The joint use of multi-source remote sensing data for earth observation missions has drawn much attention. Although the fusion of several data sources can improve the accuracy of land-cover identification, many technical obstacles exist, such as disparate data structures, irrelevant physical characteristics, and a lack of training data. In this article, a novel dual-branch method, consisting of a hierarchical convolutional neural network (CNN) and a transformer network, is proposed for fusing multi-source heterogeneous information and improving joint classification performance. First, by combining CNN with a transformer, the proposed dual-branch network can significantly capture and learn spectral-spatial features from hyperspectral image data and elevation features from light detection and ranging (LiDAR) data. Then, to fuse these two sets of data features, a cross-token attention fusion encoder is designed in a specialty. The well-designed deep hierarchical architecture takes full advantage of the powerful spatial context information extraction ability of CNN and the strong long-range dependency modeling ability of the transformer network based on self-attention mechanism. Four standard datasets are used in experiments to verify the effectiveness of the approach. The experimental results reveal that the proposed framework can perform noticeably better than state-of-the-art methods. The source code of the proposed method will be available publicly at https://github.com/zgr6010/Fusion_HCT.git.

Mobile edge computing (MEC) has become a promising technique to accommodate demands of resource-constrained mobile devices by offloading the task onto edge clouds nearby. However, most existing works only focus on whether to offload or where to offload the task but ignore the motivations of edge clouds to offer service. In this paper, we first propose an incentive mechanism in a non-competitive environment. We utilize market-based profit maximization pricing model to establish the relationship between the resources provided by edge clouds and the price charged to mobile devices. By solving the optimization problem, we provide a reasonable pricing strategy to not only ensure the profit of resource providers but guarantee the quality of experience (QoE) of mobile devices. Furthermore, we design an online profit maximization multi-round auction (PMMRA) mechanism for the resource trading between edge clouds as sellers and mobile devices as buyers in a competitive environment. The mechanism can effectively determine the price paid by buyers to use the resources provided by sellers and make the corresponding match between edge clouds and mobile devices. Finally, numerical results show that proposed mechanism outperforms other existing algorithms in maximizing the profit of edge clouds.

Network traffic forecasting is essential for efficient network management and planning. Accurate long-term forecasting models are also essential for proactive control of upcoming congestion events. Due to the complex spatial-temporal dependencies between traffic flows, traditional time series forecasting models are often unable to fully extract the spatial-temporal characteristics between the traffic flows. To address this issue, we propose a novel dual-channel based graph convolutional network (DC-STGCN) model. The proposed model consists of two temporal components that characterize the daily and weekly correlation of the network traffic. Each of these two components contains a spatial-temporal characteristics extraction module consisting of a dual-channel graph convolutional network (DCGCN) and a gated recurrent unit (GRU). The DCGCN further consists of an adjacency feature extraction module (AGCN) and a correlation feature extraction module (PGCN) to capture the connectivity between nodes and the proximity correlation, respectively. The GRU further extracts the temporal characteristics of the traffic. The experimental results based on real network data sets show that the prediction accuracy of the DC-STGCN model overperforms the existing baseline and is capable of making long-term predictions.

With the rapid expansion of data, the problem of data imbalance has become increasingly prominent in the fields of medical treatment, finance, network, etc. And it is typically solved using the oversampling method. However, most existing oversampling methods randomly sample or sample only for a particular area, which affects the classification results. To solve the above limitations, this study proposes an imbalanced data oversampling method, SD-KMSMOTE, based on the spatial distribution of minority samples. A filter noise pre-treatment is added, the category information of the near-neighbouring samples is considered, and the existing minority class sample noise is removed. These conditions lead to the design of a new sample synthesis method, and the rules for calculating the weight values are constructed on this basis. The spatial distribution of minority class samples is considered comprehensively; they are clustered, and the sub-clusters that contain useful information are assigned larger weight values and more synthetic sample numbers. The experimental results show that the experimental results outperform existing methods in terms of precision, recall, F1 score, G-mean, and area under the curve values when the proposed method is used to expand the imbalanced dataset in the field of medicine and other fields.

Satellite edge computing can provide ubiquitous and reliable connectivity to remote or disaster area networks that are difficult to serve. However, due to the explosive growth of Internet-of-Things (IoT) data traffic, satellite edge services alone make it difficult to meet the latency and energy demands of abundant IoT devices. Edge learning (EL), combined with edge computing and machine learning, is expected to be the key to solving this problem. This article constructs a Satellite-assisted IoT network model consisting of terminals, satellites, and a cloud center. Terminals can offload tasks according to actual needs for load balancing. An EL approach using cloud edge collaboration is proposed. Then, for concurrent random tasks with different service demands, a minimization problem for the weighted sum of system latency and energy consumption is formulated. Finally, a model-assisted two-tier reinforcement learning (MTRL) optimization algorithm for task offloading decisions and resource allocation is proposed to solve this problem. Simulation results show that the proposed algorithm has a good performance in convergence. The system performance is better than the existing algorithms, which can lead to a 27.5% reduction in the system cost. Furthermore, the proposed algorithm can lead to a smaller increase in system cost as the number of IoT devices increases.

Convolutional neural network (CNN)-based and Transformer-based methods for hyperspectral image (HSI) classification have rapidly advanced due to their unique characterization capabilities. However, the fixed kernel sizes in convolutional layers limit the comprehensive utilization of multi-scale features in HSI land cover analysis, while the Transformer’s multi-head self-attention (MHSA) mechanism faces challenges in effectively encoding feature information across various dimensions. To tackle this issue, this article introduces an HSI classification method, based on multi-scale convolutional features and multi-attention mechanisms (i.e., MSCF-MAM). Firstly, the model employs a multi-scale convolutional module to capture features across different scales in HSIs. Secondly, to enhance the integration of local and global channel features and establish long-range dependencies, a feature enhancement module based on pyramid squeeze attention (PSA) is employed. Lastly, the model leverages a classical Transformer Encoder (TE) and linear layers to encode and classify the transformed spatial–spectral features. The proposed method is evaluated on three publicly available datasets—Salina Valley (SV), WHU-Hi-HanChuan (HC), and WHU-Hi-HongHu (HH). Extensive experimental results have demonstrated that the MSCF-MAM method outperforms several representative methods in terms of classification performance.