Along with the progression of Internet of Things (IoT) technology, network terminals are becoming continuously more intelligent. IoT has been widely applied in various scenarios, including urban infrastructure, transportation, industry, personal life, and other socio-economic fields. The introduction of deep learning has brought new security challenges, like an increment in abnormal traffic, which threatens network security. Insufficient feature extraction leads to less accurate classification results. In abnormal traffic detection, the data of network traffic is high-dimensional and complex. This data not only increases the computational burden of model training but also makes information extraction more difficult. To address these issues, this paper proposes an MD-MRD-ResNeXt model for abnormal network traffic detection. To fully utilize the multi-scale information in network traffic, a Multi-scale Dilated feature extraction (MD) block is introduced. This module can effectively understand and process information at various scales and uses dilated convolution technology to significantly broaden the model’s receptive field. The proposed Max-feature-map Residual with Dual-channel pooling (MRD) block integrates the maximum feature map with the residual block. This module ensures the model focuses on key information, thereby optimizing computational efficiency and reducing unnecessary information redundancy. Experimental results show that compared to the latest methods, the proposed abnormal traffic detection model improves accuracy by about 2%.

Crop classification, planting areas estimation and water demand/storage prediction are critical for optimizing water resource management in agriculture. In the water management based on sky-ground collaboration, heterogeneous data from various devices such as satellite, airborne remote sensing and ground observations, are closely depended. In general, tasks required by users always contain several dependent subtasks, and tasks are required to be finished before deadline. In this paper, we introduce a smart scheduling based on graph attention network and meta-learning for tasks with deadlines, which minimizes makespan by the balance between the data transmission time and computation time of subtasks. An enhanced multi-head attention mechanism in graph attention networks is designed to extract associations among heterogeneous data. Additionally, an exponentially smoothed meta-learning approach is designed to optimize parameters of strategies. Compared to existing deep learning-based scheduling algorithms, the proposed strategy improves the average proportion of tasks completed before deadlines by 7.52%.