REV Journal on Electronics and Communications
Not a member yet
230 research outputs found
Sort by
EyeTrackDL: A Robust Deep Learning Framework for Saccade Detection via Simulated Data Augmentation
No full textSaccade detection is a fundamental task in visual behavior analysis and vestibular diagnostics. However, video Head Impulse Test (vHIT) recordings are often noisy, heterogeneous, and affected by class imbalance, particularly for covert saccades. In this paper, we propose EyeTrackDL, a lightweight yet effective deep learning framework based on a multilayer perceptron (MLP) architecture for classifying three types of eye movements: non-saccades, overt saccades, and covert saccades. Input signals are preprocessed using a fourth-order Butterworth filter, and two high-level features (onset time and duration) are extracted per saccade. To address data scarcity and imbalance, we apply SMOTE resampling and incorporate synthetic data generated from a kinematic vestibulo-ocular reflex (VOR) model. The model is evaluated using K-fold cross-validation (K = 2 to 10) on both real and simulated datasets. Results show that EyeTrackDL achieves an average accuracy of up to 96.5% on simulated data and approximately 83% in the real data, with significant improvements in the sensitivity of the covert saccades. Our findings demonstrate the potential of integrating simulation-based augmentation and class balancing for robust saccade detection in clinical environments
Performance Analysis of CR-NOMA Network with Active RIS-Enhanced System
Get PDFThis paper proposes a reconfigurable intelligent surface (RIS)-assisted underlay spectrum sharing system, in which a RIS-assisted secondary network shares the spectrum licensed for a primary network. The secondary network consists of a secondary source (BS), an RIS, and a secondary destination (SD), operating in a Rician fading environment. We study the performance of the secondary network while considering a peak power constraint at the BS and an interference power constraint at the primary receiver (PR). Building upon the SNR statistics, we analyze the outage probability and throughput, deriving novel exact expressions for these performance measures. Finally, we conduct exhaustive Monte-Carlo simulations to confirm the correctness of our theoretical analysis
Hybrid Satellite-Terrestrial Relaying Networks With Imperfect Channel State Information and Directional Antenna: the Dilemma of Facilitating Reliability and Improving Security
Get PDFThe performance of hybrid satellite-terrestrial relaying (HSTR) networks is investigated in this work. Specifically, we examine the trade-off between reliability and security in HSTR networks using two key parameters: outage probability (OP) and intercept probability (IP). Both metrics are derived in closed-form expressions under the assumption of imperfect channel state information (CSI) for the legitimate channels. Additionally, a directional antenna is employed to compensate for the significant path loss caused by the long transmission distance between the satellite and the ground terminal. Numerical computations are provided to validate the accuracy of the derived framework. Furthermore, our findings reveal that increasing the satellite's transmit power and altitude has opposite effects on security and reliability. Specifically, increasing the transmit power enhances system reliability but reduces security. In contrast, a higher satellite altitude decreases reliability but improves security. These findings are further validated through Monte Carlo-based simulations
On Performance Evaluation Of Hybrid Satellite-Terrestrial Relaying Networks With Fountain coding, NOMA and RIS
No full textIn this paper, we analyze outage probability (OP) of a hybrid satellite-terrestrial non-orthogonal multiple access (NOMA) relay system, where Fountain coding is used. In this system, Fountain-encoded packets are transmitted from the satellite to two terrestrial users. To assist the communication between a satellite and users, a reconfigurable intelligent surface (RIS) and a relay node are simultaneously utilized. To employ the advantages of the cooperative scheme, at the relay station, the fountain-encoded packets are decoded and then forwarded both directly without the RIS and indirectly with the RIS to the two users. We analyze the system’s performance through theoretical analysis and propose a closed-form formula for the OP of both users over shadowing and Nakagami fading channels. Monte Carlo simulations confirm the correctness of the theoretical analysis by providing empirical evidence that supports the predicted outcomes under various scenarios and conditions. Additionally, factors influencing OP, such as satellite-terrestrial channel conditions, the number of reflecting elements in the RIS, Fountain coding parameters, and power allocation coefficients, are thoroughly investigated. Our findings demonstrate that integrating NOMA, RIS, and Fountain coding substantially improves system performance. Moreover, the results also reveal that to achieve the specified OP target, the system can dynamically adjust suitable power allocation coefficients for users, the number of elements in the reflector, or appropriate Fountain coding parameters
Development of a Multi-Constraint Loss Function for Image Recovery from Pruned Features
Get PDFExisting image recovery methods have demonstrated that original images can be reconstructed from full features. In this work, we consider a more challenging problem of recovering images from pruned features learned by deep neural networks. Our study addresses this issue by introducing a multi-constraint loss function that integrates L2 distance, sixth-power summation, and total variation regularization to enhance reconstruction quality. The loss function enhances image smoothness and fidelity while ensuring that reconstructed images are encoded as vectors closely aligned with the pruned feature.The proposed loss function enables robust image recovery, preserving key visual features even at high pruning ratios. Additionally, this study investigates the impact of different pruning levels on reconstruction fidelity, highlighting the trade-off between pruning efficiency and recoverability. These findings provide valuable insights into inverse problems in deep learning and image processing, with implications for security risk assessment, efficient model evaluation, and feature redundancy analysis
