A5100399267- Software-Defined Networks and 5G
- Network Traffic and Congestion Control
- Network Security and Intrusion Detection
- Internet Traffic Analysis and Secure E-voting
- Advanced Malware Detection Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
- Advanced Optical Network Technologies
- Domain Adaptation and Few-Shot Learning
- IoT and Edge/Fog Computing
- Advanced Data Storage Technologies
- Caching and Content Delivery
- Cooperative Communication and Network Coding
- Cloud Computing and Resource Management
- Advanced MIMO Systems Optimization
- Network Packet Processing and Optimization
- Interconnection Networks and Systems
- Blockchain Technology Applications and Security
- Privacy, Security, and Data Protection
- Software System Performance and Reliability
- Mobile Ad Hoc Networks
- Peer-to-Peer Network Technologies
- Handwritten Text Recognition Techniques
- Graph Theory and Algorithms
- Ferroelectric and Negative Capacitance Devices
Zhengzhou University
2024
Tsinghua University
2014-2024
Carnegie Mellon University
2017-2024
Kyoto College of Graduate Studies for Informatics
2022
Kyoto University
2022
University of Ottawa
2022
Beihang University
2019-2020
National University of Defense Technology
2020
China University of Geosciences
2017
Hangzhou Dianzi University
2017
This paper investigates a phenomenon where query-based object detectors mispredict at the last decoding stage while predicting correctly an intermediate stage. We review training process and attribute overlooked to two limitations: lack of emphasis cascading errors from sequence. design present Selective Query Recollection (SQR), simple effective strategy for detectors. It cumulatively collects queries as stages go deeper selectively forwards downstream aside sequential structure. Such-wise,...
Traffic engineering is a method to balance the flows and optimize routing in network. Software defined networking new network architecture we can gain great benefit by migrating traditional IP SDN-enabled from perspective of traffic engineering. However, due economical, organizational technical challenges, with full deployment SDN routers impractical short term. It desirable choice deploy incrementally. In this paper, seek search for an optimal migration sequence legacy so that decide where...
Starting from the concept of Cloud Radio Access Network (C-RAN), continuing with virtual (vRAN) and most recently Open RAN (O-RAN) initiative, (RAN) architectures have significantly evolved in past decade. In last few years, wireless industry has witnessed a strong trend towards disaggregated, virtualized open RANs, numerous tests deployments worldwide. One unique aspect that motivates this paper is availability new opportunities arise using machine learning, more specifically multi-agent...
Backup is cumbersome. To be effective, backups have to made at regular intervals, forcing users organize and store a growing collection of backup media. In this paper we propose novel peer-to-peer system, PeerStore, that allows the user his on other people's computers instead. PeerStore an adaptive, cost-effective system suitable for all types networks ranging from LAN, WAN large unstable like Internet. The consists two layers: metadata layer symmetric trading layer. Locating blocks...
This paper focuses on a novel and challenging detection scenario: A majority of true objects/instances is unlabeled in the datasets, so these missing-labeled areas will be regarded as background during training. Previous art [1] this problem has proposed to use soft sampling re-weight gradients RoIs based overlaps with positive instances, while their method mainly two-stage detector (i.e. Faster RCNN) which more robust friendly for missing label scenario. In paper, we introduce superior...
Bot detection methods that rely on deep packet inspection (DPI) can be foiled by encryption. Encryption, however, increases entropy. This paper investigates whether adding high-entropy detectors to an existing bot tool uses DPI restore some of the visibility. We present two classifiers, and use one them enhance BotHunter. Our results show while BotHunter misses about 50% bots when they employ encryption, our classifier restores most its ability detect bots, even
Backup is cumbersome. To be effective, backups have to made at regular intervals, forcing users organize and store a growing collection of backup media. In this paper we propose novel peer-to-peer system, PeerStore, that allows the user his on other people's computers instead. PeerStore an adaptive, cost-effective system suitable for all types networks ranging from LAN, WAN large unstable like Internet. The consists two layers: metadata layer symmetric trading layer. Locating blocks...
As Internet-of-Things (IoT) devices rapidly gain popularity, they raise significant privacy concerns given the breadth of sensitive data can capture. These are amplified by fact that in many situations, IoT collect about people other than their owner or administrator, and these stakeholders have no say how is managed, used, shared. To address this, we propose a new model ownership, Ephemeral Ownership (TEO). TEO allows to quickly register with an device for limited period, thus claim...
The internet is playing an increasingly crucial role in both personal and business activities. In addition, with the emergence of real-time, delay sensitive mission-critical applications, stringent network availability requirement put forward for service providers (ISPs). However, commonly deployed intradomain link-state routing protocols react to link failures by globally exchanging state advertisements recalculating table, inevitably causing significant forwarding discontinuity after a...
Edge Computing (EC) provides delay protection for some delay-sensitive network services by deploying cloud infrastructure with limited resources at the edge of network. In addition, Network Function Virtualization (NFV) implements functions replacing traditional dedicated hardware devices Virtual (VNF) that can run on general servers. NFV environment, Service Chaining (SFC) is regarded as a promising way to reduce cost configuring services. therefore allows deploy in more flexible and...
Inter-DataCenter Wide Area Network (Inter-DC WAN) that connects geographically distributed data centers is becoming one of the most critical network infrastructures. Due to limited bandwidth and inevitable link failures, it highly challenging guarantee availability for services, especially those with stringent demands, over inter-DC WAN. We present BATE, a novel Traffic Engineering (TE) framework (BA) provision, which aims ensure each demand must be satisfied stipulated probability, when...
As a typical representative of new generation military information technology, the value and significance Internet Battlefield Things (IoBT) has been widely recognized by world's forces. At same time, is facing serious scalability security challenges. This paper presents basic concept six-domain model IoBT, explains integration framework IoBT blockchain. Furthermore, we design build novel IoT called IoBTChain based on blockchain smart contracts, which adopts credit-based resource management...
We present simple and flexible drop-in modules in feature pyramids for general object detection, which can be easily generalized to other anchor-free detectors without introducing extra parameters, only involves negligible computational cost on training testing. The proposed detector, called NCMS, inserts a norm calibration (NC) operation between the detection head alleviate balance bias caused by pyramid network (FPN). Furthermore, NCMS leverages an enhanced multi-feature selective strategy...
High entropy (HE) traffic may result from encrypted such as C&C botnet communication. Such also tends to be opaque an IDS. However, line speed calculation is expensive, especially for long flows. In this paper we introduce methodology classify flows HE or low (LE) by considering only the first M packets of flow. We use our classifiers in two ways: (a) improve effectiveness BotHunter, a bot detection tool when presented with traffic, and (b) filter reduce load on implement Snort preprocessor....
With an increasing urgent demand for fast recovery routing mechanisms in large-scale networks, minimizing network disruption caused by failure has become critical. However, a large number of relevant studies have show... | Find, read and cite all the research you need on Tech Science Press
Numerous distributed applications are deployed in the cross-datacenter networks (Cross-DC) where geographically data centers (DC) connected by wide area network (WAN). These online will generate both intra-datacenter and inter-datacenter traffic, each with distinct requirements characteristics. We find that existing combined congestion control schemes ignore interaction of two types traffic hybrid fail to accurately estimate cross center extent. In this paper, we propose IDCC a delay based...