Integrated sensing and communication (ISAC) can enhance artificial intelligence-generated content (AIGC) networks by providing efficient sensing and transmission. Existing AIGC services usually assume that the accuracy of the generated content can be ensured, given accurate input data (e.g., pose image) and command (i.e., prompt), thus only the content generation quality (CGQ) is concerned. However, it is not applicable in ISAC-based AIGC networks, where content generation is based on inaccurate sensed data. Moreover, the AIGC model itself introduces generation errors, which depend on the number of generating steps (i.e., computing resources). Thus, to assess the quality of experience (QoE) of ISAC-based AIGC services, this paper proposes a content accuracy and quality aware service assessment metric (CAQA). Since allocating more resources to sensing and generating improves content accuracy but may reduce communication quality, and vice versa, this sensing-generating (computing)-communication three-dimensional resource tradeoff must be optimized to maximize the average CAQA (AvgCAQA) across all users with AIGC (CAQA-AIGC). This problem is NP-hard, with a large solution space that grows exponentially with the number of users. To solve the CAQA-AIGC problem with low complexity, a standard linear programming (LP) guided deep reinforcement learning (DRL) algorithm with an action filter (LPDRL-F) is proposed. Through the LP-guided approach and the action filter, LPDRL-F can transform the original three-dimensional solution space to two dimensions, reducing complexity while improving the learning performance of DRL. Simulations show that compared to existing DRL and generative diffusion model (GDM) algorithms without LP, LPDRL-F converges faster and finds better resource allocation solutions, thus improving AvgCAQA by more than 10%. With LPDRL-F, CAQA-AIGC can achieve an improvement in AvgCAQA of more than 50% compared to existing schemes focusing solely on CGQ.

Mobile Edge Computing (MEC) is promising to enable low delay services with which users can offload computing intensive and delay sensitive tasks to the edge. Considering a multi-cell MEC (MC-MEC) network without sufficient resources to serve all users, user selection and non-orthogonal multiple access (NOMA) should be introduced. Then, to maximize the delay-aware average user service satisfaction degree (DA-AveUSD), user selection and resource allocation are jointly optimized (DA-JUSRA), which is modeled as a mixed integer nonlinear programming (MINLP) problem and proven to be NP-hard. To solve this problem, it is decomposed into two independent subproblems, i.e., the power allocation (PA) problem and the user selection, subchannel scheduling and computing resource allocation (USC) problem. Next, a convex evolutionary alternating optimization (CEAO) algorithm is proposed, which alternately applies the convex optimization method and the Karush-Kuhn-Tucker (KKT)-embedding enhanced elite genetic algorithm KKT-embedding E2GA to solve the PA and the USC problem, respectively. Simulations show that compared to the optimal exhaustive search algorithm, the proposed CEAO algorithm converges rapidly within a few iterations, with a gap in DA-AveUSD of less than 1% to the optimum performance. Next, compared to existing user selection schemes, DA-JUSRA with CEAO can enhance DA-AveUSD by more than 50% and yield a higher optimal load.

In integrated sensing and communication (ISAC) enabled unmanned aerial vehicle (UAV) systems, based on sensed information such as user positions, UAV scheduling could be optimized to enhance the communication performance. However, sensing errors are inevitable, leading to a performance degradation. This paper proposes a sensing-error-aware (SEA) multi-UAV scheduling scheme (SEA-scheduling). First, the impact of the sensing errors on communication performance is analyzed, and a SEA communication rate is derived. Then, targeting to maximize this SEA rate, multi-UAV collaborative scheduling is jointly optimized with sensing resource allocation. The problem is solved by decomposing into two subproblems, i.e., a joint UAV position schedule, user association and bandwidth allocation optimization subproblem (PUB) and a sensing resource optimization subproblem (SRO), which can be solved iteratively. A generative diffusion(GD)-driven multi-agent reinforcement learning (GD-MADRL) algorithm is proposed to solve PUB, and a classical simulated annealing (SA) algorithm is adopted to solve SRO. The main idea of GD-MADRL is to introduce the GD model in MADRL to generate training data with sensing errors, enhancing the robustness of generated UAV scheduling strategies. Simulation results demonstrate that when there are sensing errors, the proposed SEA-scheduling scheme improves the communication rate by up to 30% compared to existing sensing-error-unaware schemes.

