Huawei

About: Huawei is a company organization based out in Shenzhen, China. It is known for research contribution in the topics: Terminal (electronics) & Node (networking). The organization has 41417 authors who have published 44698 publications receiving 343496 citations. The organization is also known as: Huawei Technologies & Huawei Technologies Co., Ltd..

Method, device and system for monitoring network performance

Abstract: A method for monitoring network performance includes: sending correspondences between a remote network element and a plurality of IP addresses thereof from a network management device to a local network element; according to the correspondences, calculating performance parameters between the local IP address and the plurality of remote IP addresses respectively, by the local network element; making a statistics of the calculated performance parameters by the local network element. A network element for monitoring network performance, connected with a network management device and a remote network element, includes: a receiving unit, a performance parameter processing unit, a performance parameter statistics unit, and the transmitting unit. A network system for monitoring network performance is further provided. According to embodiments of this invention, the efficient measurement of network performance including IP QoS between each two MGWs of an IP network, RTP stream bandwidths, and the like, is realized, thus the message bandwidth of performance test can be saved and processing load of the network management device can be reduced.

Decoders Matter for Semantic Segmentation: Data-Dependent Decoding Enables Flexible Feature Aggregation

Abstract: Recent semantic segmentation methods exploit encoder-decoder architectures to produce the desired pixel-wise segmentation prediction. The last layer of the decoders is typically a bilinear upsampling procedure to recover the final pixel-wise prediction. We empirically show that this oversimple and data-independent bilinear upsampling may lead to sub-optimal results. In this work, we propose a data-dependent upsampling (DUpsampling) to replace bilinear, which takes advantages of the redundancy in the label space of semantic segmentation and is able to recover the pixel-wise prediction from low-resolution outputs of CNNs. The main advantage of the new upsampling layer lies in that with a relatively lower-resolution feature map such as $\frac{1}{16}$ or $\frac{1}{32}$ of the input size, we can achieve even better segmentation accuracy, significantly reducing computation complexity. This is made possible by 1) the new upsampling layer's much improved reconstruction capability; and more importantly 2) the DUpsampling based decoder's flexibility in leveraging almost arbitrary combinations of the CNN encoders' features. Experiments demonstrate that our proposed decoder outperforms the state-of-the-art decoder, with only $\sim$20\% of computation. Finally, without any post-processing, the framework equipped with our proposed decoder achieves new state-of-the-art performance on two datasets: 88.1\% mIOU on PASCAL VOC with 30\% computation of the previously best model; and 52.5\% mIOU on PASCAL Context.

Method and apparatus for intercepting performance metric packets for improved security and intrusion detection

Abstract: A method in which a border device (3) of a destination network located outside of a recipient personal computer (2) or network intercepts a performance measurement packet for a specified recipient in order to relieve problems that arise when performance metric packets are interpreted as harmful to a recipient network or server. A border device (3) intercepts the performance metric packet and returns requested information to the sender (1) while masking the source address of the response as the original destination address of the original recipient (2) or the network number of that recipient. The sender (1) of the packet receives ample information on the performance metrics to the perimeter of the recipient (2) for use in its application and the recipient network is protected as well by masking the IP addresses in use on its network. The method is applicable in both existing performance metric protocols and is adaptable to a new protocol which would also additionally assist in identifying the purpose of the performance metric packets and protecting the destination network from outside interference. The number of performance metrics queried by some applications could also be reduced through the use of CIDR network block tables. These tables would be referenced to determine if a previous response was cached from this network block or to allow for a longer cache time-out due to the static nature of CIDR blocks.

Multi-path multi-tier 360-degree video streaming in 5G networks

Abstract: 360° video streaming is a key component of the emerging Virtual Reality (VR) and Augmented Reality (AR) applications. In 360° video streaming, a user may freely navigate through the captured 360° video scene by changing her desired Field-of-View. High-throughput and low-delay data transfers enabled by 5G wireless networks can potentially facilitate untethered 360° video streaming experience. Meanwhile, the high volatility of 5G wireless links present unprecedented challenges for smooth 360° video streaming. In this paper, novel multi-path multi-tier 360° video streaming solutions are developed to simultaneously address the dynamics in both network bandwidth and user viewing direction. We systematically investigate various design trade-offs on streaming quality and robustness. Through simulations driven by real 5G network bandwidth traces and user viewing direction traces, we demonstrate that the proposed 360° video streaming solutions can achieve a high-level of Quality-of-Experience (QoE) in the challenging 5G wireless network environment.

Self-Promoted Prototype Refinement for Few-Shot Class-Incremental Learning

Abstract: Few-shot class-incremental learning is to recognize the new classes given few samples and not forget the old classes. It is a challenging task since representation optimization and prototype reorganization can only be achieved under little supervision. To address this problem, we propose a novel incremental prototype learning scheme. Our scheme consists of a random episode selection strategy that adapts the feature representation to various generated incremental episodes to enhance the corresponding extensibility, and a self-promoted prototype refinement mechanism which strengthens the expression ability of the new classes by explicitly considering the dependencies among different classes. Particularly, a dynamic relation projection module is proposed to calculate the relation matrix in a shared embedding space and leverage it as the factor for bootstrapping the update of prototypes. Extensive experiments on three benchmark datasets demonstrate the above-par incremental performance, outperforming state-of-the-art methods by a margin of 13%, 17% and 11%, respectively.