The increasing demand for wireless applications with tight deadlines and intensive computational requirements has exposed the limitations of traditional mobile edge computing (MEC). In such a tight deadline task processing environment, it is inevitable that some tasks cannot be processed within the deadline and are dropped. To tackle this, we formulate an optimization problem that minimizes a weighted sum of the dropout ratio and average delay. To solve the problem, we present an integrated method that combines MEC with cell-free networks. In addition, we propose the Joint optimization of Association, Beamforming, Task Offloading, and Resource Allocation (JAB-TORA), which accounts for realistic wireless conditions. We formulate the problem as a Markov decision process and address it using a multi-agent deep reinforcement learning approach. Simulation results demonstrate the inversely proportional relationship between the dropout ratio and the average delay when adjusting the weights. These weights can be adjusted to meet the required dropout ratio or delay level. Furthermore, we demonstrate that the proposed method achieves a lower dropout ratio and delay than other offloading methods.

We propose a novel integrated sensing and communication (ISAC) framework leveraging spatially separated sub-arrays (SAs) in the upper mid-band. While large-scale arrays in the upper-mid band enable near-field operation, which facilitates joint range-angular estimation in sensing and depth-domain spatial multiplexing in communication, the effective near-field range is constrained by the Rayleigh distance. The proposed SA structure effectively forms a large virtual aperture, enabling near-field-like beam focusing beyond this constraint and enhancing spatial resolution. We develop a joint communication and sensing beamformer to support multiple users and multiple dedicated sensing targets. Additionally, we introduce two methods for joint range-angle estimation using the SA configuration. Numerical results demonstrate that the SA deployment significantly outperforms conventional co-located arrays in both sensing accuracy and communication efficiency, achieving a broader sensing-communication trade-off region. These findings highlight the potential of spatially separated SAs for next-generation ISAC systems, offering improved performance without incurring excessive hardware complexity.

Edge deployment of large language models (LLMs) can reduce latency for interactive services, but mobility introduces service interruptions when an user equipment (UE) hands over between base stations (BSs). To promptly resume decoding, the target-side edge server must recover the UE context state, which can be provisioned either by token forwarding followed by prefill computation or by direct key-value (KV) cache transmission over backhaul. This paper proposes a unified handover (HO) design that jointly selects the prefill length and schedules backhaul KV cache delivery to minimize the worst-user LLM HO delay for multiple UEs. The resulting scheme admits a tractable step-wise solution with explicit feasibility conditions and a constructive rate-scheduling policy. Simulations show that the proposed method consistently outperforms baselines across a wide range of backhaul capacities, prefill speeds, and context sizes, providing practical guidelines for mobility-aware Edge LLM token streaming.

This letter presents an OFDM-based hybrid beam-forming (HBF) design for wideband extremely large-scale MIMO (XL-MIMO) systems in a hybrid near-field/far-field (hybrid-field) environment. As apertures grow in Sub-THz and millimeter-wave systems, the Rayleigh distance increases substantially and near-field (NF) and far-field (FF) components coexist within a single link, where conventional FF-only beamforming suffers severe inter-field interference. We formulate field separation within an oblique-projection framework, which isolates a signal subspace from an interfering one using only basis matrices of the two subspaces, without requiring orthogonality. Stage 1 builds the FF subspace from the NLoS component and extracts the NF subspace as its orthogonal-complement residual; this sequential construction makes the two subspaces exactly orthogonal, so the oblique projectors reduce to orthogonal projectors requiring no inversion or regularization. A single virtual analog matrix is then formed from the separated components—computed once from the frequency-averaged channel, since the analog beamformer is shared across subcarriers—and FRF is extracted via its SVD. Stage 2 designs per-subcarrier baseband matrices FBB[k] via SVD of the effective channel seen through FRF. Simulations show superior spectral efficiency over FF-only and NF-only HBF across the entire SNR range.