Frequency shift keying (FSK) modulation can avoid the direct path interference problem in backscatter communication (BackCom). However, due to the limited number of reflection impedances of the tag, unwanted high-order harmonics are introduced, thereby degrading bit error rate (BER) performance. In this work, we investigate the harmonic elimination strategy for FSK-based BackCom. In particular, we first propose a novel multi-stage incident signal and the corresponding reflection strategy, which allows the tag to modulate the incident signal through absolute value operations, effectively suppressing harmonic generation. Then, we develop a USRP-based platform to experimentally verify the feasibility of our harmonic elimination strategy. Finally, the results show that our design can achieve a 12 dB attenuation in the third harmonic and up to 34 dB attenuation in high-order harmonics compared to conventional FSK-based BackCom without harmonic elimination, which consequently leads to improved BER performance.

With the growing diversity and density of mobile nodes, indoor wireless channels are becoming increasingly complex. Existing channel modeling methods struggle to balance accuracy and adaptability, which calls for low-complexity models capable of capturing dynamic indoor environments efficiently. This paper proposes a novel propagation graph (PG) framework that models the channel effects of dynamic objects indoors by designing a scattering-aware point cloud neural network (SPointNet). The proposed PG framework explicitly incorporates reflection, transmission, and diffuse scattering into channel modeling, and employs a physics-aware scatterer discretization and classification strategy, which reduces the complexity of the conducted graph. Then, SPointNet enables fast estimation of scattering coefficients, allowing the model to bypass exhaustive analysis of material properties. Finally, we conduct channel measurements in real indoor environments to validate the proposed approach. Experimental results show that the proposed model accurately models channel responses while significantly reducing modeling time compared to traditional PG-based methods.

The deployment of object detection tasks on embedded or mobile platforms has become increasingly prevalent, driven by the heightened demand across various scenarios. However, for object recognition tasks such as gesture recognition, the use of overly complex network models presents a formidable obstacle in achieving real-time detection tasks, and the majority of lightweight convolution menthods based on depth-separated convolution lack accuracy. In this paper, we propose a lightweight and highly accurate convolutional neural network for gesture recognition on terminal device (GRTD-Net), specially designed for devices with scarce computing power and tight hardware resources. In GRTD-Net, we proposed the convolution method R2SGConv that masterfully harmonizes model size and accuracy, elegantly achieving a delicate balance between efficiency and lightweight design. Moreover, we propose a neck network paradigm with good feature fusion capability to compensate for the accuracy degradation due to the use of lightweight convolutional modules in neck networks. Experimental results show that the proposed GRTD-Net model improves the mAP0.5and mAP0.95by 0.8% and 2.2%, reduces model parameters by 35.2%, increases FPS by 20.7%, and reduces the inference latency by 2.9 ms, compared with the popular YOLOv5 algorithm on the dataset Gesture. We successfully deployed GRTD-Net on ARM devices and proved its practicality in constrained environments.

Orthogonal time-frequency space (OTFS) modulation has emerged as a promising solution for reliable communication in high-mobility scenarios, addressing the limitations of traditional orthogonal frequency-division multiplexing (OFDM) systems. However, existing OTFS detection methods, including linear, nonlinear, and AI-based detectors, struggle with either high computational complexity or suboptimal performance. To overcome these limitations, we propose a low-complexity graph attention network-based OTFS detector (GAT-OTFS) tailored for the reduced cyclic prefix (RCP) OTFS scenario. Our approach leverages the sparsity of the DD domain equivalent channel to reduce complexity by precisely constructing a sparse graph for GNN-based detection. By introducing an attention mechanism, the GAT-OTFS effectively captures the correlation between different channel paths, enhancing detection accuracy. Simulation results demonstrate that the GAT-OTFS offers substantial performance improvements over existing detectors, with reduced complexity, making it a viable solution for future high-mobility communication systems.

