摘要:
Vehicular Edge Computing (VEC) is a promising paradigm for autonomous driving. It can reduce delay and energy consumption of tasks. The problem of joint task offloading scheduling and resource allocation in VEC is a challenge issue. In this paper, we investigate the problem of joint task offloading, task scheduling, and resource allocation in VEC, and the fast changing channel between a vehicle and an edge server. A target problem of joint considering task offloading scheduling, resource allocation and time-varying channel in VEC is formulated. The goal is to minimize the delay and energy consumption of tasks to guarantee the Quality of Service (QoS) of VEC. Constraints on the completion time, the energy consumption, and the computing capability are considered for each task. The resulting mixed integer optimization problem is decomposed into a two-layer optimization problem. In the upper layer, we use a Deep Q-Network (DQN) to solve the task offloading scheduling problem. In the lower level, the CPU frequency allocation is determined using the Gradient Descent (GD) method. Numerical results illustrate that the proposed algorithm can minimize the delay and energy consumption of VEC for different network parameter settings.
摘要:
The current greedy iterative pursuit algorithms for sparse channel estimation in an orthogonal frequency division multiplexing (OFDM) system based on compressive sensing (CS) or distributed CS (DCS) have the disadvantages of relying on channel priori information as halting condition and having a low support searching efficiency. Under DCS framework, this letter proposes a unique halting condition for greedy algorithms by exploiting the delay correlation between adjacent symbol channels. Additionally, we present a segmented pruning strategy that supports to select multiple atoms in a single iteration to improve the support searching efficiency. Simulation results show that our algorithm can achieve more robust sparsity-adaptive channel estimation with reduced computational complexity compared to the traditional methods.
通讯机构:
[Zhou, GX ] C;Cent South Univ Forestry & Technol, Coll Comp & Informat Engn, Changsha 410004, Peoples R China.
关键词:
Road crack;Remote sensing image;Pixel-by-pixel detection;Multi-scale features
摘要:
Crack is the main manifestation of road damage, and its further deterioration will affect road traffic. The timely detection of road cracks is of great significance for ensuring road safety. In this work, starting from UAV remote sensing images,a pixel-by-pixel crack detection method named ARD-Unet is proposed based on U-Net combined with Depth Separable Residual Block (DR-Block), Atrous Spatial Pyramid Fusion Attention Module (ASAM) and Receptive Field Block (RFB). We used UAV to construct a remote sensing road crack dataset containing 1046 high-quality images. The proposed method achieves 76.41 % MIOU and 74.24 % F1-Score on the self-made dataset. Finally, we combine ARD-Unet with UAV to build a road crack detection UAV IoT system, which has been tested in practical applications and achieved excellent performance. Experiments show that ARD-Unet is effective for road crack detection in remote sensing images.
摘要:
In lossy, large-scale wireless sensor networks (WSNs), data collection protocols face the challenge of balancing expected data transmission reliability and limited resources of sensors. Packet redundancy strategies have been widely employed to guarantee transmission reliability. However, these schemes often result in energy waste, especially in sink-neared hotspot districts. We propose the APAP to balance the two competing interests in energy-constrained WSNs. APAP mainly consists of two parts: adaptive packet-reproduction routing and active packet-loss mechanism. In the first part, we design a compensation function to compensate for packet losses and employ it in the packet-reproduction routing. Through available network parameters, such a function offers an optimized number of redundant packets. In the second part, we devise a novel distributed packet-loss mechanism to detect and intercept redundant packets in the hotspot area. Hotspot nodes record the simple traffic information in their vicinity on a local micro-table, these nodes transform their mode based on the records to reach a basic consensus with one-hop neighbors, they then collaboratively intercept excess packets from outside, thus the injection traffic can be mitigated. Such an interception area has the potential to be expanded and prolonged as the redundancy becomes more severe. Our mechanism imposes low-complexity requirements on general devices. Moreover, thorough mathematical analyses are offered for the performance of APAP. Simulation results indicate that with high reliability, the maximum node energy consumption is reduced by 44.7% to 66.3% compared to conventional protocols in one round. Besides, the network lifetime is prolonged by 91.1% to 200%.
作者机构:
[Shen, Xingdong; Zhou, Cui; Zhou, C] Cent South Univ Forestry & Technol, Coll Comp & Informat Engn, Changsha 410004, Peoples R China.;[Shen, Xingdong; Zhou, Cui; Zhou, C] Cent South Univ Forestry & Technol, Coll Sci, Changsha 410004, Peoples R China.;[Zhu, Jianjun] Cent South Univ, Sch Geosci & Infophys, Changsha 410083, Peoples R China.
通讯机构:
[Zhou, C ] C;Cent South Univ Forestry & Technol, Coll Comp & Informat Engn, Changsha 410004, Peoples R China.;Cent South Univ Forestry & Technol, Coll Sci, Changsha 410004, Peoples R China.
