A5101712908- Software-Defined Networks and 5G
- Data Stream Mining Techniques
- Network Security and Intrusion Detection
- Network Traffic and Congestion Control
- Software System Performance and Reliability
- Caching and Content Delivery
- Advanced Database Systems and Queries
- Data Management and Algorithms
- Time Series Analysis and Forecasting
- Solar and Space Plasma Dynamics
- Internet Traffic Analysis and Secure E-voting
- Machine Learning and Data Classification
- Video Analysis and Summarization
- Artificial Intelligence in Games
- Anomaly Detection Techniques and Applications
- Network Packet Processing and Optimization
- Geomagnetism and Paleomagnetism Studies
- Cloud Computing and Resource Management
- Metaheuristic Optimization Algorithms Research
- Covalent Organic Framework Applications
- Age of Information Optimization
- IoT and Edge/Fog Computing
- Aeolian processes and effects
Harvard University Press
2019-2023
Peking University
2017-2019
When network is undergoing problems such as congestion, scan attack, DDoS etc., measurements are much more important than usual. In this case, traffic characteristics including available bandwidth, packet rate, and flow size distribution vary drastically, significantly degrading the performance of measurements. To address issue, we propose Elastic sketch. It adaptive to currently characteristics. Besides, it generic measurement tasks platforms. We implement sketch on six platforms: P4, FPGA,...
Finding top-k elephant flows is a critical task in network traffic measurement, with many applications congestion control, anomaly detection and engineering. As the line rates keep increasing today's networks, designing accurate fast algorithms for online identification of becomes more challenging. The prior are seriously limited achieving accuracy under constraints heavy small on-chip memory use. We observe that basic strategies adopted by these either require significant space overhead to...
Data stream processing is a fundamental issue in many fields, such as data mining, databases, network traffic measurement. There are five typical tasks processing: frequency estimation, heavy hitter detection, change distribution and entropy estimation. Different algorithms proposed for different tasks, but they seldom achieve high accuracy speed at the same time. To address this issue, we propose novel structure named HeavyGuardian. The key idea to intelligently separate guard information...
Given v sets and an incoming item e, multi-set membership query is to report which set contains e. Multi-set a fundamental problem in computer systems applications. All existing data structures cannot achieve small memory usage, fast speed high accuracy at the same time. In this paper, we propose novel probabilistic structure named Difference Bloom Filter (DBF) for query, not only more accurate than state-of-the-art, but has faster speed. There are two key design principles DBF. The first...
By moving monolithic network appliances to software running on commodity hardware, function virtualization allows flexible resource sharing among functions and achieves scalability with low cost. However, due contention, can suffer from performance problems that are hard diagnose. In particular, when many flows traverse a complex topology of NF instances, it is pinpoint root causes for flow experiencing issues such as throughput or high latency. Simply maintaining counters at individual NFs...
When network is undergoing problems such as congestion, scan attack, DDoS etc, measurements are much more important than usual. In this case, traffic characteristics including available bandwidth, packet rate, and flow size distribution vary drastically, significantly degrading the performance of measurements. To address issue, we propose Elastic sketch. It adaptive to currently characteristics. Besides, it generic measurement tasks platforms. We implement sketch on six platforms: P4, FPGA,...
Network measurement is an essential building block for a variety of network applications such as traffic engineering, quality service, load-balancing and intrusion detection. Maintaining per-flow state often impractical due to the large number flows, thus modern systems use complex data structures that are updated with each incoming packet. Therefore, designing operate at line speed significant challenge in this domain.
Sketch is a data structure used to record frequencies of items in multiset, which widely streams, graph, distributed datasets processing, etc. It works with small memory usage and high speed at the cost slight inaccuracy. In practice, many are non-uniformly distributed. Unfortunately, existing sketches can hardly work well on non-uniform datasets. To address this issue, we propose new sketch framework, namely ABC be applied most significantly improve accuracy The key idea behind our...
Finding top-k hot items in a data stream is critical problem big management. It benefits various kinds of applications, such as mining, databases, network traffic measurement, etc. However, the speed streams become increasingly large, it becomes more and challenging to design an accurate fast algorithm for this problem. There are several existing algorithms, including Space-Saving, Frequent, Lossy counting, with Space-Saving being most widely used among them. Unfortunately, all these...
The packet delivery fairness is critical in many applications the cloud, such as exchange systems, consensus protocols, and online gaming applications. However, due to nonidentical dynamic forwarding paths, well in-network queuing delays, supporting challenging a shared compute environment. In this paper, we present Octopus, first general fair service achieve arrival time variations smaller than tens of nanoseconds, with existence latency network. key ideas Octopus support good come from...
short-paper Share on DeepDiag: Detailed NFV Performance Diagnosis Authors: Junzhi Gong Harvard University UniversityView Profile , Yuliang Li Bilal Anwer AT&T AT&TView Aman Shaikh Minlan Yu Authors Info & Claims SIGCOMM Posters and Demos '19: Proceedings of the ACM 2019 Conference DemosAugust Pages 33–35https://doi.org/10.1145/3342280.3342298Published:19 August 2019Publication History 1citation329DownloadsMetricsTotal Citations1Total Downloads329Last 12 Months16Last 6 weeks3 Get Citation...
Virtualized radio access networks (vRANs), which allow running RAN processing on commodity servers instead of proprietary hardware, are gaining adoption in cellular networks. Two properties the vRAN's "Distributed Unit (DU)" that implements lower layers---its real-time deadlines and its black-box nature---make it challenging to provide resilience features such as upgrades failover without long service disruptions. These preclude use existing techniques like virtual machine migration or state...