Concurrency and Computation-Practice & Experience最新文献

In this work, privacy-preserving distributed deep learning (PPDDL) is re-visited with a specific application to diagnosing long-term illness like diabetic retinopathy. In order to protect the privacy of participants datasets, a multi-key PPDDL solution is proposed which is robust against collusion attacks and is also post-quantum robust. Additionally, the PPDDL solution provides robust network security in terms of integrity of transmitted ciphertexts and keys, forward secrecy, and prevention of man-in-the-middle attacks and is extensively verified using Verifpal. Proposed solution is evaluated on retina image datasets to detect diabetic retinopathy, with deep learning accuracy results of 96.30%, 96.21% and 96.20% for DDL, DDL + SINGLE and DDL + MULTI scenarios respectively. Results from our simulation indicate that accuracy of the PPDDL is maintained while protecting the privacy of the datasets of participants. Our proposed solution is also efficient in terms of the communication and run-time costs.

In previous studies on facial video depression recognition, although convolutional neural network (CNN) has become a mainstream method, its performance still has room for improvement due to the insufficient extraction of global and local information and the neglect of the correlation of temporal and spatial information. This paper proposes a novel dual-task enhanced global–local temporal–spatial network (DTE-GLTS) to enhance the extraction capability of global and local features and deepen the analysis of temporal–spatial information correlation. We design a dual-task learning mode that utilizes the data-efficient image transformer (Deit) as the main body to learn the global features of video sequences and guides Deit to learn local features with the pre-trained temporal–spatial fusion network (TSF). In addition, we propose the TSF mechanism to more effectively fuse temporal–spatial information in video sequences, strengthen the correlation between frames and pixels, and embed it in Resnet to form the TSF network. To the best of our knowledge, this is the first application of Deit and dual-task learning mode in the field of facial video depression recognition. The experimental results on AVEC 2013 and AVEC 2014 show that our method achieves competitive performance, with mean absolute error/root mean square error (MAE/RMSE) scores of 6.06/7.73 and 5.91/7.68, respectively, while significantly reducing the number of parameters.

This paper develops a new image encryption algorithm based on a novel two-dimensional chaotic map and bit-level dynamic scrambling. First, multiple one-dimensional chaotic maps are coupled to construct a novel two dimensions Logistic-Sine-coupling chaos map (2D-LSCCM). The performance analysis shows that the 2D-LSCCM has more complex chaotic characteristics and wider chaotic range than many extant 2D chaos maps. Second, original image matrix combines with hash algorithm SHA-256 to generate a hash value. The initial values of 2D-LSCCM are generated based on the hash value. Third, the original image matrix is divided into multiple sub-matrices by wavelet transform, followed by scrambling by an improved Knuth shuffle algorithm. Fourth, the scrambled multiple sub-matrices are stitched into an image matrix of $��M�\times �N�\times �3�$ and converted into a binary matrix. The chaotic sequence generated by 2D-LSCCM is introduced as a control sequence to control the bit-level scrambling of pixel points, which realizes the bit-level dynamic scrambling. Finally, the diffusion operation is performed by parameter par and chaotic sequence to obtain the final encrypted image. The algorithm security analysis and simulation examples demonstrate the effectiveness of the proposed encryption scheme.

Hadoop distributed file system (HDFS) performs well when storing and managing large files. However, its performance significantly decreases when dealing with massive small files. In response to this problem, a novel archive-based solution is proposed. The archive refers to merging multiple small files into larger data files, which can effectively reduce the memory usage of the NameNode. The current archive-based solutions have the disadvantages of long access time, long archive construction time, and no support for storage, updating and deleting small files in the archive system. Our method utilizes a dynamic hash function to distribute the metadata of small files across multiple metadata files. We construct a primary index that combines dynamic and static indexes for these metadata files. Regarding data files, include some read-only files and one readable–writable file. A small file's contents are written into a readable and writable file. Upon reaching a predetermined threshold, the readable–writable file transitions into read-only status, with a fresh readable–writable file replacing it. Experimental results show that the scheme improves the efficiency of archive access and archive creation and is more efficient than the original HDFS storage and update efficiency.