Improving DOA Estimation Accuracy Using Combination of U-Net Model and MUSIC Algorithm for Uniform Circular Arrays With Inactive Elements
Get PDFIn this study, a novel method is proposed to improve the accuracy of direction of arrival (DOA) estimation for radio signal sources when a uniform circular antenna array (UCA) has inactive elements. Specifically, the full-rank covariance matrix is reconstructed by integrating a U-Net deep learning model with the multiple signal classification (MUSIC) algorithm, even with incomplete array elements. To restore essential correlation information lost due to the inactive elements, a subspace-based full-rank recovery technique is employed. The reconstructed covariance matrix is then utilized by the MUSIC algorithm for accurate DOA estimation. Experimental results demonstrate significant improvements in accuracy, especially under low signal-to-noise ratio (SNR) conditions and with incomplete antenna arrays. Therefore, this approach ensures stable and precise DOA estimation even under non-ideal operating scenarios, offering a practical solution when antenna arrays experience element failures or physical obstructions
Development of Interleaver for the BICM-ID system based on the GBIm
Get PDFIn this study, the Generalized Block Interleaving method (GBIm) is proposed for the Bit-Interleaved Coded Modulation with Iterative Decoding (BICM-ID) system. Generalized Block Interleaving with Almost Regular Permutation (GBI-ARP) and Generalized Block Interleaving with Golden (GBI-Golden) are developed based on GBIm, utilizing component interleaving constructed from algebraic mathematical expressions. With this approach, the proposed interleavers maintain simplicity while ensuring high randomness and large connectivity indices. This enhances system flexibility and reduces memory requirements. Research results demonstrate that the GBI-ARP and GBI-Golden interleavers significantly improve the Bit Error Rate (BER) performance of the BICM-ID system using (8,4,4) Extended Hamming code and 16-QAM modulation, achieving a gain of 0.5 dB in Signal-to-Noise Ratio (SNR) at a BER of 10^−6 compared to a random interleaver, and over 2dB at a BER of 10^−4 compared to traditional Block interleaver and basic Golden interleaver. Furthermore, the proposed interleavers based on the GBIm meet the criteria for complexity, latency, and applicability in next-generation real-time communication systems
Performance Analysis of MIMO Full-Duplex NOMA Networks with Max-Min Relay Selection for Short-Packet Communications
No full textThis paper investigates the performance of a multiple-input multiple-output (MIMO) non-orthogonal multiple access (NOMA) network, enhanced by a pool of full-duplex (FD) relays, for short-packet communications (SPC). We propose a combined framework where the source and destinations are equipped with multiple antennas, and a max-min relay selection criterion is employed to choose the best relay among multiple candidates. This criterion maximizes the end-to-end link quality by selecting the relay that offers the best bottleneck signal-to-interference-plus-noise ratio (SINR). We derive novel, exact analytical expressions for the average block error rate (BLER) for two NOMA users, meticulously accounting for the joint effects of MIMO diversity, relay selection diversity, residual self-interference (RSI) at the FD relay, and imperfect successive interference cancellation (SIC). To facilitate computation, we also provide a highly accurate closed-form approximation of the BLER using Gauss-Chebyshev quadrature. The analysis reveals that the max-min selection strategy effectively overcomes the performance bottleneck of simpler selection schemes, ensuring that the system's diversity order scales with the number of available relays. Numerical results validate our rigorous theoretical analysis, demonstrating the significant reliability gains achieved by the proposed framework. The findings offer crucial insights into the interplay between key system parameters, providing a comprehensive guide for designing robust ultra-reliable low-latency communication (URLLC) systems
Electrocardiogram Based Heartbeat Detection Using Deep Learning
Get PDFCardiovascular diseases remain a leading cause of death worldwide, which results in important requirements for early and accurate arrhythmia diagnosis. This work proposes a novel design of automated heartbeat detection, which consists of a convolutional neural network and three-channel images using the electrocardiogram (ECG) signals. A combination of various preprocessing is applied for the elimination of interferences of the ECG signals such as band-pass filtering and wavelet transform for R-peak identification using a sliding window. Multimodal image fusion method is used to construct three-channel images from different grayscale images, which are transformed from the heartbeats by three transformation techniques namely Gramian angular field, Markov transition field, and Recurrence plot. Grid-search based optimization method in combination with 5-fold cross validation procedure are implemented for selection of the optimal hyper-parameters of the CNN models using the input three-channel images. The proposed algorithm including CNN models and MIF images is estimated the detection performance using 5-fold cross validation, which produces average accuracy of 99.63%, precision of 99.41%, recall of 99.52%, and F1-score of 99.64%. The relatively high performance of the proposed algorithm confirms the effectiveness for the arrhythmia recognition on the ECG signals