Transactions on Emerging Telecommunications Technologies最新文献

This work introduces an orthogonal frequency division multiplexing based discrete cosine transform assisted index modulation with novel signal identification technique. To take use of the design flexibility offered by the twice the number of accessible subcarriers under the same bandwidth, it combines the concepts of IM and DCT assisted Orthogonal Frequency Division Multiplexing (DCT-OFDM). The performance of DCT-OFDM-IM in contrast to OFDM-IM is enhanced in the proposed study by the employment of a deep learning detector. The Deep Learning based detector (DLD), in contrast to conventional detectors like Maximum Likelihood (ML), Greedy Detector (GD), Log Likelihood Ratio (LLR), and others, improves system performance and lowers system overhead. In order to perceive data bits at the OFDM-IM system's receiver in Rayleigh, Rician, and Nakagami-m Fading channels, the proposed DLD uses a Deep Neural Network with completely automated linking layers. To start with, DLD is trained offline by assembling datasets of simulated results in order to enhance BER performance. Next, the model is trained to recognize DCT-OFDM-IM signals at the receiver under various fading channels. The results demonstrate that the DLD outperforms conventional approaches for all multipath fading channels in terms of BER, and that BER for DCT OFDM-IM has improved over that of OFDM-IM.

The advancement of digital technologies is used for providing more enhanced healthcare services to patients in a timely and effective manner. The multi-modal data encompasses a huge amount of information when compared to the single-modal data. Fusing and analyzing various data types provides a more comprehensive understanding of the patient's condition. Fusing these multi-modal data poses several technical challenges because of its data incompatibility. Therefore, this research work focuses on implementing a deep learning-based disease prediction model using multi-modal data to generate precise prediction results regarding healthcare applications. Initially, the required multi-modal data such as signal, data and image are gathered from the standardized benchmark data sources. Then, the collected data is subjected to the implemented multi-modal data-based disease prediction network (MMPredNet). This network is developed by combining an adaptive hybrid deep learning network (AHDLN) and a multi-scale dilated residual attention network (MDRAN). Here, MDRAN performs a feature extraction process to extract the features from the input data. further, the prediction process is carried out using the AHDLN model. It is a hybridized network generated by fusing a deep Bayesian network (DBN) with a deep shallow network (DSN). The parameters of the AHDLN are optimized using the adaptive learning rate-based dove swarm optimization (ALR-DSO) algorithm to reduce FPR and enhance the precision, NPV, and accuracy of the prediction outcome. From the MMPredNet, the final disease prediction outcomes are provided. The performance of the implemented multi-modal data processing model is evaluated with various conventional methods to showcase its effectiveness in healthcare. The accuracy of the developed model on text data is 97.31%, images are 98.02%, and the signal is 97.23%, which is enhanced than the prior works. Hence, it is proved that the developed framework can accurately predict the disease at an early stage and helps to improve patient outcomes and prevent the progression of diseases in patients.

Windy conditions challenge precise payload delivery by unmanned aerial vehicle (UAV), but wind variability defined within lower and upper bounds at any instance can assist to optimize candidate release points. Therefore, at first, it is crucial to collect candidate release points caused by wind variability. In this paper, knowledge of candidate release points is drawn by applying ballistic equation. The knowledge about the points then initializes differential evolution (DE) optimization to search an optimum payload release point. Therefore, the proposed method is named as DE with knowledge-based initialization (KI), that is, DE-KI. Simulations demonstrate DE-KI's effectiveness by measuring landing error as root mean square error (RMSE) and achieve an average reduction in RMSE compared to existing methods. For instance, DE-KI outperforms two other alternative approaches by an average RMSE of $$ and $$ with varying payload weight, and $$ and $$ with varying wind speed.

