Federated semi-supervised learning (FSSL) is primarily challenged by two factors: the scarcity of labeled data across clients and the non-independent and identically distribution (non-IID) nature of data among clients. In this paper, we propose a novel approach, diffusion model-based data synthesis aided FSSL (DDSA-FSSL), which utilizes a diffusion model (DM) to generate synthetic data, bridging the gap between heterogeneous local data distributions and the global data distribution. In DDSA-FSSL, clients address the challenge of the scarcity of labeled data by employing a federated learning-trained classifier to perform pseudo labeling for unlabeled data. The DM is then collaboratively trained using both labeled and precision-optimized pseudo-labeled data, enabling clients to generate synthetic samples for classes that are absent in their labeled datasets. This process allows clients to generate more comprehensive synthetic datasets aligned with the global distribution. Extensive experiments conducted on multiple datasets and varying non-IID distributions demonstrate the effectiveness of DDSA-FSSL, e.g., it improves accuracy from 38.46% to 52.14% on CIFAR-10 datasets with 10% labeled data.

Large Language Models (LLMs) have achieved significant success in various natural language processing tasks, but how wireless networks can support LLMs has not been extensively studied. In this paper, we propose a wireless distributed LLMs paradigm based on Mixture of Experts (MoE), named WDMoE, through server-device collaboration at the wireless network edge. Specifically, we decompose the MoE layer in LLMs by deploying the gating network and the preceding neural network layer at the edge server of the base station (BS), while distributing the expert networks across the mobile devices. This arrangement leverages the parallel capabilities of expert networks on distributed devices. Moreover, to overcome the instability of wireless communications, we design an expert selection policy by taking into account both the performance of the model and the end-to-end latency, which includes both transmission delay and inference delay. Evaluations conducted across various LLMs and multiple datasets demonstrate that WDMoE not only outperforms existing models, such as Llama 2 with 70 billion parameters, but also significantly reduces end-to-end latency.

The utilization of unmanned aerial vehicles (UAVs) as base stations is an essential case in air-to-ground communication. To balance the equipment cost and performance of directional antennas, the ground terminal can be equipped with a servo phased antenna, which requires the careful design of the access mechanisms. In this paper, we consider the application in which the UAV uses an omnidirectional antenna while the ground terminals use a servo-phased array. We first proposed two access approaches, i.e., stop-and-scan and continuous scan schemes. Then, we established theoretical models for the two schemes and analyzed the performance and the factors impacting it. After that, we conducted a Monte Carlo simulation and compared the performance with numerical results. Results show that the access delay in the stop-and-scan scheme is inversely proportional to the beam width, servo acceleration, and the number of access time slots. The access delay of the continuous scan scheme is significantly lower than the other. However, in the continuous scan scheme, the beam width and the servo acceleration should satisfy a specific constraint to avoid access failure.

Deep neural networks (DNNs) have shown remarkable performance in the super-resolution (SR) task, which can upscale low-resolution images to satisfy application demands on image quality. However, the high computational intensity of DNN models poses a challenge to executing SR tasks on resource-constrained edge platforms. To leverage heterogeneous computational resources (e.g., CPU, GPU, and NPU) to speed up image reconstruction through concurrent inference, we propose a novel framework, called ESHP, for Efficient Super-resolution on edge platforms with Heterogeneous Processors. Our proposed ESHP framework boasts several advantageous characteristics: 1) it substantially speeds up SR processing over the existing approaches by leveraging all available heterogeneous hardware; 2) it uses deep reinforcement learning (DRL) to enable adaptive and optimal scheduling based on runtime states; 3) it strikes a balance between SR performance and computational cost during inference; and 4) it does not modify the original architecture of given SR model. We have conducted extensive experiments on typical edge platforms with popular SR models and resolution datasets of different scales, which verify the effectiveness and the versatility of our ESHP against other commonly-used baselines.

