
Introduction/Objective: Voice Activity Detection (VAD) plays a vital role in speech processing systems, particularly in noisy environments, where accurately detecting speech segments is challenging. This study aims to improve VAD performance by integrating a Maximum a Posteriori (MAP)-based speech enhancement technique with a Discrete Wavelet Transform (DWT)-based VAD framework. Methods: The proposed system combines a MAP-based speech enhancement module as a preprocessing stage with a DWT-based VAD algorithm. The system is evaluated using the NOIZEUS noisy speech corpus across six real-world noise environments at different signal-to-noise ratio (SNR) levels. Performance is assessed using standard metrics such as SNR improvement, Frame Error Rate (FER), precision, recall, and F1-score. Results: The proposed MAP+DWT VAD framework reduced the average FER from 31.68% to 17.02%, corresponding to a 46.3% reduction. The average precision and F1-score improved from 0.679 to 0.842 and from 0.808 to 0.885, respectively. The average recall decreased from 1.00 to 0.947, indicating a small reduction in missed-speech performance while providing a more balanced overall VAD performance. The MAP-based enhancement achieved mean SNR improvements of 2.03 dB at 10 dB input SNR and 1.22 dB at 15 dB input SNR. Discussion: The results indicate that incorporating a speech enhancement stage prior to VAD substantially improves detection accuracy by reducing noise interference and false positives. While the system performs effectively across various noise types, slight reductions in recall suggest a more conservative classification behavior after enhancement. The approach demonstrates robustness and adaptability in real-world acoustic scenarios. Conclusion: The integration of MAP-based speech enhancement with DWT-based VAD provides a reliable and effective solution for robust speech detection in noisy environments. The proposed framework enhances overall VAD performance and offers potential for real-time speech processing applications.
Introduction/Objective: The Fluctuating Nakagami-m fading is adequate for device-to-device communications, vehicle-to-vehicle communications, and underwater acoustic communications systems. The analysis of energy detection-based spectrum sensing performance for the Fluctuating Nakagami-m fading model is investigated in this paper. Methods: At first, we derive the analytical expressions of the average probability of detection and the Area Under the receiver operating characteristic Curve (AUC) for single-user spectrum sensing over Fluctuating Nakagami-m fading channels. After that, the average detection probability expressions are derived for collaborative spectrum sensing, square-law selection combining diversity and noise power uncertainty. Results: The results are presented for the number of multipath clusters, time-bandwidth product, predetermined threshold, number of collaborating users, shadowing, the number of diversity branches, noise power uncertainty condition, and average Signal-to-Noise Ratio (SNR). The achieved results are compared with the results of Rayleigh and Nakagami-m distributions, which are special cases of the fluctuating Nakagami-m distribution. The results show that the energy detection performance is more sensitive to fading severity and the shadowing parameter. We also present the results under light shadowing and heavy shadowing conditions. Discussion: The results show that fading severity and shadowing parameters have a great impact on detection performance. The detection performance also improves by employing diversity and collaborative users. In collaborative spectrum sensing, the OR-rule shows better energy detection performance than the AND-rule. Also, a degradation in detection performance was observed by increasing the predetermined threshold values. However, it is shown that the detection performance degrades with noise power uncertainty. Conclusion: The presented analytical results are compared with Monte Carlo simulation results to cross-check the validity of derived expressions.
Introduction: The analytical complexity persists due to the κ-μ/shadowed fading, which often restricts its practical applicability in diversity-aided systems and makes the interpretation of the performance metrics more challenging. This complexity becomes more pronounced when the joint effects of diversity, fading, reliability, and latency need to be quantified. materials and methods: The study begins by establishing a unified characterization of the probability density function in the vicinity of the origin for the broad class of κ–μ fading models, encompassing both single- and double-shadowed channel conditions. Building upon this generalized statistical framework, the analysis is subsequently extended to incorporate diversity combining schemes, through which the corresponding probability density functions are systematically derived. These formulations then serve as the foundation for a rigorous asymptotic investigation, enabling insightful evaluation of system behavior in the high signal-to-noise ratio regime, while explicitly quantifying key performance metrics such as diversity order and SNR gain. Method: To address this challenge, this work proposes a unified and simplified asymptotic framework for short-packet transmission over the κ-μ/shadowed fading family, covering both single- and double-shadowed scenarios in diversity-assisted communication. Initially, we derive tractable expressions for the origin probability density function (PDF) of a single-input-single-output system, followed by their evaluation under Maximal Ratio Combining (MRC) and Selection Combining (SC) schemes. Building on this, asymptotic expressions of short-packet communication metrics, including the average block error rate and delay outage rate, are derived. The framework is further extended to provide a unified characterization of the diversity and SNR gains, which offers a direct assessment of the performance benefits offered by diversity combining under generalized fading conditions. results: The obtained output expressions