Cellular-connected unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC) has emerged as a key enabler for the low-altitude economy network (LAEN), supporting task offloading from UAVs to ground base stations (GBSs) in aerial Internet of Things (IoT) applications. However, existing studies focus on conventional stochasticity such as fading channels and random task arrivals, while largely overlooking the dynamic service availability of rapidly deployed ad-hoc GBSs in extreme, high-demand scenarios, characterized by a stress-driven lifecycle including resource depletion, probabilistic service disruptions, and delayed recovery. Such dynamics may cause severe energy inefficiency and queue backlogs, undermining system reliability. To address these issues, we formulate a joint optimization problem of resource allocation and 3D trajectory planning, which explicitly integrates a multi-dimensional GBS reliability-aware model to capture this service lifecycle, aiming to minimize the weighted energy–delay system cost while ensuring long-term queue stability. We develop a dual-layer online optimization framework: the outer layer reformulates the objective function using a time-adaptive velocity-triggered penalty term (TA-VTPT) to suppress energy spikes from abrupt velocity changes, then leverages Lyapunov drift-plus-penalty to decompose the original problem into per-slot subproblems; the inner layer proposes a Lyapunov-guided multi-agent dual actor-critic (LyMADAC) algorithm to generate real-time decentralized decisions without future information. Simulations show that our approach effectively reduces the system cost while ensuring queue stability across heterogeneous traffic patterns and extreme stress scenarios, and further demonstrates robustness under representative non-idealities (e.g., wind disturbance and battery aging).

In mobile millimeter-wave channels, accurate channel prediction improves the performance of wireless communication systems. However, channel state information (CSI) obtained using pilot signals is interfered by noise, and the sequential prediction of CSI leads to error accumulation, which limits the performance of the channel prediction methods. To solve the noise interference problem and the error propagation problem in channel prediction, we propose a channel predictor based on Transformer network. By using learned positional encoding instead of the traditional sinusoidal positional encoding and introducing a convolutional module to enhance the multi-head self-attention layer, the proposed method reduces the impact of noise. And the parallel prediction capability of Transformer is utilized to address the issue of error propagation. By simulation, within the acceptable computational complexity, the proposed method can achieve better prediction performance in noisy mobile scenarios compared with other neural network-based channel prediction methods. Numerical results show that on average, the normalized mean square error (NMSE) of the proposed method is 2.46 dB lower than that of the canonical Transformer-based channel prediction method under different prediction lengths.

The detection and identification of radio signals play a crucial role in cognitive radio, electronic reconnaissance, noncooperative communication, etc. Deep neural networks have emerged as a promising approach for electromagnetic signal detection and identification, outperforming traditional methods. Nevertheless, the present deep neural networks not only overlook the characteristics of electromagnetic signals but also treat these two tasks as independent components, similar to conventional methods. These issues limit overall performance and unnecessarily increase computational consumption. In this paper, we have designed a novel and universally applicable integrated radio detection and identification deep architecture and corresponding training method, which organically combines detection and identification networks. Furthermore, we extract signal features using only one‐dimensional horizontal convolution based on the characteristics of the impact of wireless channels on time‐domain signals. Experiments show that the proposed methods perform signal detection and identification more efficiently, which can not only reduce unnecessary computational consumption but also improve the accuracy and robustness of both detection and identification simultaneously. More specifically, the ability to distinguish different modulated signal categories tends to increase with the rise in SNRs, and the upper limit of detection accuracy can exceed 95% at SNRs above 0 dB. The proposed method can improve both signal detection and identification accuracy from 83.44% to 83.56% and from 61.27% to 62.32%, respectively.

Utilizing integrated waveforms to achieve communication and sensing functions simultaneously is the key technology of Integrated Sensing and Communication (ISAC). With the development of ISAC, sensing will play a greater role than ever before, especially in location/environment sensing scenarios. In this paper, we utilize the OFDM echo signal from a single antenna to perform range and velocity estimation based on the modulation symbol domain. We propose an optimized two-dimensional MUSIC algorithm capable of jointly estimating the range and relative velocity of multiple targets. Additionally, we employ the Fast Fourier Transform to accelerate the algorithm. Our algorithm overcomes the range and velocity matching problem, surpassing the limitations of 1D-MUSIC. Compared to 2D-FFT, it offers advantages such as super-resolution, high precision, and robustness against noise interference. The range and velocity resolution of our algorithm are respectively up to 9 times and 3 times that of 2D-FFT. In scenarios with multiple targets having similar ranges and velocities, the RMSE of range and velocity can be reduced to less than 1/10 of 2D-FFT. Furthermore, the normalized noise amplitude of the power map is 40dB to 60dB lower than that of 2D-FFT.

