteristics of these optimal solutions, a general low-complexity iterative algorithm is then designed to solve the original non-convex problem. Simulation results demonstrate that our proposed SC-based computation -offloading scheme can substantially reduce users’ energy consumption compared to the conventional offloading and full offloading, especially with scarce wireless resources.

With the rapid advancements in medical imaging technology, super-resolution reconstruction techniques are crucial for enhancing disease diagnosis and treatment precision. However, the inherent resolution limitations of imaging devices and the clinical need for low-dose imaging often result in insufficient resolution. This deficiency leads to a loss of detailed information, adversely affecting diagnostic accuracy. Improving image resolution without increasing radiation dose has therefore become a critical issue in medical imaging. Addressing this challenge, this study proposes an innovative super-resolution reconstruction method for medical images, the Multi-Scale adaptive Convolutional Neural Network (CNN) and Hybrid Transformer model (MSCT). The core of this method is a multi-scale adaptive convolutional module with residual connections that effectively extracts local features of multi-scale images, overcoming traditional CNNs’ limitations in handling complex textures. Additionally, a hybrid Transformer module enhances the perception of global features and captures texture details by leveraging long-range dependencies and global information modeling. By combining the adaptive convolutional module with the Transformer module, this method effectively fuses local and global information, significantly improving image reconstruction accuracy. Experimental results on CT and MRI datasets show that the proposed method achieved average values of 21.96 dB in Peak Signal-to-Noise Ratio (PSNR) and 0.8053 in Structural Similarity Index (SSIM). These results surpass most mainstream methods, demonstrating superior performance in recovering image details and texture information, providing strong technical support for the diagnosis and analysis of medical images.

To provide sustainable energy support while meeting the demands for serving a rapidly increasing number of devices, in this paper, we propose a new Reconfigurable intelligent surface (RIS) Assisted simultaneous wireless information and Power transfer (SWIPT) and over-the-aIr computation (AirComp) feDerated learning system which is termed RAPID. Specifically, at each training iteration, an access point (AP) first simultaneously broadcasts global model and transfers wireless energy to the selected devices via the RIS-assisted SWIPT. These devices then use the harvested power to compute and upload their local gradients to the AP for aggregation via the RIS-assisted AirComp. To identify the performance improvement to federated learning, we first analyze and derive the expected convergence rate for the proposed RAPID system taking into account factors of device selection and wireless communication. We then formulate a joint learning-communication optimization problem in terms of device selection, transmit beamforming, power splitting, receive beamforming, and RIS coefficients design. To solve the formulated non-convex problem, we propose a new two-stage algorithm by successively solving the downlink SWIPT and the uplink AirComp sub-problems based on alternating optimization and successive convex approximation techniques. Simulation results are presented to demonstrate that our proposed RAIPD system can significantly improve the convergence and accuracy of federated learning compared with benchmark schemes.

In this paper, we investigate the application of rate-splitting multiple access (RSMA) in a two-tier wireless backhaul heterogeneous network (HetNet), where a macro base station (MBS) simultaneously transmits wireless access signals to multiple macro-cell users (MCUs) and wireless backhaul signals to small base stations (SBSs) by leveraging RSMA. Furthermore, to explore the potential advantage of common streams in RSMA systems, we develop a “Hybrid RSMA” scheme in which the MBS only employs RSMA to encode the backhaul messages while each MCU’s message is directly encoded without rate-splitting. We formulate an optimization problem to maximize the system’s spectral efficiency (SE) by jointly considering transmit precoding and rate allocation at the MBS and SBSs. To solve the formulated non-convex problem, we first propose an iterative centralized algorithm based on successive convex approximation (SCA). Then, we further develop an efficient decentralized algorithm that can be executed in parallel at the MBS and each SBS based on the local channel state information with fewer signaling exchanges. Simulation results show that the application of RSMA can achieve higher SE over conventional non-orthogonal multiple access (NOMA) and space division multiple access (SDMA) under different network loads. Particularly, “Hybrid RSMA” has a greater performance improvement than “RSMA” in the underloaded system.

In this paper, we investigate the application of rate-splitting multiple access (RSMA) in a two-tier wireless backhaul heterogeneous network (HetNet), where a macro base station (MBS) simultaneously transmits wireless access signals to macro-cell users (MCUs) and wireless backhaul signals to small base stations (SBSs) by leveraging RSMA. In order to improve the system spectral efficiency (SE) while guaranteeing the quality-of-service (QoS) of each user, we formulate an optimization problem to maximize the sum SE by jointly considering transmit precoding and rate allocation at the MBS and SBSs. To solve the formulated non-convex problem, we propose an iterative algorithm based on successive convex approximation (SCA). Simulation results show that the application of RSMA can achieve higher SE over conventional non-orthogonal multiple access (NOMA) and space division multiple access (SDMA) under different network loads. In addition, RSMA has better flexibility than other benchmark schemes in meeting the increasing QoS requirements of users.

Mobile edge computing (MEC) has been regarded as a promising technology to liberate the resource-limited users from computation-intensive and latency-sensitive tasks by computation offloading. Furthermore, implementing non-orthogonal multiple access (NOMA) technology in heterogeneous networks (HetNets) has become a trend to improve system throughput and spectrum efficiency. Exploiting these benefits, we investigate the joint task offloading and resource allocation problem for MEC in NOMA-based HetNets. To minimize the energy consumption of all users, we jointly consider task offloading decision, local CPU frequency scheduling, power control, computation resource and subchannel resource allocation. The optimization problem is challenging due to the strong coupling between offloading decision and resource allocation. We thus decouple the problem into two sub-problems of offloading decision and resource allocation, and propose an efficient approach to find the joint solution by solving these two sub-problems iteratively. Simulation results show that the proposed approach can efficiently lower energy consumption of users compared to other benchmark schemes with an acceptable complexity.

Mobile edge computing (MEC) is regarded as a key technology to reduce the network pressure from the computing-intensive and latency-sensitive applications in future wireless networks. Non-orthogonal multiple access (NOMA) can achieve high spectral efficiency by allowing multiple users to reuse the same resources. In this paper, we consider a novel NOMA-based MEC system to improve the energy efficiency during task offloading process. With multiple access points (APs) being deployed, the optimization problem is joint user association and resource allocation while the objective is to minimize the total energy consumption of all users subject to the task execution deadline. We formulate the problem as a many-to-one matching game with externality due to the co-channel interference among users, and then, propose a matching-coalition approach coupled with computing resource allocation and power control. Simulation results show that the proposed approach can efficiently reduce the total energy consumption in comparison to other simplified approaches.