In this article, a system for automatic recognition, including detection and classification, of emergency vehicles based on images from video recorders is proposed. The system utilizes YOLO v8 artificial neural networ...
详细信息
ISBN:
(数字)9788362065486
ISBN:
(纸本)9798350373806
In this article, a system for automatic recognition, including detection and classification, of emergency vehicles based on images from video recorders is proposed. The system utilizes YOLO v8 artificial neural network. Recognition accuracy tests were conducted on a specially prepared database of frames from real recordings. It includes marked and unmarked police vehicles (which are very difficult to distinguish and typically not included in automatic recognition systems), fire, ambulance, military vehicles and other reference images. The experiments concerned the recognition of vehicle types and the state of emergency lighting. The highest value of total (for all classes) F1 measure is over 86%. In selected classes the F1 reaches 91%, while precision is 94% and sensitivity is 89%. They can be considered as satisfactory results, what indicates the possibility of using the presented system in practice.
Integrated sensing and communication (ISAC) system has been emerged as a crucial paradigm for addressing the growing demand of emerging wireless applications that require both ultra-reliable data transmission and high...
详细信息
ISBN:
(数字)9798350374216
ISBN:
(纸本)9798350374223
Integrated sensing and communication (ISAC) system has been emerged as a crucial paradigm for addressing the growing demand of emerging wireless applications that require both ultra-reliable data transmission and high-precision sensing. However, due to the limited computation capability of ISAC devices, the large amount of collected sensing data is difficult to be timely processed. In this paper, we consider that the ISAC device can offload the sensing data to a group edge helper nodes via multi-access edge computing to improve the efficiency of sensing data processing. We propose a multi-access edge computing empowered ISAC with hybrid active and passive sensing, in which the ISAC device can perform passive sensing through the sensing reflected signal from the edge helper nodes while performing active sensing. To investigate this problem, we formulate a joint optimization of the transmit beamforming for active sensing, passive sensing and offloading, as well as the computation rates of both the ISAC device and the edge helper nodes, with the objective of maximizing the total computation rates for the sensing data. Despite the non-convexity of the formulated problem, we propose an efficient algorithm to obtain its solutions. Simulation results validate the performance advantages of our multi-access edge computing empowered integrated hybrid sensing and communication.
In this paper, a machine learning (ML)-based channel allocation algorithm is proposed to form a secure communication zone in indoor visible light communication (VLC) systems. The algorithm first employs the probabilis...
详细信息
ISBN:
(数字)9798350348743
ISBN:
(纸本)9798350348750
In this paper, a machine learning (ML)-based channel allocation algorithm is proposed to form a secure communication zone in indoor visible light communication (VLC) systems. The algorithm first employs the probabilistic neural network (PNN), which classifies the VLC transmitter (Tx) based on its proximity to the user's location. Subsequently, the selected Tx is used to establish a point-to-point channel allocation, hence forming a closed-access zone within a certain effective communication range. Through numerical simulations, it is observed that the single Tx-based VLC transmission confines the legitimate user in a pre-defined trust boundary for a secure transmission.
Estimating the direction-of-arrival (DOA) is a crucial problem in most array signalprocessing applications, including wireless communication, radar, sonar, astronomical observation, and acoustics. The traditional dir...
详细信息
ISBN:
(数字)9798331517786
ISBN:
(纸本)9798331517793
Estimating the direction-of-arrival (DOA) is a crucial problem in most array signalprocessing applications, including wireless communication, radar, sonar, astronomical observation, and acoustics. The traditional direction-of-arrival (DOA) estimate methods, which are based on subspace decomposition, need for the eigenvalue decomposition, resulting in greater computation complexity. Different adaptive algorithms, including fixed-step-size least mean square (FSS-LMS), variable step-size least mean square (VSS-LMS), and bias-compensated LMS (BC-LMS), have recently been developed for the DOA estimation by using the adaptive nulling antenna techniques in an effort to reduce the computational complexity. The aforementioned algorithms are designed upon the minimization of Mean Square Error (MSE), which proves to be effective in the presence of Gaussian noise. However, their performance will degrade and leads to inaccurate DOA estimation when non-Gaussian/impulsive noise is present. In order to improve the DOA estimation performance in the presence of non-Gaussian/impulsive noise environment, we propose a variable-step-size generalized modified Blake-Zisserman (VSS-GMBZ) algorithm in this letter. The VSS-GMBZ is evaluated for various non-Gaussian noise scenarios to determine DOA estimation accuracy in the Matlab environment. Numerical results demonstrate the superiority of VSS-GMBZ over existing methods.
This paper aims at analyzing the use of Gaussian Process Regression (GPR) in working and creating effective time management system. It incorporates the GPR model that helps in predicting the future time usage by disse...
详细信息
ISBN:
(数字)9798350390759
ISBN:
(纸本)9798350390766
This paper aims at analyzing the use of Gaussian Process Regression (GPR) in working and creating effective time management system. It incorporates the GPR model that helps in predicting the future time usage by dissecting the time-study data this assists the users in scheduling. The approach adopted entails harvesting data from automated tools and manual logs, pre-processing the data collected, using k-fold cross-validation in training and validating the model. The forecasting accuracy features a low Mean Absolute Error and a high R-squared value, which signifies that the models perform well in terms of predictive capability. This research proves that GPR can actively help with data on time management and benefit time management systems as a whole. Some of the challenges that faced the research include quality of the data and the computation required to process the data was enhanced by effective preprocessing and refined algorithms respectively.