Diffusion models (DM) can gradually learn to remove noise, which have been widely used in artificial intelligence generated content (AIGC) in recent years. The property of DM for eliminating noise leads us to wonder whether DM can be applied to wireless communications to help the receiver mitigate the channel noise. To address this, we propose channel denoising diffusion models (CDDM) for semantic communications over wireless channels in this paper. CDDM can be applied as a new physical layer module after the channel equalization to learn the distribution of the channel input signal, and then utilizes this learned knowledge to remove the channel noise. We derive corresponding training and sampling algorithms of CDDM according to the forward diffusion process specially designed to adapt the channel models and theoretically prove that the well-trained CDDM can effectively reduce the conditional entropy of the received signal under small sampling steps. Moreover, we apply CDDM to a semantic communications system based on joint source-channel coding (JSCC) for image transmission and design a three-stage training algorithm for combining them. Extensive experimental results demonstrate that CDDM can further reduce the mean square error (MSE) after minimum mean square error (MMSE) equalizer, and the joint CDDM and JSCC system achieves better performance than the JSCC system, the traditional JPEG2000 with low-density parity-check (LDPC) code approach and other benchmarks in diverse scenarios.

Diffusion models (DM) can gradually learn to re-move noise, which have been widely used in artificial intelligence generated content (AIGC) in recent years. The property of DM for removing noise leads us to wonder whether DM can be applied to wireless communications to help the receiver eliminate the channel noise. To address this, we propose channel denoising diffusion models (CDDM) for wireless communications in this paper. CDDM can be applied as a new physical layer module after the channel equalization to learn the distribution of the channel input signal, and then utilizes this learned knowledge to remove the channel noise. We design corresponding training and sampling algorithms for the forward diffusion process and the reverse sampling process of CDDM. Moreover, we apply CDDM to a semantic communications system based on joint source-channel coding (JSCC). Experimental results demonstrate that CDDM can further reduce the mean square error (MSE) after minimum mean square error (MMSE) equalizer, and the joint CDDM and JSCC system achieves better performance than the JSCC system and the traditional JPEG2000 with low-density parity-check (LDPC) code approach.

The broadcast scheme is a basic and important data dissemination mechanism in FANET. In this paper, we propose an efficient neighbor quality based mMultil- Ppoint Rrelay (MPR) broadcast scheme based on local neighbor status. The local neighbor status includes two aspects, i.e., the physical link status and the congestion status. The physical link status is indicated by the expected successful packet delivery ratio and link expiration time calculated from the Received Signal Strength Indication (RSSI). The congestion status is determined by the queue length of each neighboring nodes. Then an efficient MPR set can be selected from one-hop neighbors by local neighbor status. Each node uses its MPR set to rebroadcast messages. We use RSSI to calculate the packet successful expectation and link expiration time, and MAC layer queue length to indicate how busy the node is. Simulation result shows that our method performs better in terms of end-to-end delay and packet delivery rate than probability methods and other deterministic methods.

Industrial Wireless Sensor Networks (IWSNs) are mainly used for critical monitoring and control applications in industrial automation systems. The applications require realtime data delivery within strict delay constraints. Exceeding the required delay bound for emergency traffic could result in economic losses or even safety incidents. In this paper, we propose a Priority-enhanced slot Allocation Medium Access Control protocol (PriAlloc-MAC), which can guarantee the transmission of different traffic categories of IWSNs in given time constraints. In this protocol, the nodes with emergency traffic are allowed to preempt slots to reduce the channel access delay. Multiple nodes with emergency traffic access the channel without conflicts according to their access sequence numbers (ASNs). In addition, the protocol utilizes a short time slot design and allocates a different number of time slots to each node according to the length of the data packet to be transmitted to improve channel utilization. In this paper, the performance of PriAlloc-MAC is evaluated and compared with WirelessHART. Results show that PriAlloc-MAC is significantly better than WirelessHART in terms of traffic delay and channel utilization.