关键词:
least squares collocation method;systematic error;random error;TanDEM-X DEM;ICESat-2
摘要:
The TanDEM-X Digital Elevation Model (DEM) is limited by the radar side-view imaging mode, which still has gaps and anomalies that directly affect the application potential of the data. Many methods have been used to improve the accuracy of TanDEM-X DEM, but these algorithms primarily focus on eliminating systematic errors trending over a large area in the DEM, rather than random errors. Therefore, this paper presents the least-squares collocation-based error correction algorithm (LSC-TXC) for TanDEM-X DEM, which effectively eliminates both systematic and random errors, to enhance the accuracy of TanDEM-X DEM. The experimental results demonstrate that TanDEM-X DEM corrected by the LSC-TXC algorithm reduces the root mean square error (RMSE) from 6.141 m to 3.851 m, resulting in a significant improvement in accuracy (by 37.3%). Compared to three conventional algorithms, namely Random Forest, Height Difference Fitting Neural Network and Back Propagation in Neural Network, the presented algorithm demonstrates a reduction in the RMSEs of the corrected TanDEM-X DEMs by 6.5%, 7.6%, and 18.1%, respectively. This algorithm provides an efficient tool for correcting DEMs such as TanDEM-X for a wide range of areas.
摘要:
Concrete cracks are one of the most harmful flaws on the road, threatening traffic safety. In this paper, an effective crack segmentation network MOACA-CrackNet that strives to boost both the model generalization rate and segmentation accuracy of crack segmentation is proposed to segment various types of cracks rapidly and accurately in a variety of acquisition conditions. First, a multi-frequency OctaveRes dual encoder is designed to reduce spatial redundancy by sharing information from neighboring locations. Then, an average weight cross-attention mechanism is designed to suppress redundant background information and improve information exchange between frequencies. Finally, depthwise separable convolution is used to reduce the number of parameters. A dataset with a total of 2062 crack images is constructed in this research, MOACA-CrackNet is trained and tested on this dataset. The experimental results show that MOACA-CrackNet has a good segmentation performance for tiny cracks, the F1-score and mIoU reached 89.2% and 91.32%, respectively.
摘要:
Grain yield prediction affects policy making in various aspects such as agricultural production planning, food security assurance, and adjustment of foreign trade. Accurately predicting grain yield is of great significance in ensuring global food security. This paper is based on the MODIS remote sensing image data products from 2010 to 2020, and adds band information such as vegetation index and temperature to form composite remote sensing data as a dataset. Aiming at the lack of models for large-scale forecasting and the need for human intervention in traditional models, this paper proposes a grain production estimation model based on deep learning. First, image cropping and yield mapping techniques are used to process the data to generate training samples. Then the channel and spatial attention mechanism (convolutional block attention module, CBAM) is added to extract spatial information in different remote sensing bands to improve the efficiency of the model. Long short-term memory (LSTM) neural networks are added to obtain feature information in the time dimension. Finally, a national-scale grain yield prediction model is constructed. After the study, it was found that the LSTM model using a combination of multi-source satellite images and an attention mechanism can effectively predict grain yield in China. Furthermore, the proposed model was tested on data from 2018 to 2020 showing an average R-2 of 0.940 and an average RMSE of 80,020 tons, indicating that it can predict Chinese grain yield better. The model proposed in this paper extracts grain yield information directly from the composite remote sensing data, and solves the problem of small-scale research and imprecise yield prediction in an end-to-end manner.
通讯机构:
[Jizheng Yi] C;College of Computer and Information Engineering, Central South University of Forestry and Technology, Changsha, China<&wdkj&>Institute of Artificial Intelligence Application, Central South University of Forestry and Technology, Changsha, China
摘要:
As the center of the financial market, the stock market is popular with the public attention of investors. It is of great significance for investors that an effective analytic method of stock public opinion is proposed. As the main communication platform, the forum not only provides the investors with investment information but also comments related to the stock market. In view of the defects of text emotions and investment problems, this paper proposes a framework based on web crawler and deep learning technologies including one-dimensional convolutional neural networks (1DCNN) and long short-term memory (LSTM), to evaluate the stock market volatility. Among them, the extracted features include not only the stock price but also the text information. Firstly, we develop the crawler technology to grab large-scale text data from the internet and they are manually labeled their emotions by analyzing the relevant financial knowledge. Secondly, as the character-level text classification method, the 1DCNN is designed for text sentiment classification to detect the reliability of text annotation. Finally, considering the time sequence of price and the continuity of post influence, the emotional and technical features are combined to estimate the fluctuation of the stock market in different industries by the LSTM model. We test four evaluation indexes, the classification accuracy of the model is 74.38%, the accuracy rate is 76.83%, the recall rate is 70%, and the F1 value is 72.8%. The results show that the combination of characteristics of internet public opinion more effectively evaluates the changes in the stock market.
摘要:
In this paper, a wide supercontinuum of multi-order orbital angular momentum (OAM) modes is generated theoretically in the mid-infrared region based on a ring-core As2S3-based photonic crystal fiber (PCF). This PCF supports six OAM modes with two orders and exhibits an ultra-high effective refractive index difference of 10(-2) between the vortex modes. After customizing the nonlinear coefficient and dispersion, all the supported OAM modes can be used for wide-band supercontinuum (>3100 nm) generation at low input pulse power of 1000 W. In particular, the supercontinuum of OAM(1,1 )mode covers a wavelength range from 2860 nm to 6120 nm at-40 dB level. This work has great potentials in the OAM-based mid-infrared source in optical communication and sensor systems.