Tokens are fundamental processing units of generative AI (GenAI) and large language models (LLMs), and token communication (TC) is essential for enabling remote AI-generate content (AIGC) and wireless LLM applications. Unlike traditional bits, each of which is independently treated, the semantics of each token depends on its surrounding context tokens. This inter-token dependency makes TC vulnerable to outage channels, where the loss of a single token can significantly distort the original message semantics. Motivated by this, this paper focuses on optimizing token packetization to maximize the average token similarity (ATS) between the original and received token messages under outage channels. Due to inter-token dependency, this token grouping problem is combinatorial, with complexity growing exponentially with message length. To address this, we propose a novel framework of semantic packet aggregation with lookahead search (SemPA-Look), built on two core ideas. First, it introduces the residual semantic score (RSS) as a token-level surrogate for the message-level ATS, allowing robust semantic preservation. Second, instead of full search, SemPA-Look applies a lookahead search-inspired algorithm, thereby achieving linear complexity. Experiments on an MS-COCO dataset (text captioned images) demonstrate that SemPA-Look achieves high ATS and LPIPS comparable to exhaustive search, while reducing computational complexity by up to 40 & times; . Compared to other algorithms such as the genetic algorithm (GA), SemPA-Look achieves 10 & times; lower complexity, demonstrating its practicality for remote AIGC and other TC applications.

Unmanned aerial vehicle (UAV)-assisted networks are a promising technology in future wireless communication networks. Numerous applications related to UAV-assisted networks include mobile users (or targets) such as people, selfdriving vehicles, and robots. The location information of mobile devices is an essential element for realizing these applications. Furthermore, for providing services to mobile devices that move over time, real-time multi-target tracking is necessary. To ensure fair service provision to multiple users, unnecessary duplicated support or unsupported users should be avoided. For this reason, we consider multiple-target assignment constraints to fairly provide services. These motivations lead to the design of multiple UAV target assignment and tracking (MUTAT). In this paper, we propose a joint multi-UAV target assignment and tracking scheme to minimize target tracking errors while ensuring multiple target assignment constraints. Our proposed approach, named deep reinforcement learning-based multi-UAV target assignment and tracking (DeepMUTAT), consists of two distinct stages. In the first stage, to reduce computational complexity and ensure multiple target assignment constraints, we adopt a deep reinforcement learning (DRL)-based multi-target assignment for efficient multitarget tracking. In the second stage, to minimize tracking errors without requiring knowledge of the dynamics of targets and to avoid collisions with surrounding obstacles, we propose a DRL-based multi-target tracking approach. Based on extensive simulations, we demonstrate the performance of the proposed scheme compared to baseline schemes in terms of tracking error and tracking success probability. The proposed scheme achieved similar performance to the baseline scheme but had lower computational complexity.

One of representative strategies for hybrid analog-and-digital beamforming (HBF) in massive multiple-input multiple-output (MIMO) systems is the two-stage design, where radio frequency (RF) and baseband (BB) beamformers are optimized sequentially. However, the two-stage approach overlooks the impact of BB beamforming (BF) on RF BF design, resulting in loss of optimality of RF BF. To address this, we propose an iterative two-stage HBF design that considers the influence of BB BF on RF BF design. The BB BF is designed per subcarrier for orthogonal frequency division multiplexing (OFDM) systems and optimized under per-RF chain power constraints. Simulation results verify superiority of the proposed HBF design.