The expanding popularity of Voice over WiFi (VoWiFi) necessitates a concerted effort to identify novel ways to increase VoWiFi cell capacity. The primary objective of this study is to increase the capacity of VoWiFi cells by means of frame aggregation of aggregate MAC protocol data unit (A-MPDU) for variable bit rate (VBR) traffic. Taking into account Arbitration Inter-frame Spacing (AIFS), Compressed RTP (cRTP) and A-MPDU frames, we devised a formula to calculate an approximate number of concurrent VoWiFi users that can coexist with no detriment to the quality-of-service (QoS) of existing VoWiFi calls over the Wireless Fidelity (WiFi) standards. Here, we used AIFS to determine the channel's health before sending Voice over WiFi data and Short Inter-frame Spacing (SIFS) to transfer frames such as Request To Send (RTS)/Clear To Send (CTS) and Acknowledgement (ACK). We have used our suggested model to analyse the capacity of VoWiFi cells in IEEE 802.11b/g/n/ac/ax/be Wireless Local Area Network (WLAN)s with VBR traffic utilising DCF Inter-frame Spacing (DIFS) and AIFS. For IEEE 802.11b/g/n/ac/ax/be, we also determined the most number of MAC protocol data unit (MPDU)s that may be combined into a single A-MPDU. We have also studied the impact of voice packet retransmission on the cell capacity of a WLAN standard that offers VoWiFi service while taking A-MPDU method into account. We have compared the results gained using IEEE 802.11be with earlier WLAN standards like IEEE 802.11b/g/n/ac/ax considering the constant bit rate (CBR) and VBR traffics.

UAVs are emerging as a critical asset in the field of data collection from extensive wireless sensor networks (WSNs) on a large scale. UAVs can be used to deploy energy-efficient nodes or recharge nodes, but it should not compromise the network's coverage and connectivity. This paper proposes a comprehensive approach to optimize UAV trajectories within large-scale WSNs, utilizing Multi-Objective Reinforcement Learning (MORL) to balance critical objectives such as coverage, connectivity, and energy efficiency. This research investigates the configuration of a Wireless Sensor Network (WSN) assisted by a pen_spark UAV. In this network, Cluster Heads (CHs) act as central points for collecting data from their assigned sensor nodes. A predefined path is established for the UAV to efficiently gather data from these CHs. The Hybrid Threshold-sensitive Energy Efficient Network (Hy-TEEN) encompasses sophisticated algorithms for CH selection, dynamic A* for 3D trajectory planning and leverages reinforcement learning for multi-objective optimization. The experimental results and analysis demonstrate the effectiveness and efficiency of the proposed approach in improving UAV performance and energy efficiency. The results demonstrate that the proposed methodology's trajectories are capable of achieving a time savings of 3.52% in mission completion when contrasted with conventional baseline methods.

The cloud computing environment is subject to unprecedented cyber-attacks as its infrastructure and protocols may contain vulnerabilities and bugs. Among these, Distributed Denial of Service (DDoS) is chosen by most cyber extortionists, creating unusual traffic that drains cloud resources, making them inaccessible to customers and end users. Hence, security solutions to combat this attack are in high demand. The existing DDoS detection techniques in literature have many drawbacks, such as overfitting, delay in detection, low detection accuracy for attacks that target multiple victims, and high False Positive Rate (FPR). In this proposed study, an Artificial Neural Network (ANN) based hybrid GBS (Grey Wolf Optimizer (GWO) + Back Propagation Network (BPN) + Self Organizing Map (SOM)) Intrusion Detection System (IDS) is proposed for intrusion detection in the cloud computing environment. The base classifier, BPN, was chosen for our research after evaluating the performance of a comprehensive set of neural network algorithms on the standard benchmark UNSW-NS 15 dataset. BPN intrusion detection performance is further enhanced by combining it with SOM and GWO. Hybrid Feature Selection (FS) is made using a correlation-based approach and Stratified 10-fold cross-validation (STCV) ranking based on Weight matrix value (W). These selected features are further fine-tuned using metaheuristic GWO hyperparameter tuning based on a fitness function. The proposed IDS technique is validated using the standard benchmark UNSW-NS 15 dataset, which consists of 1,75,341 and 82,332 attack cases in the training and testing datasets. This study's findings demonstrate that the proposed ANN-based hybrid GBS IDS model outperforms other existing IDS models with a higher intrusion detection accuracy of 99.40%, fewer false alarms (0.00389), less error rate (0.001), and faster prediction time (0.29 ns).