期刊:
IEEE Transactions on Cybernetics,2023年54(5):3286-3298 ISSN:2168-2267
作者机构:
[Lv, Mingjie; Yang, Chunhua; Fan, Maosen; Liu, Shengyu; Huang, Keke; Liu, Gengchen; Sun, Bei] Cent South Univ, Sch Automat, Changsha 410083, Hunan, Peoples R China.;[Sun, Bei] Peng Cheng Lab, Ind Intelligence Basic Res Studio, Shenzhen 518000, Guangdong, Peoples R China.;[He, Mingfang] Cent South Univ Forestry & Technol, Sch Comp & Informat Engn, Changsha 410004, Peoples R China.
关键词:
Estimation;Logic gates;Observers;Inductors;Testing;Computational modeling;Production;Dynamics learning;long short-term memory;multirate estimation;quality-related indices (QRIs)
摘要:
In this study, we propose a dynamics-learning multirate estimation approach to perceive the quality-related indices (QRIs) of the feeding solution of a unit process. A quality-related index for estimation is an intermediate technical indicator between a unit process and a proceeding unit process; hence, the estimation problem is formulated as a two-stage estimation problem utilizing the production data of both unit processes. Dynamics-learning bidirectional long short-term memory (BiLSTM) with different inputs for the forward and backward layers is proposed to manage the input data from the different unit processes. In the dynamics-learning BiLSTM, a cycle control gate is added in the memory cell to learn the dynamics of the QRIs, thereby enabling a high-rate estimation under multirate conditions. A Bayesian estimation model is then combined with the dynamics-learning BiLSTM model to manage the process delay. Ablation and comparative experiments are conducted to evaluate the feasibility and effectiveness of the proposed estimation approach. The experimental results illustrate the performance and high-rate estimation ability of the proposed approach.
摘要:
【目的】森林火灾识别是避免森林火灾大面积蔓延的一项重要研究。随着深度学习的快速发展,基于卷积神经网络的模型因其在图像识别领域的优异表现,被广泛应用到森林火灾识别任务当中。然而,基于卷积神经网络的方法通常在标签数据不充分时...展开更多 【目的】森林火灾识别是避免森林火灾大面积蔓延的一项重要研究。随着深度学习的快速发展,基于卷积神经网络的模型因其在图像识别领域的优异表现,被广泛应用到森林火灾识别任务当中。然而,基于卷积神经网络的方法通常在标签数据不充分时,难以取得令人满意的森林火灾识别结果。【方法】本研究提出了一种基于视觉变换网络的自监督森林火灾识别模型(Self supervised forest fire identification model based on visual transformation network),来提高模型在标签稀缺情况下的森林火灾识别精度。具体来说,该模型采用视觉变换网络作为主干网络,通过视觉变换网络中的多头自注意力机制来捕获森林火灾图像的全局信息特征。并且引入自监督学习中的图像重建任务来辅助模型训练,从而减少模型对标签数据的依赖。模型通过对掩盖图像的特征恢复和重建学习相关语义信息。同时,本研究还提出了一种基于傅里叶低频混合变换的数据增强方法来提高模型的泛化性和鲁棒性。【结果】通过开展详细的试验来验证模型的有效性,结果表明,与其他常见的网络模型相比,FFDM模型在森林火灾识别任务中取得了最佳的识别效果,其识别准确率为89.51%,比VGG16网络高13.7%,比ResNet50网络高8.2%,比InceptionV3网络高7.2%。【结论】通过自监督学习辅助模型训练的方法,FFDM模型即使在标签稀缺下依然可以取得不错的森林火灾识别效果。收起
通讯机构:
[Li, XY ] C;Cent South Univ Forestry & Technol, Sch Comp & Informat Engn, Hunan, Peoples R China.
关键词:
Vehicle Suspension System;Fractional-Order System;Asymptotic Stability;Stabilization Control
摘要:
Our study focuses on the analysis of asymptotic stability and the development of asymptotic stabilization control methods for fractional vehicle suspension systems (FVSS). We begin by constructing a mathematical model for FVSS using the state-space equations of Caputo fractional calculus. Initially, we utilize the fractional Routh-Hurwitz criterion to derive the necessary conditions for asymptotic stability and instability in the open-loop system of the fractional vehicle suspension. Subsequently, we propose a novel control strategy for FVSS and establish an associated asymptotic stabilization criterion by combining a new vector Lyapunov function with the M-matrix method. Moreover, we extend the fractional-order vehicle suspension model to include time delay resulting from the interactions between different variables in the real system, thus creating a FVSS with time delay. Based on the vector Lyapunov function, M-matrix measure, and Razumikhin interpretation, we develop a control strategy specifically tailored for FVSSs with time delay. Lastly, we compare two numerical simulations of the FVSS, one with time delay and one without, to demonstrate the accuracy, effectiveness, and applicability of the proposed method presented in our paper.