Backscatter communication and harvest-then-transmit (HTT) communication are regarded as promising technologies for enabling green Internet of Things (IoT). The current works on the joint use of backscatter communication and HTT are limited in single cell scenarios with the fixed backscatter-then-HTT transmission structure. In this work, we propose a transmission scheme with flexible mode selection for the hybrid backscatter-HTT multi-cell system to achieve much improved communication performance, and then study the joint design for such a system. Specifically, by utilizing multi-antenna technology and enabling the flexible mode selecting between backscatter and HTT, a novel transmission scheme is developed. With the aim to maximize the sum rate of the considered system, we formulate a joint optimization problem for the base station transmission beamforming (TB), the transmission mode (TM), and the transmit power (TP) of the hybrid backscatter-HTT devices. To address the formulated non-convex problem, we propose a block coordinate descent-based algorithm, namely J3TO, to jointly optimize TB, TM, and TP, by decoupling the original problem into three sub-problems. Therein, the weighted minimum mean square error approach, matching theory, and the fractional programming technique are leveraged to deal with the sub-problems efficiently. Simulation results show that the proposed algorithm flexibly integrates the merits of backscatter and HTT technologies, achieving superior performance across various scenarios, compared with the benchmark schemes, e.g., backscatter-only SDMA, HTT-only SDMA, and backscatter-HTT TDMA.

Task-oriented communications with semantic encoders are promising to enhance the communication efficiency, by selecting and transmitting valuable data according to task requirements/queries. However, existing semantic encoders lack the capability to track the changing in queries, leading to biased data selection. This paper proposes a query-aware semantic encoder, i.e., Query-Data Cross (QDC) encoder for task-oriented communications. By consistently focusing on data features that are most relevant to the current query at the transmitter, QDC can adapt to changing queries. Based on the dynamic semantic relevance obtained by QDC, a relevance-based data selection and bandwidth allocation optimization (RDSBA) problem is formulated, considering a multi-device task-oriented communication system, where devices should transmit valuable data with high relevance to the queries broadcasted by the base station (BS). RDSBA aims to maximize the data profit of all devices, which is defined as the difference between the relevance of data selected for the BS and the cost of obtaining the data. Then, a DRL-based data selection and bandwidth allocation (DRL-DB) algorithm is proposed to solve the NP-hard optimization problem. Simulation results demonstrate that QDC can smartly track the changing in queries and achieve an accuracy of at least 85% in relevance evaluation, more than 8% higher than existing schemes. Based on the relevance provided by QDC, the proposed RDSBA scheme with DRL-DB can increase the data profit by at least 18%, comparing to existing schemes.

The deployment of mobile edge computing services in LEO satellite networks achieves seamless coverage of computing services. However, the time-varying wireless channel conditions between satellite–terrestrial channels and the random arrival characteristics of ground users’ (GUs) tasks bring new challenges for managing the LEO satellite’s communication and computing resources. Facing these challenges, a stochastic computation offloading problem of joint optimizing communication and computing resources allocation and computation offloading decisions is formulated for minimizing the long-term average total power cost of the GUs and the LEO satellite, with the constraint of long-term task queue stability. However, the computing resource allocation and the computation offloading decisions are coupled within different slots, thus making it challenging to address this problem. To this end, we first employ the Lyapunov optimization to decouple the long-term stochastic computation offloading problem into the deterministic subproblem in each slot. Then, an online algorithm combining deep reinforcement learning and conventional optimization algorithms is proposed to solve these subproblems. Simulation results show that the proposed algorithm can achieve the superior performance while ensuring the stability of all task queues in LEO satellite networks.