Automatic monitoring of heterogeneous devices across Space-Air-Ground Integrated Networks (SAGIN) remains a significant challenge due to the complex temporal dependencies inherent in multivariate time series and the vast amount of data generated across space-based, aerial, and ground sensors. Hybrid models have proven effective for time series anomaly detection by identifying abnormal segments through high reconstruction errors, a strategy particularly valuable for multi-source data streams in SAGIN scenarios. However, these methods typically fall short in addressing the non-stationarity and noise inherent in multivariate time series, as they use fixed thresholds and lack mechanisms to adapt to changing data distributions—an issue exacerbated in SAGIN environments with widely varying network conditions. In contrast, our approach employs a dynamic threshold selection strategy that automatically adjusts based on the statistical properties of the reconstruction error, thus effectively mitigating these issues in SAGIN's dynamic environment. Consequently, these earlier models fail to extract rich differential features from both local and long-term sequences, thereby limiting detection performance—particularly under the multi-scale, distributed conditions of SAGIN. This study introduces VAML-Net, a composite architecture designed for unsupervised detection of anomalies within multivariate time series, and especially tailored to the heterogeneous data and distributed nature of SAGIN environments. The framework incorporates a Variational Autoencoder to derive compact representations from localized temporal segments, which are subsequently utilized for data reconstruction. To model extended and hierarchical temporal dependencies, the architecture integrates a multilevel LSTM configuration, enhanced with a cross-layer information aggregation mechanism, mirroring the multi-tier structure of SAGIN. Furthermore, we propose a dynamic threshold selection approach that adapts to the inherent non-stationarity and noise present in real-world time series data by continuously recalculating the threshold based on the evolving statistical properties of the reconstruction errors. Extensive experiments conducted on six anomaly detection benchmark datasets demonstrate that the proposed method consistently outperforms other state-of-the-art techniques.

Recently, underwater wireless communication (UWC) networks have garnered significant attention. In specific application scenarios, underwater electrocommunication technology exhibits distinct advantages over traditional acoustic and optical communication methods, emerging as a viable alternative for communication among autonomous underwater vehicles (AUVs). Most AUVs depend heavily on battery power, where the energy is highly precious. Given that the reliability of AUVs communications is tethered to limited energy storage, the imperative for energy-efficient communication strategies is paramount. The issue of power consumption control in underwater electrocommunication systems is addressed in this research by proposing an adaptive power control strategy based on transfer learning for transferring power. The method can predict the minimum voltage across the transmitting electrodes required to satisfy the communication task according to the changes in the operating environment and adjust the transmitting power level accordingly. To verify the effectiveness of this method, this paper establishes a transfer network based on simulation data obtained by finite element simulation combined with the theory and technique of transfer learning. It uses experimental samples to verify the effectiveness of this network in shallow waters. According to the findings, the transfer network outperforms the ordinary backpropagation neural network trained solely on experimental samples in terms of performance.

In recent years, the development of distributed learning systems and advancements in deep learning have led to significant improvements in music style transfer techniques. However, these improvements face significant challenges when implemented in space-air-ground integrated network (SAGIN) environments, due to issues such as high latency, limited bandwidth, and privacy concerns. This research explores a distributed, cross-domain music style transfer model based on SAGIN environments, proposing a federated learning (FL) approach to mitigate these challenges. The proposed method facilitates efficient music style transformation while maintaining high content and style fidelity, ensuring privacy protection by keeping sensitive data localized to edge devices. We analyze and compare the performance of several models on different music datasets (including classical, jazz, and rock genres), demonstrating that our method outperforms traditional centralized models in terms of latency, communication efficiency, and privacy preservation. Moreover, we present several ablation experiments, illustrating the contribution of each component in the model. The proposed method demonstrates its applicability in distributed, real-time environments, offering a solution for scalable and privacy-preserving music style transfer applications in the SAGIN framework.