Due to its simplicity and lack of channel state information (CSI) feedback requirements, time division duplexing (TDD) has been the preferred duplexing method in wireless-powered communication networks (WPCNs), while the advantages of frequency-division duplexing (FDD) has remained largely unexplored. Yet, the decision between TDD and FDD goes beyond CSI considerations, as it depends on various system parameters and operational trade-offs not previously considered. In FDD, the transmitter remains active throughout the entire frame duration, enabling more effective utilization of the maximum instantaneous transmit power of the hybrid access point (HAP). In contrast, TDD exploits the full bandwidth (BW), thereby making better use of the feasible maximum power spectral density (PSD). Finally, while the constraint of maximum time-averaged transmit power in FDD closely resembles the effect of the maximum instantaneous transmit power constraint, they have different effects on the operation of the TDD-WPCN. To analyze these effects, we thoroughly investigate both TDD-WPCN and FDD-WPCN and characterize their respective operating regions. Our extensive theoretical and simulation results reveal that selecting between the two schemes involves a complex, multidimensional trade-off, warranting careful consideration in system design. We demonstrate that, under certain assumptions, the throughput of an FDD-WPCN can be substantially greater than that of the same WPCN operating in TDD mode. Furthermore we prove that, under certain conditions, a single-node FDD-WPCN can achieve a two-fold increase in throughput compared to a single-node TDD-WPCN.

Cell-free multiple-input-multiple-output (MIMO) is poised to enable scalable next-generation cellular networks. To this end, it is crucial to optimize the cell-free MIMO link configuration, including user associations, data stream allocation, and beamforming (BF). However, the scalability of link configuration optimization is significantly challenged as signaling and computational costs increase with the number of base stations (BSs) and user equipments (UEs). To address this scalability issue, this paper proposes a distributed multi-agent deep reinforcement learning (MADRL)-based cell-free MIMO link configuration framework that leverages interference approximation to minimize signaling overhead required for channel state information (CSI) exchange. Our proposed framework reduces the solution search space suitable for distributed MADRL, by decomposing the original sum rate maximization problem into BS-specific tasks. Simulation results show that our proposed method achieves scalability, as the sum rate increases with the number of BSs and UEs.

In this paper, we aim to design an ISAC beamforming technique that is robust to angle estimation errors of both target and clutters, in multi-user and single-target MIMO ISAC systems. To this end, the proposed ISAC beamformer is designed to maximize the average signal-to-clutter-plus-noise ratio (SCNR) with respect to the angular distributions of the target and clutters, while ensuring signal-to-interference-plus-noise ratio (SINR) threshold for each communication user. Since calculating the average SCNR is highly complex, we derive a lower bound for the average SCNR and approximated it with simple closed form expression. Additionally, we use Dinkelbach’s transform and SDR (semidefinite relaxation) to resolve the non-convexity of the optimization problem. Numerical simulations demonstrate that the proposed beamformer achieves higher sensing performance in average sense, compared to the estimated angle-based beamformer, which indicates that the proposed beamforming framework is robust to angle estimation errors for both target and clutters.

This paper proposes a novel, unified approach to design both radio frequency (RF) and baseband beamforming (BF) solutions for wideband hybrid multiple-input multiple-output (MIMO) systems using the Riemannian manifold optimization. The proposed two-stage optimization framework effectively addresses various power constraints of the wideband hybrid MIMO systems, including total sum power, per-RF-chain, and per-antenna power constraints. By transforming constrained optimization problems into unconstrained ones on Riemannian submanifolds, we achieve notable improvements in spectral ef- ficiency (SE) and power consumption. In addition, we account for the log-determinant structure of the objective function, which forms an unconstrained solution subspace. This allows for a more efficient retraction process towards the manifold, representing the set of solutions that satisfy the constraints. This unified design approach provides a more accurate and efficient solution to non-convex optimization problems with complex constraints. Simulation results demonstrate the superior performance of the proposed approach over conventional approaches.

Token communication (TC) is poised to play a pivotal role in emerging language-driven applications such as AI-generated content (AIGC) and wireless language models (LLMs). However, token loss caused by channel noise can severely degrade task performance. To address this, in this article, we focus on the problem of semantics-aware packetization and develop a novel algorithm, termed semantic packet aggregation with genetic beam search (SemPA-GBeam), which aims to maximize the average token similarity (ATS) over erasure channels. Inspired from the genetic algorithm (GA) and the beam search algorithm, SemPA-GBeam iteratively optimizes token grouping for packetization within a fixed number of groups (i.e., fixed beam width in beam search) while randomly swapping a fraction of tokens (i.e., mutation in GA). Experiments on the MS-COCO dataset demonstrate that SemPA-GBeam achieves ATS and LPIPS scores comparable to exhaustive search while reducing complexity by more than 20x.