In Aeronautical Ad Hoc Network (AANET), attacks are often diverse, and a lack of preventive measures may cause serious losses. Therefore, compared with traditional MANET, AANET has higher security requirements. In this paper, we propose a trust-based partially distributed key management scheme to consider both capability and integrity trust of network. Capability trust reflects the link state of the routing path and the metric is used to choose reliable intermediate nodes for transmission. The integrity trust distinguishes compromised nodes from well-behaving nodes. The key management scheme only issues certificates to the well-behaving nodes to verify their legitimacy. We perform simulations and compare the proposed scheme with a composite trust-based public key management scheme (CTPKM). The simulation results demonstrate the performance advantages of our scheme in terms of overhead, packet delivery ratio and evaluation accuracy.

Despite the substantial progress of communication with Shannon theory in recent years, there are still many development opportunities in the transmission process. In this paper, we focus on the convergence of communication and computing to reduce transfer bandwidth, which transfers feature instead of information. Specifically, we propose a framework based on the Generative Adversarial Networks (GANs) for the first time to transfer feature. The simulation test and result analysis show that the joint process of communication and the computing relied on knowledge architecture opens up new possibilities for improving communication capability and surpass the Shannon limit under certain circumstances.

In this paper, we firstly establish the model of inductive power transfer (IPT) through metal barriers by using the electromagnetic model. Then we theoretically analyze the operating frequency range of IPT through metal barriers, which includes the optimal operating frequency. After that, the theoretical results are proved by simulation in MAXWELL and SIMPLORER.

Caching and multicasting are two promising methods to support massive content delivery in multi-tier wireless networks. In this paper, we consider a random caching and multicasting scheme with caching distributions in the two tiers as design parameters, to achieve efficient content dissemination in a two-tier large-scale cache-enabled wireless multicasting network. First, we derive tractable expressions for the successful transmission probabilities in the general region as well as the high signal-to-noise ratio (SNR) and high user density region, respectively, utilizing tools from stochastic geometry. Then, for the case of a single operator for the two tiers, we formulate the optimal joint caching design problem to maximize the successful transmission probability in the asymptotic region, which is nonconvex in general. By using the block successive approximate optimization technique, we develop an iterative algorithm, which is shown to converge to a stationary point. Next, for the case of two different operators, one for each tier, we formulate the competitive caching design game where each tier maximizes its successful transmission probability in the asymptotic region. We show that the game has a unique Nash equilibrium (NE) and adopt an iterative algorithm, which is shown to converge to the NE under a mild condition. Finally, by numerical simulations, we show that the proposed designs achieve significant gains over existing schemes.

Recent studies have preliminarily investigated and proven the feasibility and effectiveness of applying Content Centric Networking (CCN) principles to Mobile Ad-hoc Networks (MANETs) for content-oriented wireless communications in infrastructure-less mobile environments. However, existing designs may not achieve the full potential of CCN-MANETs. In this paper, we develop a content-centric communication system, named CC-OffGrid, for efficient content delivery (over possibly long distances) in infrastructure-less mobile environments, by deploying a CCN layer directly on top of the media access control (MAC) and physical (PHY) layers. In CC-OffGrid, each user is equipped with a mobile installing a designed APP and an integrated chip with Bluetooth4.0, MSP430 and Lora™ module. We propose a MAC frame structure to achieve unicast, broadcast as well as multi-hop transmissions. We also design an interface protocol for the communication between the APP and the MAC layer. In addition, to effectively alleviate the broadcast storm of Interest Packets, we propose two optimization-based next-hop broadcasting node selection algorithms for Interest Packet broadcasting, based on one-hop and two-hop neighbor information, respectively, obtained using global positioning system (GPS). By adding a hop counter in Interest Packets and a data dissemination limit (DDL) counter in Data Packets, we can obtain node speed-based DDL and use it for alleviating the broadcast storm of Data Packets in mobile environments. We also propose an efficient caching algorithm for data caching by effectively exploiting content popularity information. Finally, we evaluate the performance of CC-OffGrid in ns-3 and test CC-OffGrid on hardware testbeds.

Detection probability is one of the most important criteria in the satellite-based automatic identification system. It represents how many vessels can be detected in the monitoring area of the system. However, existing research often neglects practical factors in S-AIS (satellite-based AIS) and cannot assess the performance of a practical system. In this paper, we propose a much detailed collision model by considering both the effects of SINR (Signal to Interference plus Noise Ratio) caused by antenna patterns and the demodulation performance. Through this model, we can construct much more accurate and reasonable detection probability calculation method. Simulation results show that the detection probability from the proposed practical model is much larger than that from the legacy model.