exhibit a notably simple and analytically tractable form, which not only enhances their practical utility but also facilitates deeper theoretical insight. More importantly, these expressions demonstrate a strong asymptotic consistency with the exact formulations in the high transmit power regime, thereby validating the soundness and precision of the derivations. This close agreement under such limiting conditions substantiates the correctness of the analytical framework and confirms that the proposed results reliably capture the system behavior, even when operating in extreme power scenarios. Results: The present analysis provides simple and tractable asymptotic expressions for key performance indicators of finite-block-length transmission for a broad range of κ-μ/shadowed (KMS) fading conditions. The derived expressions accurately capture the high-SNR regime and provide useful insight into the impact of diversity and fading parameters on reliability and latency. The analytical findings are validated by Monte-Carlo simulations and found to be in close agreement, confirming the accuracy of the derived expressions. Discussion: Apart from simplifying analysis, the presented framework provides a unified asymptotic perspective for both single- and double-shadowed fading channels under different diversity combining schemes. It also offers useful insight into the trade-off involving diversity, fading, reliability, and latency. The outcomes demonstrate the role of diversity in mitigating the adverse impact of fading and improving system reliability. Conclusion: In this paper, a unified asymptotic framework is introduced for finite-block-length communications over In this paper, a unified asymptotic framework is introduced for finite-block-length communications over κ−μ single- and double-shadowed fading models. The presented expressions enable an efficient and proficient performance assessment, while quantifying both diversity and SNR gains. These approximations were corroborated through simulations and are expected to be beneficial for the design of future low-latency wireless networks. single- and double-shadowed fading models. The presented expressions enable an efficient and proficient performance assessment, while quantifying both diversity and SNR gains. These approximations were corroborated through simulations and are expected to be beneficial for the design of future low-latency wireless networks.
Introduction: Road anomalies such as potholes, surface cracks, and speed bumps present a persistent hazard to vehicles, motorists, and pedestrians. Their automatic detection is a significant issue because they damage cars, disrupt traffic, accelerate the deterioration of road infrastructure, and reduce driver comfort. Advanced driver assistance systems address this need by allowing autonomous and semi-autonomous vehicles to identify and respond to road anomalies in real time. The present work proposes a deep learning-based smart driver-assistance system for the automated detection of road anomalies. Methods: A 13-layer convolutional neural network (CNN) with 232,034 trainable parameters is used both as an end-to-end classifier and, in a separate hybrid configuration, as a feature extractor whose 512-dimensional output is passed to a light gradient boosting machine, histogram-based gradient boosting, extra trees, random forest, and k-nearest-neighbour classifiers. Images from four public datasets are resized to 256 x 256 x 3, scaled to [0, 1], and class distributions are balanced using the synthetic minority over-sampling technique. Results: The proposed CNN attains 99.00% accuracy, 99.13% precision, and a 98.98% F1 score in the two-class classification of cracks and potholes and 97.67% accuracy, 97.54% precision, and a 97.61% F1 score in the three-class classification of cracks, potholes, and speed bumps. Its computational cost is additionally compared with that of transformer-based classifiers and contemporary convolutional backbones. Discussion: The proposed network reaches this accuracy with 232,034 parameters, which is between one and three orders of magnitude fewer than the established convolutional and transformer backbones evaluated here. This combination of accuracy and low parameter count is the property that matters for an embedded driver assistance unit, where memory and inference latency are constrained. Conclusion: The results indicate that a compact, purpose-designed CNN is a practical basis for onvehicle road anomaly detection. Thus, this approach could contribute to safer travel, reduced vehicle deterioration, and enhanced performance of autonomous vehicles. The evaluation is confined to offline public image datasets, and on-road validation remains to be carried out.
Objective: To design an adaptive millimeter-wave (mmWave) Wireless Power Transfer (WPT) system that dynamically optimizes carrier frequency and beamforming parameters to maintain Unmanned Aerial Vehicle (UAV) operations in Beyond-5G (B5G) networks. Methods: The proposed framework utilizes a reinforcement learning (Q-learning) controller to adapt transmission parameters in real time. The system was evaluated using MATLAB simulations integrating ITU-R standardized atmospheric channel models (accounting for rain, fog, and molecular absorption) across diverse environmental profiles including urban, rural, desert, and subzero settings. Results: The adaptive scheme improved received power by up to 10 dB compared to staticfrequency baselines, yielded an additional 3 to 5 dB gain in urban Non-Line-of-Sight (NLoS) conditions, and demonstrated a 15-20% efficiency increase over traditional sub-6 GHz systems. Discussion: By synchronizing intelligent frequency switching with adaptive beam steering, the system effectively mitigates severe environmental impairments and spatial misalignments caused by UAV mobility. Conclusion: The reinforcement-learning-driven joint optimization framework significantly enhances the robustness, efficiency, and reliability of mmWave WPT, offering a resilient solution to support persistent UAV operations in dynamic conditions.