Nowadays, this society relies more and more on large-scale intelligent networking to operate the desired functions, which leads to a great and tremendous amount of Internet traffic, particularly at peak time. Such exponential traffic increase has caused a great challenge to network infrastructure, which in turn reflects the importance of network traffic prediction. Reliable traffic prediction can help the Internet service provider to manage their resource efficiently, so as to guarantee the service quality even at high demand and prevent the network congestion from happening. However, with the access to a tremendous smart devices, the corresponding traffic grows exponentially. In this way, it becomes vitally important to accurately capture and predict such traffic status. Traditional prediction models are usually applied to short-term prediction scenarios, which are not suitable for real-world scenarios, because the traffic is more complex. On the other hand, deep learning has been frequently used for network prediction due to its non-linear modeling capability. Nevertheless, these methods may encounter trouble when dealing with the problems related to the long dependence relationship and dynamic space relevance among traffic data with a large amount of data. To address these challenges well, we propose a novel prediction architecture using the transformer structure. It takes the time and space factors into consideration when fulfilling traffic prediction. Specifically, on one hand, we separate the input sequence along the timeline, so as to better capture the dynamic space relevance to traffic. For each part of the captured sequence, we build the sub-model for relevance modeling. Then, with these discrete traffic models with time and space features, we introduce the multi-head attention mechanism to integrate them, so as to finally build the perfect relevance matching among the local and global traffic space. Our experiments indicate that the proposed transformer-based architecture implement a highly accurate traffic prediction model while reducing the training time. Compared to the state-of-the-art methods, the proposed one achieves high performance in terms of the mean absolute error, root mean square error, R-squared, and efficiency in training time.

The Space–Air–Ground Integrated Network (SAGIN) enables seamless connectivity across satellite, aerial, and terrestrial nodes. However, its heterogeneous architecture, resource-constrained nodes, and dynamic link conditions pose significant challenges for deploying deep learning models-particularly for detecting AI-generated artistic content. These challenges include limited on-board computational capacity, fluctuating bandwidth, and the requirement for fine-grained visual-semantic reasoning under weak supervision. To overcome these limitations, we propose the Weakly-Aligned Region-Language Transformer (WARL-Transformer), a novel framework designed for robust AI-generated content detection under realistic SAGIN constraints. WARL-Transformer incorporates: (1) A vision-language alignment mechanism that integrates local visual features with high-level semantic cues derived from textual descriptions, and (2) a weakly supervised local feature alignment strategy that learns region-language correspondences without relying on costly fine-grained annotations. Laboratory-based SAGIN emulation experiments further verify that WARL-Transformer maintains high detection accuracy across diverse artistic styles while preserving robustness against network interference. In particular, WARL-Transformer achieves an F1-score of 96.78%, outperforming the baseline by +0.63 percentage points, and even under bandwidth-constrained SAGIN emulation still reaches 99.7% of the full-model F1, demonstrating strong robustness. This work establishes a foundation for reliable AI-generated content detection in resource-limited SAGIN settings, bridging the gap between visual content authentication and practical network-driven constraints.

Internet of Vehicles (IoV) networks face significant security and privacy challenges due to dynamic topologies and high mobility, exposing them to threats like unauthorized tracking and data tampering. This study aims to develop a robust, privacy-preserving authentication protocol for IoV using blockchain technology. We propose a novel approach integrating a Cascaded and Dilated Residual Recurrent Neural Network (CD-RRNN) for malicious attack detection, blockchain for secure data storage, and an Optimized Key in El-Gamal Encryption (OKEE) with keys selected via Modified Manta-Ray Foraging Optimization (MMRFO). Results demonstrate a 94.34% accuracy in attack detection and a 28.57% reduction in decryption time compared with baselines, validated against state-of-the-art methods. This protocol enhances IoV security, privacy, and scalability, offering a practical solution for smart transportation systems. The rapid expansion of the IoV has introduced significant challenges related to security, privacy, and efficient data management. Traditional centralized architectures struggle with scalability and vulnerability to cyber threats. This article proposes a blockchain-based security framework to enhance trust, authentication, and data integrity in IoV networks. The proposed model leverages Cascaded and Dilated Residual Recurrent Neural Network (CD-RRNN) for detecting malicious activities, coupled with El-Gamal encryption optimized using the Modified Manta-Ray Foraging Optimization (MMRFO) algorithm for secure communication. The system effectively balances privacy and traceability in vehicular networks. Performance evaluations demonstrate improved attack detection accuracy, reduced computational overhead, and higher efficiency compared with existing methods. The proposed solution ensures a decentralized, scalable, and robust security model, making it a viable framework for next-generation IoV ecosystems.