A5009477401- Software-Defined Networks and 5G
- Interconnection Networks and Systems
- Cloud Computing and Resource Management
- Network Traffic and Congestion Control
- Caching and Content Delivery
- Advanced Data Storage Technologies
- Advanced Optical Network Technologies
- International Business and FDI
- Global trade and economics
- GNSS positioning and interference
- Embedded Systems Design Techniques
- Internet Traffic Analysis and Secure E-voting
- Energy, Environment, Economic Growth
- IoT and Edge/Fog Computing
- Firm Innovation and Growth
- Video Coding and Compression Technologies
- Global Trade and Competitiveness
- Network Packet Processing and Optimization
- IPv6, Mobility, Handover, Networks, Security
- Parallel Computing and Optimization Techniques
- Economic Zones and Regional Development
- E-commerce and Technology Innovations
- Optical Network Technologies
- Peer-to-Peer Network Technologies
- Advanced Memory and Neural Computing
Hunan University
2008-2024
Hubei University
2022-2023
Wuhan Business University
2023
Hubei University Of Economics
2022
Tsinghua University
2014-2022
Hubei Normal University
2020
Shanghai University of Engineering Science
2020
Nanjing University
1997-2019
Microsoft Research (United Kingdom)
2017
Center for Life Sciences
2017
Inspired by the brain's structure, we have developed an efficient, scalable, and flexible non-von Neumann architecture that leverages contemporary silicon technology. To demonstrate, built a 5.4-billion-transistor chip with 4096 neurosynaptic cores interconnected via intrachip network integrates 1 million programmable spiking neurons 256 configurable synapses. Chips can be tiled in two dimensions interchip communication interface, seamlessly scaling to cortexlike sheet of arbitrary size. The...
Despite many years of improvements to it, TCP still suffers from an unsatisfactory performance. For services dominated by short flows (e.g., web search and e-commerce), the flow startup problem cannot fully utilize available bandwidth in modern Internet: starts a conservative static initial window (IW, 2-4 or 10), while most are too converge best sending rate before session ends. long video streaming file downloading), congestion control (CC) scheme manually statically configured might not...
Abstract Current state-of-the-art stretchable/flexible sensors have received stringent demands on sensitivity, flexibility, linearity, and wide-range measurement capability. Herein, we report a methodology of strain based graphene/Ecoflex composites by modulating multiscale/hierarchical wrinkles flexible substrates. The sensor shows an ultra-high sensitivity with gauge factor 1078.1, stretchability 650%, response time ~140 ms, superior cycling durability. It can detect physiological signals...
As mobile Internet is now indispensable in our daily lives, WiFi's latency performance has become critical to applications' quality of experience. Unfortunately, WiFi hop the wild remains largely unknown. In this paper, we first propose an effective approach break down round trip network latency. Then provide systematic study on based and factors collected from 47 APs T university campus for two months. We observe that can be weakest link latency: more than 50% (10%) TCP packets suffer...
In this paper, the bibliometrics analysis software Cite Space is used for quantitative of retrieved literature, aiming to provide a systematic and complete knowledge graph study change physical activity level Chinese college students. academic journals papers from 2004 2024 in National Database (CNKI) were selected as literature sources, 484 articles research objects after screening. way, annual publication volume, international cooperation characteristics, institutions cooperation, authors...
RDMA is becoming prevalent because of its low latency, high throughput and CPU overhead. However, in current datacenters, remains a single path transport which prone to failures falls short utilize the rich parallel network paths. Unlike previous multi-path approaches, mainly focus on TCP, this paper presents for RDMA, i.e. MP-RDMA, efficiently utilizes paths datacenters. MP-RDMA employs three novel techniques address challenge limited NICs on-chip memory size: 1) ACK-clocking mechanism...
Nowadays, with the rapid development of hardware and network technologies, various types smart devices are released by different vendors, resulting in emergence Internet Things (IoT). However, most existing approaches for IoT device management designed a centralized way, whose efficiency meets challenges recently because large scale heterogeneous modules highly dynamic essence networks. To tackle this challenge, we propose distributed approach service composition from social point view....