Text-based communication is expected to be prevalent in 6G applications such as wireless AI-generated content (AIGC). Motivated by this, this paper addresses the challenges of transmitting text prompts over erasure channels for a text-to-image AIGC task by developing the semantic segmentation and repeated transmission (SMART) algorithm. SMART groups words in text prompts into packets, prioritizing the task-specific significance of semantics within these packets, and optimizes the number of repeated transmissions. Simulation results show that SMART achieves higher similarities in received texts and generated images compared to a character-level packetization baseline, while reducing computing latency by orders of magnitude compared to an exhaustive search baseline.

In this paper, we propose deep learning-based channel estimation and pilot reduction for mmWave point-to-point multi-input multi-output systems. The proposed scheme consists of a two-step approach where the first step is applying a denoising autoencoder for channel estimation. With the denoising characteristic of autoencoder, sparse channel estimation can be conducted although the orthogonality of pilot sequences is not guaranteed due to shorter pilots. The second step is exploiting the temporal correlation of the channel, using the previous estimate to extract information for the current estimate. Through simulation, the proposed scheme shows superior performance with reduced pilots.

Integrated localization and communication (ILC) will be a key enabler for providing accurate location information and high data rate in next generation networks. This paper proposes a transmission frame structure and a soft information (SI)-based localization algorithm for position-assisted communications. The proposed ILC achieves improved localization accuracy and enhanced communication rate simultaneously by accounting for the statistical characteristics of the wireless environment. Results in 3rd Generation Partnership Project (3GPP) industrial scenarios show that the SI-based localization algorithm can achieve decimeter-level accuracy. Moreover, the position-assisted communication enhances the achievable rate, especially in scenarios with high mobility.

In order to provide flexible radio access technologies, the fifth generation (5 G) New Radio (NR) employs various choices of orthogonal frequency division multiplexing (OFDM) numerologies. Nevertheless, a comprehensive signal model of massive multiple-input multiple-output (MIMO) uplink (UL) systems with multiple numerologies remains unclear. Most previous studies have solely considered the signals of two numerologies within only one least common multiplier (LCM) symbol duration, which is the time duration for synchronizing distinct numerologies, and aimed to address inter-numerology interference (inter-NI). However, the inter-NI from OFDM symbols in the previous LCM symbol duration should also be considered in UL systems. Accordingly, this paper provides a generalized baseband signal model with an arbitrary number of numerologies that encompasses all the inter-NI. In addition, we conduct an analysis of signal power and derive a tractable form of the signal-to-interference-plus-noise ratio (SINR) expression. Based on the analysis, we propose linear combining designs with respect to the zero-forcing (ZF) and minimum mean square error (MMSE) criteria for multi-numerology systems. Simulation results verify the superiority of the proposed linear combiners. In particular, we demonstrate that the proposed ZF combiner is able to perfectly eliminate all the inter-NI. Furthermore, the simulation results verify that the derived lower bound of the achievable rate with the proposed ZF combiner is very tight.

Variable Coding and Modulation (VCM), one of the Orthogonal Multiple Access (OMA) schemes and also known as channel adaptive Time-sharing Division Multiplexing (TDM) in satellite broadcasting and communication systems, has been widely utilized to mitigate heavy rain fading in Ku/Ka-band to enhance link availability in DVB-S2x (Digital Video Broadcasting -Satellite 2nd generation eXtension) standard. For next-generation satellite broadcasting and communication, we exploit Layer Division Multiplexing (LDM) technology, which is also referred to as Non-Orthogonal Multiple Access (NOMA). We conduct a performance assessment of VCM in terms of the total achievable data rate under Additive White Gaussian Noise (AWGN) channel and nonlinear satellite High Power Amplifier (HPA) impairments. In addition, we consider realistic Radio Frequency (RF) inaccuracies characterized by timing and carrier offsets. Through the identification of the performance impacts, we propose a robust carrier phase synchronization scheme to mitigate phase noise impairment. Numerical results demonstrate that our proposed scheme can enhance Packet Error Rate (PER) performance compared to the conventional one in a phase noise environment.