Multi-cell massive MIMO is a key technology in the future communication due to its ability to compensate the fading of high-frequency channels. However, the extensive use of antennas causes higher power consumption. In general, base stations (BS) prefer low-resolution analog-to-digitalldigital-to-analog converters (ADC/DAC) to reduce power consumption, which also leads to the quantization effects. In this work, we propose a coordinated beamforming method to maximize energy efficiency (EE) in multi-cell communication systems with the impact of quantization effects on user signal-to-interference-plus-noise ratio (SINR). Through fractional programming (FP) and convex optimization approximation, the non-convex EE maximization problem is transformed into a convex problem ultimately. Simulation results demonstrate that the system reaches an optimum EE by adjusting the number of antennas. Furthermore, appropriately deploying the number of users and cells can simultaneously improve network capacity and EE.

In this letter, we address the challenge of employing multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) to simultaneously estimate the range, velocity, and angle of multiple moving targets at the transmitter, while maintaining communication. Specifically, we present a mathematical model for radar sensing signal processing and elaborate on a successive estimation scheme. We reveal that the compensation processing algorithm proposed in existing successive estimation scheme suffers from the problem of ambiguous matching, which fails to properly match the angle and the range-velocity for targets with close angles. To address the ambiguous matching problem, this letter proposes a novel compensation processing algorithm for range-velocity estimation. Based on minimum variance unbiased estimation, the proposed compensation processing algorithm fully exploits the estimated angle information to extract corresponding range-velocity information for eliminating the interference from targets at close angles.

Low Earth orbit (LEO) satellite networks will become the core infrastructure of space-air–ground integrated networks in 6G wireless technology, providing ubiquitous broadband internet services for global users in a cost-effective manner. However, timing synchronization is a significant problem in large-scale multisatellite (MSAT) communications. Due to the different propagation delays, signals transmitted from multiple satellites to a user terminal are always misaligned. In this paper, we study asynchronous satellite-terrestrial interference in LEO MSAT systems, aiming to mitigate asynchronous interference through cooperative beamforming. Specifically, we derive an MSAT coherent transmission model and analyse the power degeneration of interference due to the offset sample time. We also derive the closed-form expression for the achievable rate by considering asynchrony and propose a cooperative beamforming algorithm based on the weighted minimum mean square error (WMMSE) criterion. To realize accurate timing advance (TA), we design a delay estimation method based on pseudorandom code. Finally, we approximate the achievable rate under an imperfect TA and propose a timing-error–robust algorithm. The simulation results show that the proposed asynchronous algorithm achieves a 21.47% performance improvement over the conventional algorithm, indicating that by exploiting time asynchrony, we can naturally eliminate some interference and achieve substantial performance improvements.

In the field of machine vision and pattern recognition, the convolutional neural network (CNN) is one of the hottest research topics. However, the application of CNN, which requires complicated operations, appears to be exceptionally difficult in resource-constrained embedded devices. In this article, a parallel processing CNN accelerator based on optimized MobileNet is presented. By modifying the fully connected layer of the MobileNet network topology with the convolution process, and postponing the global pooling layer, the model topology is unified, which is conducive to the design of the hardware accelerator. After using the 8-bit quantization strategy of network model parameters, the process of depthwise separable convolution is accelerated by parallel processing between channels and pipelined processing between layers. Thus, the processing speed and throughput of the accelerator can be improved. The designed accelerator classification performance on ImageNet achieved 580.6 frames per second (fps) on a ZYNQ AXZU5EV platform and a system power consumption of only 6.51 W. This result represents a$22.3\times $speedup compared to CPU and a$1.7\times $speedup compared to graphics processing unit (GPU), while the design has a lower power consumption than CPU and GPU, providing a reference for the application of CNN in embedded devices.

