Software Defined Networking (SDN) has become a popular paradigm for centralized control in many modern networking scenarios such as data centers and cloud. For large hosting hundreds of thousands servers, there are few switches that need to be managed fashion, which cannot done using single controller node. Previous works have proposed distributed architectures address scalability issues. A key limitation these works, however, is the mapping between switch statically configured, may result...

Netflix and Hulu are leading Over-the-Top (OTT) content service providers in the US Canada. alone accounts for 29.7% of peak downstream traffic 2011. Understanding system architectures performance can shed light on design such large-scale video streaming platforms, help improving future systems. In this paper, we perform extensive measurement study to uncover their strategies. bear many similarities. Both platforms rely heavily third-party infrastructures, with migrating that majority its...

We study the Hulu online video service via active measurements. It is known that utilizes multiple CDNs to serve users' requests. The focus of this on how selects and each CDN allocates resources (i.e., servers) user Based our analysis measurement data, we find frequently changes preferred for users. However, once a selected, clients try stay with same during entire length movie even when performance degrades. While selection not fixed, observe attempts divide requests among attain fixed...

Since the outbreak of COVID-19 pandemic, videoconferencing has become default mode communication in our daily lives at homes, workplaces and schools, it is likely to remain an important part post-pandemic world. Despite its significance, there not been any systematic study characterizing user-perceived performance existing systems other than anecdotal reports. In this paper, we present a detailed measurement that compares three major systems: Zoom, Webex Google Meet. Our based on 48 hours'...

Gradient descent strategy, viewed as an important model optimization method, has been widely used for various tasks (such model-based image denoising) of computer vision. In the gradient denoising model, learning rate (LR) and residual component are two parts to be adaptively estimated its stable point. This letter proposes a deep network (DGDNet), including key points: one is that LR designed with eigenvalues Hessian matrix remotely sensed images (RSIs) their local weighted factor (LWF),...

Remotely sensed images degraded by additive white Gaussian noise (AWGN) are not beneficial for the analysis of their contents. Such a phenomenon is usually modeled as an inverse problem which can be solved model-based optimization methods or discriminative learning approaches. The former pursue pleasing performance at cost highly computational burden while latter impressive fast testing speed but limited application range. To join merits, this letter proposes nonlocal self-similar (NSS)...

In this paper, we consider the problem of designing a data structure that can perform fast multiple-set membership testing in deterministic time. Our primary goal is to develop hardware implementation uses only embedded memory blocks. Prior efforts solve involve hashing into multiple Bloom filters. Such approach needs priori knowledge number elements each set order size filter. We use single-Bloom-filter-based and sets hash functions code for (group) id. Since single filter used, it does not...

Recently, there has been much renewed interest in developing compact data structures for packet processing functions such as longest prefix-match IP lookups. This motivated by several factors: (1) The advent of 100 Gbps interfaces necessitating correspondingly fast algorithms with a memory footprint; (2) network virtualization leading to physical router platforms making it critical reduce high-speed needs per virtual router; (3) software routers built on multi-core processors requiring the...

Mixed (random and stripe) noise will cause serious degradation of optical remotely sensed image quality, making it hard to analyze their contents. In order remove such noise, various inverse problems are usually constructed with different priors, which can be solved by either model-based optimization methods or discriminative learning methods. However, they have own drawbacks, as the former flexible but time-consuming for pursuit good performance; while later fast limited extensive...

Optical flow represents vector motion of targets, and its estimation is a hot research in computer vision. The traditional prior-based methods build mathematical models using different regularizers on object characteristics. These are straightforward easily understood, but they suffer from high-computational burden. Deep learning-based approaches greatly improve their efficiency, filter sizes fixed receptive fields (RFs) large. To solve these problems, this article proposes novel recurrent...

Recent approaches have demonstrated the effectiveness of Vision Transformer (ViT) with attention mechanisms for domain generalization Face Anti-Spoofing (FAS). However, current algorithms highlight all salient objects (e.g., background objects, hair, glasses), which results in feature learned by model containing face-irrelevant noisy information. Inspired existing Vision-language works, we propose VL-FAS to extract more generalized and cleaner discriminative features. Specifically, leverage...

Per-flow network traffic measurement is an important component of management, performance assessment, and detection anomalous events such as incipient DoS attacks. In [1], the authors developed a mechanism called RATE where focus was on developing memory efficient scheme for estimating per-flow rates to specified level accuracy. The time taken by estimate function estimation accuracy this acceptable several applications. However some applications, quickly detecting worm related activity or...