Transactions on Emerging Telecommunications Technologies最新文献

Video anomaly detection is a critical task in surveillance, industrial quality control, and anomaly monitoring systems. Recognizing anomalous events or behaviors within video sequences is challenging due to the diverse and often vague nature of anomalies. A novel temporal convolutional network-based anomaly detection (TCN-AnoDetect) is proposed that leverages TCNs and self-supervised learning. In this, TCNs are employed to model the temporal context within video sequences effectively, capturing short and long-term dependencies. The algorithm integrates TCNs with pretrained models to encode rich spatiotemporal features. The core of TCN-AnoDetect lies in self-supervised feature learning, where a neural network is pretrained on unlabeled video data to capture high-level spatiotemporal features. The anomaly detection module combines reconstruction-based and temporal context–aware approaches, using reconstruction errors and temporal context deviations for anomaly scoring and classification. To enhance model robustness, TCN-AnoDetect incorporates domain adaptation technique to handle domain shifts and evolving anomalies. The proposed algorithm is evaluated on three different benchmark datasets and ShanghaiTech Campus, demonstrating its superior performance. The extensive experiments performed in terms of different evaluation measures show the efficiency of the TCN-AnoDetect algorithm. The TCN-AnoDetect, an innovative approach, thereby provides promising solutions in video anomaly detection and in various applications.

This paper presents a channel estimation method for an intelligent reflecting surface (IRS)-aided orthogonal time-frequency spacing (OTFS) system in a dynamic scenario. Current channel estimation techniques for IRS-aided OTFS systems are built upon explicit channel model assumptions, which can constrain their adaptability in intricate environments. Furthermore, their reliance on pilot signals introduces significant pilot overhead in high-speed scenarios. To address these issues, we propose a dilated attention generative adversarial network (DAGAN) that has a novel architecture for capturing long-range dependency among data symbols separated in the delay-Doppler (DD) domain for estimating channels. Furthermore, the DAGAN includes an attention block to extract essential features from data symbols for channel information generation. This mechanism is guided by least square (LS) estimates of specific DD paths, serving as additional information for the DAGAN. Experimental results illustrate that the DAGAN method performs the best with the least NMSE with limited pilot overhead in comparison to other methods.

In today's wireless communication systems, the integration of 5G millimeter-wave (mmWave) Massive Multiple Input–Multiple Output (M-MIMO) technology offers significant advancements and capabilities that address the growing demand for higher data rates, increased capacity, and improved user experiences. In three-dimensional (3D) environments, the design of beamforming for uplink multi-user M-MIMO relies on accurate uplink channel state information (CSI) at the transmitter/receiver. In fact, it is difficult for the Base Station (BS) and the User Equipment (UE) to obtain beam patterns due to computational complexity, multiple 3D beams, and regulating the weight of antenna elements, which leads to significantly low sum-rate. Hence, a robust, deep adaptive learning framework in the case of 3D beamforming is needed. This paper proposes a deep adaptive learning-based beam combining framework using User-Wise Attention-Assisted Deep Adaptive Neural Network (UWA-DANN) for mmWave-3DM-MIMO systems. In this, a UWA mechanism learns a set of beam features and corresponding weights for each user. This mechanism allows the system to focus on different users independently and adapt the beamforming process accordingly. Also, it allows the network to dynamically focus on relevant information from the input channels and user constraints. To this end, a dynamic beam pattern is adapted using the DANN model to learn user positions, channel measurements, and beamforming weights. This approach learns to map input parameters such as user positions, channel measurements, and corresponding beam patterns to extract relevant features for beam pattern adaptation. Thus, the UWA-DANN approach provides higher data rates, low complexity, and improved link stability for users. Experimental results show that the proposed UWA-DANN model obtains robust performance over existing schemes in terms of achievable rate and sum-rate under field trial sites in urban scenarios.