Recent years have witnessed an increasing interest in leveraging reconfigurable intelligent surfaces (RISs) to enhance the capabilities of integrated sensing and communication (ISAC) systems. RISs are advantageous in improving detection and communication performance, especially in challenging environments characterized by nonLine of Sight (NLOS) conditions and dense urban settings. In this article, a hybrid RIS, comprising passive reflecting elements and active sensors, and multiple fully passive RISs are deployed to enhance an ISAC system, where the direct paths between the base station (BS) and users/targets are blocked. The signal sent from the BS and reflected by RISs is received by the communication user, and simultaneously scattered by the target toward the sensors of the hybrid RIS. A joint optimization of the transmit covariance matrix at the BS and phase-shifting matrices at RISs is formulated, which considers the tradeoff between the communication and sensing performance. The optimization is based on the derived closed-form communication achievable rate by leveraging the free probability theory and positioning error bound (PEB) via the Cramér-Rao lower bound (CRLB) analysis. The block coordinate descent (BCD) algorithm is utilized to tackle the nonconvex problem, where the transmit covariance matrix and phase-shifting matrices are optimized iteratively. Therein, the Riemannian gradient descent algorithm is exploited for optimizing the phase-shifting matrices. Numerical results verify the effectiveness of the proposed algorithm, and both communication and sensing performance gains increase with the number of RIS panels and RIS elements.

The low earth orbit (LEO) satellites have received extensive attention as an essential supplement to the terrestrial network for supporting global Internet of Things (IoT) services. Considering the rapid growth of IoT devices and the significant satellite-to-ground latency, proposing low-latency, low-overhead access protocols for LEO satellite IoT systems is challenging. In this article, we propose a multibeam random access (RA) framework and deploy the deep reinforcement learning (DRL) algorithm to control the nonorthogonal multiple access (NOMA) aided RA strategy. First, we divide the satellite coverage region into multiple beams and assume that the adjacent beams share parts of regions. Hence, the devices in the sharing region are allowed to transmit packets in two periods allocated for the two beams. Then, packets in multiple beams can be decoded jointly by an interslot successive interference cancelation (SIC) decoder. In addition, we consider the heterogeneity among devices and assign different power levels for heterogeneous types of devices, which enables power-domain NOMA and the intraslot SIC decoder in this system to mitigate the collision resolution. To maximize the average throughput, the deep deterministic policy gradient (DDPG) algorithm is adopted to achieve an online decision to optimize the RA protocol where the packet repetition strategies of devices are adjusted dynamically. The simulation results show that the proposed scheme outperforms the traditional benchmark schemes with significant throughput gain.

In the uplink machine-type communication (MTC) system, a combination of grant-free transmission and non-orthogonal multiple access (NOMA) emerges to reduce the control overhead and transmission latency. In the grant-free scenario, the base station needs to identify the active devices and estimate the channel state information before the data detection. However, due to the lack of a scheduling process, the user activity detection (UAD) and channel estimation (CE) are both challenging, especially when short non-orthogonal preambles are adopted. In this paper, by exploiting the framework of the compressive sensing-based algorithm, we propose a novel deep learning architecture, namely UAD and CE Neural Network (UAD-CE-NN), to effectively solve the joint UAD and CE problem for grant-free NOMA. In the proposed scheme, the user activity and channel state information hidden in the received data signals are also exploited to aid the preamble for higher detection accuracy. Specifically, UAD-CE-NN is composed of two stages: we first build a preamble detection neural network for a tentative UAD-CE; a data detection neural network is then deployed to exploit the data signals. Compared with the conventional schemes, the proposed scheme obtains much higher accuracy for both the UAD and CE, especially when short preamble sequences are employed.

