计算机科学最新文献

This paper proposes a novel method for network latency estimation. Network latency estimation is a crucial indicator for evaluating network performance, yet accurate estimation of large-scale network latency requires substantial computation time. Therefore, this paper introduces a method capable of enhancing the speed of network latency estimation. The paper represents the data structure of network nodes as matrices and introduces a time dimension to form a tensor model, thereby transforming the entire network latency estimation problem into a tensor completion problem. The main contributions of this paper include: optimizing leveraged sampling for tensors to improve sampling speed, and on this basis, introducing the Qatar Riyal (QR) decomposition of tensors into the Alternating Direction Method of Multipliers (ADMM) framework to accelerate tensor completion; these two components are combined to form a new model called LNLS-TQR. Numerical experimental results demonstrate that this model significantly improves computation speed while maintaining high accuracy.

The ability of orthogonal frequency division multiplexing (OFDM) to counteract frequency-selective fading channels has made it a popular modem technology in contemporary communication systems. But maintaining dependable signaling is still difficult, especially when the signal-to-noise ratio (SNR) is low. In order to increase the dependability of OFDM systems, this study presents an enhanced LSTM-based autoencoder architecture. The suggested autoencoder efficiently utilizes temporal dependencies and reduces the impacts of channel distortion by encoding and decoding OFDM signals utilizing one-hot encoding employing long short-term memory (LSTM) networks. The outcomes of the simulation show notable gains in performance indicators. The average block error rate (BLER) of the suggested model is 0.0150, as opposed to 0.0296 for traditional autoencoders and 0.0886 for convolutional OFDM systems. Comparably, the average packet error rate (PER) is decreased to 0.0017, surpassing convolutional OFDM systems' 0.2260 and traditional autoencoders' 0.0070. These outcomes highlight the LSTM-based autoencoder's efficacy in enhancing OFDM systems' dependability, especially in demanding settings. This study lays the groundwork for employing cutting-edge deep learning methods to create reliable and effective communication systems.

Fog computing extends cloud computing to the edge of the network, bringing processing and storage capabilities closer to end users and Internet of Things (IoT) devices. This paradigm helps to reduce latency, improve response time, and optimize bandwidth usage. In the cloud computing environment, service availability is a criterion for determining user satisfaction, which is strongly influenced by response time and optimal allocation of network resources (communication bandwidth). Service placement in fog computing refers to the process of determining optimal locations for placing services in the network. In this paper, the service placement is done by being aware of the volume of user requests from fog nodes by using neural networks, reinforcement learning, and the improved gray wolf optimization (IGWO) method. Based on the results obtained from simulation, the proposed approach has less response time (between 5% and 21%), more favorable load balance, more utility value (12%) and lower Energy consumption by a minimum of 10% and a maximum of 25%.

Human emotion recognition (HER) has rapidly advanced, with applications in intelligent customer service, adaptive system training, human–robot interaction (HRI), and mental health monitoring. HER’s primary goal is to accurately recognize and classify emotions from digital inputs. Emotion recognition (ER) and feature extraction have long been core elements of HER, with deep neural networks (DNNs), particularly convolutional neural networks (CNNs), playing a critical role due to their superior visual feature extraction capabilities. This study proposes improving HER by integrating EfficientNet with transfer learning (TL) to train CNNs. Initially, an efficient R-CNN accurately recognizes faces in online and offline videos. The ensemble classification model is trained by combining features from four CNN models using feature pooling. The novel VGG-19 block is used to enhance the Faster R-CNN learning block, boosting face recognition efficiency and accuracy. The model benefits from fully connected mean pooling, dense pooling, and global dropout layers, solving the evanescent gradient issue. Tested on CK+, FER-2013, and the custom novel HER dataset (HERD), the approach shows significant accuracy improvements, reaching 89.23% (CK+), 94.36% (FER-2013), and 97.01% (HERD), proving its robustness and effectiveness.

As quantum technology advances, classical digital signatures exhibit vulnerabilities in preserving security properties during the transmission of information. Working toward a reliable communication protocol, we introduce a proxy blind signature scheme to teleport a single particle qubit state with a message to the receiver, employing a three qubit GHZ entangled state. The blindness property is utilized to secure the message information from the proxy signer. A trusted party, Trent, is introduced to supervise the communication process. Alice blinds the original message and sends the Bell measurements with her entangled particle to proxy signer Charlie. After receiving measurements from Alice and Charlie, Bob verifies the proxy blind signature and performs appropriate unitary operations on his particle. Thereafter, Trent verifies the security of the quantum teleportation setup by matching the output data with the original data sent by Alice. Security analysis results prove that the proposed scheme fulfils the basic security necessities, including undeniability, unforgeability, blindness, verifiability, and traceability.