Space-based Automatic Identification System (AIS) uses Gaussian minimum shift keying modulation (GMSK) as the modulation scheme. Thus, Viterbi Algorithm (VA) is commonly implemented considering the tradeoff between performance and complexity. For the further improvement of the performance of Viterbi decoding, we propose a Soft-Output Viterbi Algorithm for AIS receiver (SOVA-AIS) in this paper. The proposed SOVA-AIS introduces soft values to quantify the error probability of demodulated bits and use them to update the survivor path in VA to get more precise estimation of the signal phase. Simulation results show that, compared with VA, SOVA-AIS can provide 0.5-1.5dB gain in the case with only one AIS signal and 1-3dB gain in the case with two collided AIS signals.

Single Carrier-Frequency Division Multiple Access (SC-FDM) and Tone Reservation (TR) are used independently to reduce Peak to Average Power Ratio (PAPR) in wireless communications. In this paper, we jointly consider TR and SC-FDM to achieve much lower PAPR. We first present the PAPR of SC-FDM system with different DFT/IDFT size, and then theoretically analyze the PAPR gain of TR in SC-FDM. By considering the impact of TR on transmission power, we propose a novel metric, the effective signal power, to measure the performance of TR in SC-FDM. Afterwards, the TR optimization problem is formulated and solved by TR gradient algorithm. Finally, the performance of TR with SC-FDM is evaluated.

Because of the wide field of view (FOV) to monitor vessel movements, satellite-based automatic identification system (S-AIS) has been promoted in recent years. However, synchronization is a tough work in S-AIS receiver because of the large Doppler shift and serious signal collision resulting from simultaneous transmission in the FOV of a satellite. In this paper, we propose a pipelined synchronization approach with good performance and low complexity. It consists of three parts, i.e., packet detection, windowing and timing recovery, and frequency and timing offset estimation, which can be implemented in feedforward structure with low complexity. Simulation results show that the proposed synchronization approach has significant performance gain compared with existing algorithms and provides an acceptable estimation range to cope with large Doppler shift and message collision.

In this paper, we propose a Differential Overlap Decoding (DOD) algorithm to solve the hidden-terminal problems in OFDM based WLAN systems. DOD exploits the retransmission and random-jitter features of WLAN to decode the collided packets as a whole. DOD provides with new methods and views to jointly separate and decode collided packets in OFDM based system. In DOD, we formulate IFFT/FFT, channel influence and packet collision as linear processes. Thus we can express the received collided packets as linear equations, and simplify this problem into solving linear equations. We evaluate the performance of DOD through simulation, and results show that when DOD is applied, hidden-terminal problems can be viewed as a 3-5 dB BER performance degradation rather than network contention.

The problem of jointly estimating clock skew and offset for wireless sensor networks (WSNs) in a pairwise broadcast synchronization (PBS) protocol is considered. The random part of the delay is supposed to be an exponential random variable. We consider two estimators, i.e., joint maximum-likelihood estimator (JMLE) and generalized ML-like estimator (GMLLE) proposed by Leng and Wu . For both estimators, the corresponding algorithms are explicitly derived and presented. For the GMLLE, the corresponding performance bound based on the reduced set of observations is derived and the optimal value of a user-defined parameter is identified accordingly. At last, analytical results are corroborated by numerical experiments. We observe that: (i) JMLE usually outperforms GMLLE at the cost of larger computational complexity; (ii) JMLE, while achieving the same estimation accuracy as that of the LP method presented in , enjoys significantly lower computational complexity than that of the latter.