Existing research in the field of reconfigurable intelligent surface (RIS)-aided physical layer security assumed Gaussian signal inputs, which is inapplicable to practical communication systems, where finite-alphabet inputs are used. This paper considers an RIS-aided secure multiple-input multiple-output wireless communication system with finite-alphabet inputs, where artificial noise (AN) is invoked at the transmitter to enhance the secure performance. In order to maximize the secrecy rate (SR), the data precoder, the AN precoder, and RIS's reflection coefficients are jointly optimized subject to the constraints of the maximum transmit power and the finite resolution of the phase shifts of RIS. Particularly, due to the finite-alphabet input, the exact expression of the SR involves multiple integrals and lacks a closed-form expression. To tackle this, a closed-form lower bound of the SR is derived as the objective function, which is theoretically proved to be equal to the SR in the high signal-to-noise ratio region. Numerical results show that the RIS can significantly improve the secure performance, and the maximum possible SR (due to the finite-alphabet inputs) can be achieved by increasing the number of the RIS's elements or by increasing the transmit power, which shows the performance advantage of the proposed optimization algorithm.

Rapid growth and technological improvement in wireless communication, driven by engineers from various disciplines, have reached significant milestones. Unmanned Aerial vehicles (UAVs) and flying ad hoc networks (FANET) have undergone one of the biggest innovations. UAVs have drawn a lot of attention from research institutions. They are increasingly employed in various application fields, such as real-time monitoring, precision farming, wireless coverage, military surveillance, climate monitoring, disaster surveillance, and monitoring and rescue operations. The primary characteristics of disasters are their unpredictability and the scarcity of resources in the affected areas. To reduce the loss of lives and livelihoods, disaster management has received much attention. Numerous methodologies and technologies have been developed to predict and handle disasters. UAVs are increasingly being used in disaster management. Additionally, artificial intelligence and collaborative machine learning techniques are gaining prominence among researchers, who are investigating the possibility of their use in disaster management tasks to better cope with the severe and frequently devastating effects of natural catastrophes. This paper provides a review of the relevant FANET research activities in disaster management and emerging artificial intelligence techniques, along with several observations and research challenges. The papers are categorized based on the disaster scenario-related problems and their proposed solutions. FANET problems are receiving less attention from the research community, and the main challenges with FANETs are also highlighted. Finally, significant insights are presented that can aid in improving research related to the application of FANETs in disaster management.

Mobile edge computing (MEC) has achieved significant attention due to the availability of computational tasks in specific scenarios such as emergency applications like forest fire and earthquake remedies. The computationally demanding policy and user offloading policy are challenging problems to address in the energy constrained unmanned aerial vehicle (UAV) network. In this work, the computational task offloading, and power management is solved by using the multi-agent deterministic power management algorithm (MADPM) based on deep reinforcement learning. Every UAV works together as a team to understand the actor critic environment and to make decisions that will help achieve the goals. This involves transferring computational tasks from UAVs to more powerful ground stations or other UAVs to save energy and enhance performance. It requires intelligent decision-making to determine which tasks to offload and when. The joint optimization problem is verified with the simulation results and the proposed work is enabled with MEC in achieving the emergence of UAV related applications. Our simulations show that the MADPM algorithm, as suggested, enhances task offloading efficiency by 35% and reduces power consumption by 25% when compared with current methods. These findings underscore the ability of our method to greatly improve the UAV operational capacities.

In this paper, a full-duplex (FD) space-air ground integrated network (SAGIN) system with passive and active eavesdroppers (PE/AE) and a friendly jammer (FJ) is investigated. The shadowing side information (SSI)-based unmanned aerial vehicle relay node (URN) selection strategy is considered to improve signal-to-interference plus noise power ratio (SINR) at the ground destination unit. To quantify the secrecy performance of the considered scenario, outage probability (OP), interception probability (IP), and transmission secrecy outage probability (TSOP) are investigated in the presence of FJ and PE/AE. The results have shown that aerial AE is an important threat since it can severely degrade the OP of the main transmission link. Furthermore, the FJ can decrease the IP of the eavesdropper by causing interference with the cost of power consumption of URNs. Simulations are performed to verify the theoretical findings.