Super Spreader Identification Using Geometric-Min Filter
Autor: | Youlin Zhang, Qingjun Xiao, Olufemi Odegbile, Chaoyi Ma, Shigang Chen |
---|---|
Rok vydání: | 2022 |
Předmět: |
Computer Networks and Communications
business.industry Computer science Hash function Real-time computing Internet traffic Computer Science Applications Network management Identification (information) Software Filter (video) Overhead (computing) System on a chip Electrical and Electronic Engineering business |
Zdroj: | IEEE/ACM Transactions on Networking. 30:299-312 |
ISSN: | 1558-2566 1063-6692 |
DOI: | 10.1109/tnet.2021.3108033 |
Popis: | Super spreader identification has a lot of applications in network management and security monitoring. It is a more difficult problem than heavy hitter identification because flow spread is harder to measure than flow size due to the requirement of duplicate removal. The prior work either incurs heavy memory overhead or requires heavy computations. This paper designs a new super-spreader monitor capable of identifying all flows whose spreads are greater than a user-specified threshold with a probability that can be arbitrarily set. It introduces a generalized geometric hash function, a generalized geometric counter, and a novel geometric-min filter that blocks out the vast majority of small/medium flows from being tracked, allowing us to focus on a small number of flows in which super spreaders are identified. We provide an analytical way of properly setting the system threshold to meet probabilistically guaranteed identification of super spreaders, and implement it on both hardware (FPGA) and software platforms. We perform extensive experiments based on real Internet traffic traces from CAIDA. The results show that with proper parameter settings, the new monitor can identify more than 99% super spreaders with a low memory requirement, better than the prior art. |
Databáze: | OpenAIRE |
Externí odkaz: |