To achieve low TCP flow completion time (FCT) in data center networks (DCNs), it is critical and challenging to rapidly recover loss without adding extra congestion. Therefore, this paper, we propose a novel recovery approach fast multi-path (FUSO) that exploits diversity DCN for transport recovery. In FUSO, when sender suspects on one sub-flow, packets are immediately sent over another sub-flow not or less lossy has spare congestion window slots. FUSO does need wait timeout the cautious...
With the acceleration of urbanization process, country’s strong support for healthy development logistics industry has made urban become a hot topic in recent years. increase number vehicles, traffic accidents have more frequent. Intelligent vehicle collision avoidance system is an important part advanced safety technology. To veracity and practicability avoidance, this paper presents distance model longitudinal vehicles considering road slope adhesion coefficient. Based on kinetic theory,...
Latency, which directly affects the user experience and revenue of web services, is far from ideal in reality, due to well-known TCP flow startup problem. Specifically, since starts a conservative static initial window (IW, 2~4 or 10), most flows are too short have enough time find its best congestion before session ends. As result, cannot fully utilize available bandwidth modern Internet. In this paper, we propose use group-based reinforcement learning (RL) enable server, through...
Due to its superior performance, Remote Direct Memory Access (RDMA) has been widely deployed in data center networks. It provides applications with ultra-high throughput, ultra-low latency, and far lower CPU utilization than TCP/IP software network stack. However, the connection states that must be stored on RDMA NIC (RNIC) small memory result poor scalability. The performance drops significantly when RNIC needs maintain a large number of concurrent connections.We propose StaR (Stateless...
Protein-nucleic acid interactions play a very important role in variety of biological activities. Accurate identification nucleic acid-binding residues is critical step understanding the interaction mechanisms. Although many computationally based methods have been developed to predict residues, challenges remain. In this study, fast and accurate sequence-based method, called ESM-NBR, proposed. we first use large protein language model ESM2 extract discriminative properties feature...
Mixed traffic flow composed of autos and non-autos widely exists in developing countries areas. To investigate the operational characteristics mixed consisting vehicles different types (large vehicles, cas, bicycles), we develop a cellular automaton model to replicate travel behaviors on bi-directional road segment with respect physical mechanic ftures différent vehicle types. By implementing eesential parameters calibrated through field data collection, numerical study is carried out...
Institutional bias is an important topic in social justice and the presence of institutional widely recognized China's urban housing sector. This paper proposes a new framework for measuring assessing system based on structural equation modeling Pratt index relative importance linear regression. The proposed analyzes pathways through which nonmarket forces affect housing, provides quantitative measures to evaluate both direct indirect effects biased institutions outcomes. A case study...
Failures are not uncommon in production data center networks (DCNs) nowadays, and it takes long time for the network to recover from a failure find new forwarding paths, significantly impacting real interactive applications at upper layer. The slow recovery is due two primary reasons. First, there lacks immediate backup paths downward links DCN with multi-rooted tree topology. Second, distributed routing protocols take converge after failures. In this paper, we present fault-tolerant...
Current TCP is very inefficient for web services. Web transactions are often short-lived. flow starts with a conservative initial congestion window (IW), which causes multiple round-trip times to finish the transmission even if end-to-end bandwidth sufficient transaction be finished in one time. Previous research efforts have been focusing on finding overall best IW entire Internet or service company. However, we observe that one-IW-fits-all suboptimal after year of online measurement Baidu,...
Failures are not uncommon in production data center networks (DCNs) nowadays. It takes long time for the DCN routing to recover from a failure and find new forwarding paths, significantly impacting realtime interactive applications at upper layer. In this paper, we present fault-tolerant solution, called F <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> Tree, which is readily deployed existing DNCs. Tree can improve recovery only through...
We use both theoretical analysis and simulations to study the impact of crosspoint-queued (CQ) buffer size on CQ switch throughput delay performance under different traffic models, input loads, scheduling algorithms. In this paper, we present following. 1) prove stability using any work-conserving algorithm. 2) an exact closed-form formula for a nonclosed- form but convergent its static nonwork- conserving random algorithms with given independent Bernoulli traffic. 3) show that above results...