Low probability of intercept (LPI) radar waveform recognition (LWR) based on deep learning can significantly enhance the recognition accuracy. However, most existing LWR networks overlook the algorithmic complexity and disparities between visual images and time-frequency images (TFIs). In practical applications, it is critical to enhance the recognition accuracy while minimizing the computational complexity to satisfy the reliability and timeliness requirements of an LWR network. To address this issue, we propose an efficient depth-tunable network (EDTN) for LWR. The EDTN reduces the computational complexity and enhances the recognition accuracy by adopting a flexible network structure, using convolution methods appropriate for TFIs, and combining several beneficial designs. Our experiments with a classical LPI radar waveform dataset demonstrate that the EDTN reduces the computational complexity by more than 95% over that of the state-of-the-art networks while maintaining the recognition accuracy.

Carrier Sense Multiple Access (CSMA) has been widely applied to various kinds of wireless networks, such as Wi-Fi, to serve portable devices which are usually greedy in terms of throughput, but with finite battery budget. Accordingly, how to optimize the usage of finite battery budget to get the best possible throughput performance is of great importance. This paper aims to address this issue by focusing on a saturated CSMA network. Explicit expressions of maximum energy efficiency and the corresponding optimal backoff parameter with or without throughput constraint are derived. It is revealed that optimizing the energy efficiency leads to throughput performance degradation. With a stringent throughput constraint, the energy efficiency has to be sacrificed. The energy efficiency and the throughput can be optimized at the same time only in special cases, e.g., the network size is large. The analysis is verified by simulations and sheds important light on performance optimization of practical CSMA-based networks such as Wi-Fi 6 networks.

In recent years, many communication standards have adopted the LDPC (Low-Density Parity Check) code due to its excellent error correction performance, but it also induces some problems. For example, the use of 5G in mobile phones dramatically reduces battery life. Therefore, it is particularly significant for LDPC code to design a decoding algorithm that does not increase hardware implementation complexity but has superior error correction performance and iteration speed faster than traditional decoding algorithms. Inspired by several network structures in artificial intelligence, bias is used to improve the performance and robustness of the network. In order to enhance decoding algorithm robustness and accuracy, a method is shown by changing the offset in OMSA to the bias and combining NMSA in the near-zero range in this paper. It significantly avoids the log-likelihood ratio information loss when less than the offset value and diminishes too quickly or sluggishly. A Segmented Bias/Normalize Min-Sum(SB/NMS) LDPC decoding algorithm is proposed with a layered structure. This algorithm can avoid the impact of the problems mentioned above and significantly improve the performance and iteration speed of the decoding process. Compared with the BLER (resp. BER) performance the traditional BP, NMS and OMS algorithm achieves a gain of 0.04, 0.11, and 0.15dB (resp. 0.035dB, 0.06, and 0.13dB). Meanwhile, the minimum SNR requirements of BLER =0.001 can be reduced by 0.02 dB compared with the BP algorithm. The parameters of SN/NMSA can be pre-processed on the computer, so there will be no additional increase in the overhead in resources and get better performance while getting faster iteration speed, while greatly guaranteeing the reliability and effectiveness of data transmission.

Coded caching has been studied extensively and extended to many scenarios because it can exploit multicast opportunities to reduce the downlink traffic on the network. The basic model of coded caching is that a server with$N$files of the same size, is connected to several users with cache size$M$through a shared link. Prior work [1] has derived the optimal rate-memory tradeoff of this system with the constraint of uncoded prefetching. However, when heterogeneity of file sizes is taken into account, there remains open questions. In this paper, we consider two users and$N$files with distinct file sizes. When the cache size is large or small compared to the overall file size, i.e., 0 ≤$M$≤ ½ N F1or$M$≥ H(W) - ½ N F1where$F$1denotes the minimum file size and H (W) represents the overall information entropy of all files in the database, we show that our proposed scheme is optimal under all possible schemes. When the cache size is mediate, i.e., ½ N F1$M$1we show that our proposed scheme is optimal under the constraint of uncoded prefetching. To obtain these results, we introduce new techniques to extend the novel cut-set bound in [2], and propose a new achievable scheme based on both files and caches sizes.