Non-orthogonal multiple access (NOMA) promises to fulfill the fast-growing connectivities in future Internet of Things (IoT) using abundant multiple-access signatures. While explicitly notifying the utilized NOMA signatures causes large signaling cost, blind signature classification naturally becomes a low-cost option. To accomplish signature classification for NOMA, we study both likelihood- and feature-based methods. A likelihood-based method is firstly proposed and showed to be optimal in the asymptotic limit of the observations, despite high computational complexity. While feature-based classification methods promise low complexity, efficient features are non-trivial to be manually designed. To this end, we resort to artificial intelligence (AI) for deep learning-based automatic feature extraction. Specifically, our proposed deep neural network for signature classification, namely DeepClassifier, establishes on the insights gained from the likelihood-based method, which contains two stages to respectively deal with a single observation and aggregate the classification results of an observation sequence. The first stage utilizes an iterative structure where each layer employs a memory-extended network to explicitly exploit the knowledge of signature pool. The second stage incorporates the straight-through channels within a deep recurrent structure to avoid information loss of previous observations. Experiments show that DeepClassifier approaches the optimal likelihood-based method with a reduction of 90% complexity.

Grant-free Non-Orthogonal Multiple Access (NOMA) techniques are able to reduce the signaling overhead and the transmission latency in multi-user communications system. However, most of the existing code-domain grant-free NOMA schemes reuse the spreading signatures designed for the grant-based scenarios. Considering the sparsity and randomness nature of user activities in the uplink transmissions, we propose a deep learning-based signature design, where the non-equal user activation probabilities are exploited to optimize the code-domain NOMA signature. In addition, the conventional grant-free NOMA signatures are not specifically designed over finite Galois field, which hinders the implementation of the encoder/decoder using practical hardware. To address these challenges, we utilize the quantized deep learning framework for the NOMA signature training, which jointly optimizes the sequence generation and the quantization. The numerical results reveal that the obtained signatures outperform the conventional ones especially when the users has unequal activation probabilities.

Non-orthogonal multiple access (NOMA) will provide massive connectivity for future Internet of Things. However, the intrinsic non-orthogonality in NOMA makes it non-trivial to approach the performance limit with only conventional communication-theoretic tools. In this paper, we resort to deep multi-task learning for end-to-end optimization of NOMA, by regarding the overlapped transmissions as multiple distinctive but correlated learning tasks. First of all, we establish a unified multi-task deep neural network (DNN) framework for NOMA, namely DeepNOMA, which consists of a channel module, a multiple access signature mapping module, namely DeepMAS, and a multi-user detection module, namely DeepMUD. DeepMAS and DeepMUD are automatically trained in a data-driven fashion, and a multi-task balancing technique is then proposed to guarantee fairness among tasks as well as to avoid local optima. To further exploit the benefits of communication-domain expertise, we introduce constellation shape prior and inter-task interference cancellation structure into DeepMAS and DeepMUD, respectively. These sophisticated designs help to reduce the implementation complexity without sacrificing DNN's universal function approximation property, which makes DeepNOMA a universal transceiver optimization approach. Detailed experiments and link-level simulations show that higher transmission accuracy and lower computational complexity can be simultaneously achieved by DeepNOMA under various channel models, compared with state-of-the-art.

Tactile Internet of Things (IoT) requires ultraresponsive and ultrareliable connections for massive IoT devices. As a promising enabler of tactile IoT, grant-free nonorthogonal multiple access (NOMA) exploits the joint benefit of grant-free access and nonorthogonal transmissions to achieve low latency massive access. However, it suffers from the reduced reliability caused by random interference. Hence, we formulate a variational optimization problem to improve the reliability of grant-free NOMA. Due to the intractability of this problem, we resort to deep learning by parameterizing the intractable variational function with a specially designed deep neural network, which incorporates random user activation and symbol spreading. The network is trained according to a novel multiloss function where a confidence penalty based on the user activation probability is considered. The spreading signatures are automatically generated while training, which matches the highly automatic applications in tactile IoT. The significant reliability gain of our scheme is validated by simulations.