A cognitive network refers to the one where two overlaid structures, called primary and secondary networks coexist. The primary network consists of primary nodes who are licensed spectrum users while the secondary network comprises unauthorized users that have to access the licensed spectrum opportunistically. In this paper, we study the percolation degree of the secondary network to achieve k-percolation in large scale cognitive radio networks. The percolation degree is defined as the number of nearest neighbors for each secondary user when there are at least k vertex-disjoint paths existing between any two secondary relays in the percolated cluster. The percolated cluster is formed when there are an infinite number of mutually connected secondary users spanning the whole network. Each user in the cluster is possibly connected to several neighbors, inducing more communication links between any two of them. Since nodes located near the boundary have fewer neighbors, the boundary effect becomes a bottleneck in determining the percolation degree. For cognitive networks, when the primary node density becomes considerably large, the boundary effect spreads inside the network. The transmission area of most secondary users who are located near the primary nodes decreases due to the restriction of the primary network. Therefore, to ensure k-connectivity in the percolated cluster, each secondary user must be connected to more neighbors, and the percolation degree of the secondary network yields a function of the primary node density. We specify the relationship into three regimes regarding the topology variation of the cognitive network. A closed-form expression of the percolation degree under different primary node densities is presented. The expression characterizes the connectivity strength in the secondary percolated cluster, therefore providing analytical insight on fault tolerance improvement in cognitive networks.

Vehicular Ad Hoc Networks (VANETs) with base stations are called hybrid VANET, where base stations are deployed to improve the throughput capacity. In this paper, we study the multicast throughput capacity for hybrid wireless VANET with a directional antenna on each vehicle and the end-to-end delay is constrained. In the hybrid VANET, there are n mobile vehicles (or nodes) distributed in a unit area with m strategically deployed base stations connected using high-bandwidth wire links. There are n_s multicast sessions and each multicast session has one source which transmits identical data to its associated p destinations. We investigate the multicast throughput capacity for two mobility models with two mobility scales, respectively, while each vehicular node is equipped with a directional antenna and with a tolerant delay D. That is, a source node transmits to its p destinations only with the help of normal nodes within D consecutive time slots. Otherwise, the transmission will be performed with in the infrastructure mode, i.e., with the help of base stations. We demonstrate that the one dimensional i.i.d. slow mobility pattern catch the main feature of VANETs. And we find that the multicast throughput capacity of the hybrid wireless VANET greatly depends on the delay constraint D, the number of base stations m, and the beamwidth of directional antenna θ. In the order of magnitude, we obtain the closed form of the multicast throughput capacity of the hybrid directional VANET, where the impact of D, m and θ on the multicast throughput capacity is analyzed. Moreover, we derive the lower bound of the muticast throughput using a similar raptor coding approach.

In this paper, we propose an agent-based spectrum trading model, where an agent can play a third-party role in the spectrum trading process. Providing service to Secondary Users (SUs) with spectrum bought from Primary Users (PUs), the agent can make profits during the process by providing service to secondary users. During each trading period, the agent has to decide how much spectrum it should lease from PUs and what price it should charge SUs. Therefore, the most significant challenge to implement this spectrum trading model is finding the most profitable strategy for agent(s). We address this challenge under two scenarios in which: 1) a single agent and 2) multiple agents. Instead of quantifying SUs' spectrum demand by a deterministic function of price, we take the randomness of secondary users' demand or demand uncertainty into consideration. To the best of our knowledge, this is the first solution to agent-based spectrum trading considering demand uncertainty.

We focus on the problem of single image super-resolution in this paper. Given a low-resolution image, we seek to synthesize its underlying high-resolution details using a learning based method. Inspired by recent progress in compressive sensing, we use sparse representation prior to regularize this ill-posed problem. On the other hand, with natural image statistics taken into consideration, we enforce the prior only on those image patches associated with image primitives rather than on arbitrary ones. Specifically, each patch from primitive layer of the lowresolution image, which can be viewed as a low-dimensional projection of a high-resolution primitive patch, is conjectured to have a sparse representation concerning an over-complete dictionary. Under mild conditions, the sparse representation can be correctly restored from the low-dimensional projection according to the theory of compressive sensing. We also construct a dictionary using image primitive patches which works well on generic input images. Experiment results show the efficiency of our method by outperforming other learning-based methods both subjectively and objectively.