In this article, we study a wirelessly powered communication network (WPCN) composed of several wireless devices (WDs) which rely on an energy access point (EAP) for their supply of energy which is used to transfer data to a data access point (DAP). While the dominant focus of the literature in this area has been on frame-based data rate optimization, it has been known that this approach is greedy and only optimal in additive white Gaussian noise (AWGN) channels. Hence, in this work, we propose an online algorithm that, based on the current state of the batteries and channels, adaptively calculates the transmission time and power allocations to maximize the long-term performance of this system in fading channels. In order to accomplish such a goal, we employ the Twin delayed DDPG (TD3) approach, which is a deep reinforcement learning (DRL) technique designed for continuous state and action spaces. Additionally, we use the multitask learning (MTL) techniques to train a DRL agent that creates a composite policy capable of dynamically optimizing not just one specific WPCN configuration, but a continuum of WPCN problems, which we call a composite environment. Each of these environments has different parameters, such as WDs' distances, to the DAP and EAP, batteries' size, rectifying antenna (rectenna) energy harvesting (EH) model linearity, as well as different uplink (UL) and downlink (DL) channel specular components. Simulation results confirm that the TD3 algorithm can easily learn to achieve a much higher common throughput compared to the traditional optimization schemes.

Cloud Radio Access Network (C-RAN) refers to the virtualization of base station functionalities by means of cloud computing. With centralized processing at the cloud, the amount of hardware and infrastructure in a network can be reduced and it allows for operators to save expenses needed to deploy and maintain the network. Along with high spectral and energy efficiency, C-RAN becomes a key technology evolving to 5G network and it is also expected to play a critical role in the next generation of wireless network. In this paper, theoretical expressions are derived for throughput performance at Medium Access Control (MAC) layer with and without Cloud Radio Random-Access (CRRA). We first develop a CRRA protocol by accounting for the Multi-Packet Reception (MPR) capabilities afforded thanks to the deployment of C-RAN. We then analyze the throughput performance based on error exponent analysis. Numerical results show the performance advantages of C-RAN and verify theoretical expressions.

To alleviate severe attenuation of millimeter-wave (mmWave) channel, massive multi-input multi-output (MIMO) in which a large number of antennas are utilized in base station (BS) is a promising solution. However, due to a large number of antennas and limited coherence time of mmWave mobile channel, acquisition of a precise channel state information (CSI) requires a large channel estimation overhead, especially for frequency division duplex (FDD) systems. In addition, due to limited saturation level of mmWave power amplifiers, a flexible pilot length adaptation is required for mmWave communication systems. Likewise, because of a large number of BS antennas, it is difficult to feed back all CSI with high accuracy in FDD massive MIMO systems. There exists a trade-off between feedback overhead and the accuracy of recovered CSI. To resolve these issues, in this paper, we propose a deep learning-based joint optimization of closed-loop FDD massive MIMO systems. Unlike previous deep learning-based approach in which only beamforming (BF) from observation of CSI is composed with neural network, we formulate the entire process of generating BF matrix as a functional optimization problem. We compose a deep learning-based FDD massive MIMO system where the functional blocks including pilot length adaptation, CSI compression, and BF are replaced with neural networks. The neural networks are connected and trained as a single network toward a direction which maximizes network utility. During the training process, each functional block learns its best strategy to maximizes network utility. Through simulation, we confirm that each network learned strategies to maximize network utility by itself.