In this paper, we are interested in the joint estimate of user equipment (UE) location and orientation for millimeter-wave multi-input-multi-output (mmWave MIMO) systems. In practice, mmWave signals suffer from small-scale fading, which degrades UE localization. Moreover, mmWave MIMO-based UE localization is a non-convex optimization problem, and the bruteforce application of conventional optimization methods will result in a poor solution or lead to large computational cost. In order to address the above challenges, we propose a novel simultaneous localization and channel estimate (SLCE) algorithm, where the UE location parameters and small-scale fading coefficients are jointly optimized. In such a case, the disturbance of small-scale fading on UE localization is alleviated. Thanks to our problem-specific update rule design, the proposed SLCE algorithm achieves a large performance gain over existing baseline methods.

Coded caching is an effective method to reduce the traffic load on the network bottleneck link by exploiting the joint optimization of caching and transmission. In most of previous works, one user may request only one file. However, users could have distinct number of requests in practice. The number of files requested by the users may also follow different distributions. In this paper, we investigate coded caching with distinct number of user requests. We propose a decentralized coded caching transmission scheme for user requests whose number may follow an arbitrary distribution and analyze the upper bound of the transmission rate. We also derive the corresponding lower bound for any achievable scheme. We show that the gap between these two rates is within a constant factor of 12. And this result holds for any distribution on the number of user requests. For the proposed transmission scheme and a given distribution, we show that when the number of users exceeds a threshold, the transmission rate will be bounded without regard to the number of users, due to the benefit of multicast opportunities brought by coded caching. Simulation results demonstrate the superior performance of the proposed scheme.

Traffic congestion, a common and complicated phenomenon in urban transportation systems, is attracting increasing attention from researchers in Geographical Information Science (GIS) and other fields. In this study, we illustrate a general mechanism that reveals the relationship between travel time and dynamic traffic conditions. We measure a vehicle’s travel time to its destination along any path, where the travel time is calculated based on the path length and on the real-time traffic volume and transport capacity of each road segment on the path. On the basis of this measurement, we present a simple dynamic routing strategy that allows each vehicle to dynamically choose the path to its destination while imposing the minimum travel time. The application of our routing algorithm to the Chengdu street network, Barabási–Albert scale-free network and Erdös–Rényi random network shows that the proposed strategy remarkably improves network throughput and balances traffic load distribution. Our findings suggest that mining the time mechanism of network transport is important to explore efficient time-optimization routing algorithms to enhance the transport capacity of urban street networks and other kinds of networks.

Sentiment analysis of short texts is challenging because of the limited contextual information they usually contain. In recent years, deep learning models such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) have been applied to text sentiment analysis with comparatively remarkable results. In this paper, we describe a jointed CNN and RNN architecture, taking advantage of the coarse-grained local features generated by CNN and long-distance dependencies learned via RNN for sentiment analysis of short texts. Experimental results show an obvious improvement upon the state-of-the-art on three benchmark corpora, MR, SST1 and SST2, with 82.28%, 51.50% and 89.95% accuracy, respectively.

Based on least squares wavelet support vector machines (LS-WSVM) with quantum-inspired evolutionary algorithm (QEA), the prediction model of urban passenger transport is proposed , that can provide the theoretical foundation of forecasting passenger volume of urban transport accurately. The prediction model of urban passenger transport is established by using LS-WSVM, whose regularization parameter and kernel parameter are adjusted using quantum-inspired evolutionary algorithm. QEA with quantum chromosome and quantum mutation has better global search capacity. The parameters of LS-WSVM can be adjusted using quantum-inspired evolutionary optimization. Combining with the data of the urban volume of passenger transport of Xipsilaan over years, the prediction model of urban passenger transport is validated, the simulation results indicate that the prediction model is effective, and based on LS-WSVM has more improvement than LS-SVM with Gaussian kernel in predicting precision, and then the improved LS-WSVM with QEA is efficient than with cross-validation method for tuning parameters.

In this paper, a method based on maximum likelihood to estimate the target true height in 2-D radar network is presented, and it is mathematically proved that the estimator is unbiased. The Cramer-Rao low bound (CRLB) of estimation error is also derived. Some conclusions are drawn based on simulations. It’s concluded that the bearing error only has a bounded influence on CRLB, however, the influence on CRLB exerted by ranging error is unbounded. It’s also found that CRLB is superlinear function of target-radar distance, and CRLB affected by target altitude or baseline length acutely as target altitude or baseline length is small.