The expansion of the scope of industrial Internet of Things (IIoT) technology is largely due to the advent of low-orbit satellite Internet (LEO SI). LEO SI uses OFDM technology to provide broadband access and data exc...
详细信息
ISBN:
(数字)9798350349818
ISBN:
(纸本)9798350349825
The expansion of the scope of industrial Internet of Things (IIoT) technology is largely due to the advent of low-orbit satellite Internet (LEO SI). LEO SI uses OFDM technology to provide broadband access and data exchange between subscribers. One of the methods that allows to increase the speed of information transfer in OFDM systems is based on replacing the fast Fourier transform (FFT) with discrete wavelet transforms (DWT). The use of parallel DWT calculations based on modular residue class codes (MCRC) makes it possible to increase the speed of orthogonal signal transformations. This result is achieved due to the parallel and independent execution of arithmetic operations based on MCRC bases. However, failures and malfunctions of the digital signalprocessing computing device may occur during signalprocessing. This leads to errors in the calculation process. When redundancy is introduced, modular codes are able to detect and correct such errors. Thus, the use of MCRC makes it possible to increase the fault tolerance of the OFDM system by fending off the consequences of failures. Therefore, the development of a mathematical model of the OFDM system that fends off the consequences of failures using modular codes is an urgent task.
Video streaming over the Internet has become a prevalent aspect of daily social interactions. As the number of users streaming videos increases, so does the importance of guaranteeing the maximum quality of experience...
详细信息
ISBN:
(数字)9798350384819
ISBN:
(纸本)9798350384826
Video streaming over the Internet has become a prevalent aspect of daily social interactions. As the number of users streaming videos increases, so does the importance of guaranteeing the maximum quality of experience (QoE) to the end users. Multiple QoE metrics, including buffer starvation and video quality, were defined to guide the development of video streaming protocols and adaptive playback mechanisms. Some of these mechanisms include encoding bitrate adaptation and frame prefetching. However, the impact of head of line (HOL) blocking on QoE has not been extensively examined in earlier studies. Therefore, this paper proposes an algorithm to limit HOL blocking at the client side to reduce buffer starvation. Simulation results show that the proposed algorithm reduces the number of starvation instances and increases inter-starvation duration in two out of three tested network conditions.
Feature points based image registration is a technique used in computer vision and image processing to align two or more images by identifying and matching distinctive feature points between them. The purpose is to es...
详细信息
ISBN:
(数字)9798350309249
ISBN:
(纸本)9798350309256
Feature points based image registration is a technique used in computer vision and image processing to align two or more images by identifying and matching distinctive feature points between them. The purpose is to establish a geometrical transformation between two images to be compared that spatially aligns each point from one image to its corresponding of the other image. Recently, local feature key-points such as scale invariant feature transform (SIFT), speeded-up robust features (SURF) and oriented FAST and rotated BRIEF (ORB) are widely used due to their inherent properties such as invariance, changes in illumination, and noise. This study provides an automated feature points based registration technique applied to satellite images. The registration task is solved by an iterative heuristic optimization method namely genetic algorithm (GA) to find the best rigid transformation parameters by minimizing the distance between the two feature point sets. An adaptive GA based on fitness sharing and elitism techniques is performed to enhance the capability of GA. A comparative study is conducted between different extraction feature techniques. Therefore, the results demonstrate the effectiveness of the SIFT algorithm compared to the other competing techniques for registering satellite images.
In line with the needs of life, new solution methods developed against the object recognition problem are gaining importance. In this paper, a literature search was conducted on object recognition studies using ultras...
详细信息
ISBN:
(数字)9798350388961
ISBN:
(纸本)9798350388978
In line with the needs of life, new solution methods developed against the object recognition problem are gaining importance. In this paper, a literature search was conducted on object recognition studies using ultrasonic signals. It is aimed to contribute to the literature by proposing an integrated method that can perform object recognition by passing ultrasonic signal data obtained from different objects through pre-processing, feature extraction and classification processes. As a result of the study, a pre-processing technique was applied to extract the signal envelope by editing the data set obtained by making measurements from objects of different diameters and shapes at a certain angle and distance. The data obtained as a result of preprocessing was passed through feature extraction methods known as waveform shape descriptors. Comparisons were made with the most appropriate hyperparameters of deep learning algorithms to achieve high-performance classification rates and comprehensive object recognition with the developed system. Consequently, high performance classification was achieved with MLP and CNN deep learning models used in the classification stage.
In recent years, the proliferation of synthetic speech has raised concerns regarding its potential misuse for unethical activities including voice impersonation and deep fake generation. Addressing this challenge requ...
详细信息
ISBN:
(数字)9798350365597
ISBN:
(纸本)9798350365603
In recent years, the proliferation of synthetic speech has raised concerns regarding its potential misuse for unethical activities including voice impersonation and deep fake generation. Addressing this challenge requires robust methods for detecting synthetic speech, which often exhibits subtle but discernible differences from natural speech. In this paper, three approaches for synthetic speech detection are proposed, two based on deep neural networks (DNNs), namely multilayer perceptrons (MLPs), convolutional neural networks (CNNs), and one based on an EfficientNetV2 model and transfer learning. The proposed system was trained on the Fake-or-Real (FoR) dataset, comprising utterances generated by some of the latest speech synthesis algorithms, and is able to generalize well on unseen samples generated with algorithms not encountered during training, yielding a validation accuracy of 98.9% and a test accuracy of 83.9%.
暂无评论