arXiv subjects

Xiong Wei

Publications and source records attributed to Xiong Wei.

3 recordsLinked to original sources

arXiv2024

Joint Load and Capacity Scheduling for Flexible Radio Resource Management of High-Throughput Satellites

This work first explores using flexible beam-user mapping to optimize the beam service range and beam position, in order to adapt the non-uniform traffic demand to offer in high-throughput satellite (HTS) systems. Second, on this basis, the joint flexible bandwidth allocation is adopted to adapt the offer to demand at the same time. This strategy allows both beam capacity and load to be adjusted to cope with the traffic demand. The new information generated during the load transfer process of flexible beam-user mapping can guide the direction of beam optimization. Then, the proposed strategies are tested against joint power-bandwidth allocation and joint optimization of bandwidth and beam-user mapping under different traffic profiles. Numerical results are obtained for various non-uniform traffic distributions to evaluate the performance of the solutions. Results show that flexible joint load and capacity scheduling are superior to other strategies in terms of demand satisfaction with acceptable complexity. Our source code along with results are available at crystal-zwz/HTS_RRM_Joint-Load-and-Capacity-Scheduling (github.com).

eess.SY↗

arXiv2018

A 2.5D Cascaded Convolutional Neural Network with Temporal Information for Automatic Mitotic Cell Detection in 4D Microscopic Images

In recent years, intravital skin imaging has been increasingly used in mammalian skin research to investigate cell behaviors. A fundamental step of the investigation is mitotic cell (cell division) detection. Because of the complex backgrounds (normal cells), the majority of the existing methods cause several false positives. In this paper, we proposed a 2.5D cascaded end-to-end convolutional neural network (CasDetNet) with temporal information to accurately detect automatic mitotic cell in 4D microscopic images with few training data. The CasDetNet consists of two 2.5D networks. The first one is used for detecting candidate cells with only volume information and the second one, containing temporal information, for reducing false positive and adding mitotic cells that were missed in the first step. The experimental results show that our CasDetNet can achieve higher precision and recall compared to other state-of-the-art methods.

cs.CV↗

arXiv2018

Depth Not Needed - An Evaluation of RGB-D Feature Encodings for Off-Road Scene Understanding by Convolutional Neural Network

Scene understanding for autonomous vehicles is a challenging computer vision task, with recent advances in convolutional neural networks (CNNs) achieving results that notably surpass prior traditional feature driven approaches. However, limited work investigates the application of such methods either within the highly unstructured off-road environment or to RGBD input data. In this work, we take an existing CNN architecture designed to perform semantic segmentation of RGB images of urban road scenes, then adapt and retrain it to perform the same task with multichannel RGBD images obtained under a range of challenging off-road conditions. We compare two different stereo matching algorithms and five different methods of encoding depth information, including disparity, local normal orientation and HHA (horizontal disparity, height above ground plane, angle with gravity), to create a total of ten experimental variations of our dataset, each of which is used to train and test a CNN so that classification performance can be evaluated against a CNN trained using standard RGB input.

cs.CV↗