Grant‐free transmission, which simplifies the signaling procedure via uplink instant transmission, has been recognized as a promising multiple access protocol to address the massive connectivity and low latency requirements for future machine type communications. The major drawback of grant‐free transmission is that the contaminations among uncoordinated transmissions can reduce the data throughput and deteriorate the outage performance. In this paper, we propose a rate‐adaptive multiple access (RAMA) scheme to tackle the collision problems caused by the grant‐free transmission. Different from the conventional grant‐free (conv‐GF) scheme which transmits a single signal layer, RAMA transmits the signals with a multilayered structure, where different layers exhibit unequal protection property. At the receiver, the intra‐ and interuser successive interference cancellation (SIC) receiving algorithm is employed to detect multiple data streams. In RAMA, the users can achieve rate adaptation without the prior knowledge of the channel conditions, since the layers with high protection property can be successfully recovered when the interference is severe, while other layers can take advantage of the channel when the interference is less significant. Besides, RAMA also facilitates the SIC receiving since the multiple layers in the transmission signals can provide more opportunities for interference cancellation. To evaluate the system performance, we analyze the exact expressions of the throughout and the outage probability of both conv‐GF and RAMA. Finally, theoretical analysis and simulation results validate that the proposed RAMA scheme can simultaneously achieve higher average throughput and lower outage performance than conv‐GF. Meanwhile, RAMA shows its robustness with large user activation probability, where the collisions among users are severe.

The capacity bound of the Gaussian interference channel (IC) has received extensive research interests in recent years. Since the IC model consists of multiple transmitters and multiple receivers, its exact capacity region is generally unknown. One well‐known capacity achieving method in IC is Han‐Kobayashi (H‐K) scheme, which applies two‐layer rate‐splitting (RS) and simultaneous decoding (SD) as the pivotal techniques and is proven to achieve the IC capacity region within 1 bit. However, the computational complexity of SD grows exponentially with the number of independent signal layers, which is not affordable in practice. To this end, we propose a scheme which employs multi‐layer RS at the transmitters and successive simple decoding (SSD) at the receivers in the two‐transmitter and two‐receiver IC model and then study the achievable sum capacity of this scheme. Compared with the complicated SD, SSD regards interference as noise and thus has linear complexity. We first analyze the asymptotic achievable sum capacity of IC with equal‐power multi‐layer RS and SSD, where the number of layers approaches to infinity. Specifically, we derive the closed‐form expression of the achievable sum capacity of the proposed scheme in symmetric IC, where the proposed scheme only suffers from a little capacity loss compared with SD. We then present the achievable sum capacity with finite‐layer RS and SSD. We also derive the sufficient conditions where employing finite‐layer RS may even achieve larger sum capacity than that with infinite‐layer RS. Finally, numerical simulations are proposed to validate that multi‐layer RS and SSD are not generally weaker than SD with respect to the achievable sum capacity, at least for some certain channel gain conditions of IC.

Massive Machine Type Communication (mMTC) is one key usage scenario in future 5G. To fulfill the massive connection requirement in mMTC, and reduce the signaling overhead, a novel grant free Random Non-Orthogonal Multiple Access (RNOMA) scheme is proposed in this paper, inspired by the concept of conventional NOMA. In RNOMA, one physical resource area is reserved and divided into orthogonal resource units (RU), where each user is allowed to persuade-randomly transmit the same packet on each RU according to optimized probability. At the receiver, inter- and intra-RU Successive Interference Cancelation (SIC) is applied for multi-user receiving. A frame structure is also designed to enable RNOMA, where signaling overhead on informing each UE with assigned resources is reduced. Simulation results show, the proposed method is robust and is able to achieve significantly better performance than novel random access method, i.e. irregular repetition slotted aloha, where its access degree distribution is optimized for binary erasure channel.