- Author:
Lisa Yan
(Stanford University) Computing Methodologies , Machine Learning, Security and Privacy
重点
任务:
这篇文章最关键的是告诉我们怎么学习Advanced Network,怎样学习是最有效的,并且对网络科研有帮助的。
CS224
Wordpress
- [-] CS244 Reproducibility Project
- [-] Week 1 Day 1 Project proposal
- [-] Week 2 Day 14 Intermediate report
- [-] Week 4 Day 23 Final report
我决定选择PCC这篇文章作为第一篇复现文,在与直接作者讨论后,我们相信,AWS容器的可视化环境降低了PCC性能。PCC依赖于在精确时间发送包裹。TCP在Mininet中会有稍微差一些的表现。
环境:
- Mininet
- Mahimahi
- NS-2
- NS-3
- Emulab
我们鼓励学生在他们的实验平台上使用MiniNet或Mahimahi仿真系统。MiniNet最适合多节点拓扑,而Mahimahi在修改和测试单个链路上运行的拥塞控制协议时很好。虽然我们通常更喜欢学生使用模拟器,因为模拟器具有更真实的网络特性,例如给定节点拓扑的实时流量处理,例如NS-3。我们为所有学生提供亚马逊网络服务(AWS)计算云(EC2)上的计算资源。
学生经验:
Some Anecdotes in Reproduction:
- Encounter an unknown facet of networking
- Understanding a cutting-edge research
- Digging deep into workshop papers
- [-] There is a level you can only get by reproducing it or implementing it.
- [-] Which is something that I wouldn’t have done if I had just read this paper
- [-] Because in a sense you already did the set up for them and you wrote it in a more lucid way
slef-clocking 如果把发送端看做一个时钟(clock),这个时钟的走动是通过发送端主动发送数据和获得应答来推进的,这个过程叫做self-clocking
ramp up 逐步加大