Large-scale data centers enable the new era of cloud computing and provide core infrastructure to meet storage requirements for both enterprise information technology needs cloud-based services. To support ever-growing needs, number servers in today's are increasing exponentially, which turn leads enormous challenges designing an efficient cost-effective center network. With availability security at stake, issues with networks more critical than ever. Motivated by these issues, many novel...

Network monitoring and measurement are crucial in network management to facilitate quality of service routing performance evaluation. Software Defined Networking (SDN) makes easier by separating the control plane data plane. SDN is lightweight as operators only need install a module into controller. Active techniques usually introduce too many overheads network. The state-of-the-art approaches utilize sampling method, aggregation flow statistics passive reduce overheads. However, little work...

In order to mine latent semantics from text data, word embedding and topic modeling are two major methodologies in industry. From a pragmatic perspective, each of these lines semantic models faces increasing challenges real-life applications. However, modern mining tasks typically require panoramic view the semantics. Hence, discovering heterogeneous (e.g., types topics) is critical for performance tasks, it necessary design model that meets this demand. Furthermore, with arrival big data...

This paper presents the design, analysis, and implementation of a novel data center network architecture, named NovaCube. Based on regular Torus topology, NovaCube is constructed by adding number most beneficial jump-over links, which offers many distinct advantages practical benefits. Moreover, in order to enable achieve its maximum theoretical performance, probabilistic oblivious routing algorithm PORA carefully designed. both deadlock livelock free algorithm, achieves near-optimal...

Cholangiocarcinoma (CCA) is the second most common primary liver cancer and characterized by huge heterogeneity, difficult diagnosis, poor prognosis. Fibrosis-associated heterogeneity in CCA serves as an indicator of malignant progression cancer; however, a precise approach to deciphering fibrosis still lacking. Typically, tissue proteome profiled analyzing bulk tissues, which gives average results different cell types, especially for tissues cells occupy very small proportion. Laser...

This paper presents the design, implementation and evaluation of SprintNet, a novel network architecture for data centers. SprintNet achieves high performance in capacity, fault tolerance, latency. is also scalable, yet low-diameter where maximum shortest distance between any pair servers can be limited by no more than four independent number layers. The specially designed routing schemes strengthen its merits. Both theoretical analysis simulation experiments are conducted to evaluate...

The rapid development of software defined measurement has significantly improved network and monitoring. key challenge for is to design a low-cost framework which minimum impact on the network. state-of-the-art approaches mainly focus reducing overhead by sampling or aggregation. However, little attention been devoted eliminating this issue in physical layer. We observe that placement controllers affects measurement. Based observation, we rethink frameworks propose novel scheme minimize...

The data center network connecting the servers in a plays crucial role orchestrating infrastructure to deliver peak performance users. In order meet high and reliability requirements, is usually constructed of massive number devices links achieve 1:1 oversubscription for workload. However, traffic rarely ever hits capacity practice are underutilized most time, which results an enormous waste energy. Therefore, aiming energy proportional without compromising throughput fault tolerance too...

Software defined networking (SDN), which enables programmability, has the advantage of global visibility and high flexibility. However, when forwarding new flows in SDN, interaction between switch controller imposes extra latency such as round-trip time routing path search time. Even though is acceptable for elephant since it only takes limited ratio total transmission flows, an overkill to pay certain overheads mice due their short Moreover, frequently invoked by number accounts a large...

In this paper, we present the design, analysis, and implementation of a novel data center network architecture named CLOT, which delivers significant reduction in diameter, latency, infrastructure cost. CLOT is built based on switchless torus topology by adding only number most beneficial low-end switches proper way. Forming servers close proximity each other well implements locality. The extra layer largely shortens average routing path length network, increases communication efficiency. We...

General-purpose topic models have widespread industrial applications. Yet high-quality modeling is becoming increasingly challenging because accurate require large amounts of training data typically owned by multiple parties, who are often unwilling to share their sensitive for collaborative without guarantees on privacy. To enable effective privacy-preserving multiparty modeling, we propose a novel federated general-purpose model named private and consistent discovery (PC-TD). On the one...

IPv6 is growing in popularity internationally nowadays. In this paper, we propose a secure DHCPv6 system that uses RSA authentication and Self-Certified Address to improve the access safety. By appending Certificate Option messages, guarantees integrity of messages. addition, allocates addresses which can be verified by address itself. With benefits Address, prevents from various network attack such as Man-in-Middle Denial-of-Service attack. It also provides strict control checking whether...

Sketch-based measurement provides traffic data summary in a memory-efficient way with provable accuracy bound. Recent advances software defined networking (SDN) facilitate the development and implementation of sketch-based applications. However, usually requires TCAMs which are precious resource switch to match packet fields. The key challenge for is how accept more concurrent tasks minimum usage. Existing proposals attempt achieve this goal by exploring different task assignment algorithms....

Data centers are considered as the indispensable infrastructure for cloud services nowadays. However, large-scale data using traditional TCP protocol may experience some performance issues, such Incast and long query completion times. It is shown that fundamental cause of these problems existing protocols unable to effectively detect packet losses in network, which always causes timeouts centers. In this paper, we present drop notification, a mechanism can accurately inform senders about...

Reducing the power consumption of packet switches is becoming increasingly significant to future networks. However, previous research all focused on reducing in crossbar-based switches, which either complex or not effective, especially some extreme cases. This paper proposes leverage dynamic voltage and frequency scaling (DVFS) technique buffered more flexible simple. The basic idea decrease working frequencies crosspoint buffers while still preserving maximum throughput satisfactory delay....

The advancement of low to zero carbon fuels for internal combustion engines has made Nitrogen oxides (NOx) emissions a focal point engine aftertreatment. Lean NOx Trap (LNT) is one the most promising technologies reducing in light-duty vehicles. proximity barium (Ba) platinum (Pt) sites plays significant role adsorption, reduction, and desorption inside LNT. A simultaneous loading Pt clusters (Pt3) oxide (BaO)2 on γ-Al2O3(100) was innovatively established this study, effect carrier -...

Flow latency monitoring is a fundamental task in network measurement. The development of software defined networking enables flexible flow the control plane. Existing approaches mainly focus on direct probe-based measurement, which has high measurement overhead, especially accuracy systems. In this paper, we revisit framework and explore cost-effective approach named CFlam to produce results. We observe large scale generates too many duplicate probe packets. Based observation, attempt...