About

This page contains information for the PacketLab project, including the problem PacketLab addresses, the PacketLab design, and the People behind PacketLab. Readers looking for more in-depth information can also refer to the original IMC’17 paper that first introduced PacketLab to the measurement community.

Part of the following text is adapted from the extended abstract PacketLab - Tools Alpha Release and Demo for the IMC’22 poster session.

The Problem

One tricky problem for distributed network measurement studies is having access to the proper vantage points for experiment data collection. Due to the high cost of establishing and maintaining a large distributed measurement infrastructure, researchers have relied on sharing of measurement infrastructure. This nevertheless introduces new difficulties:

• Compatibility: Varying measurement platform designs require the taxing process of porting experiments across platforms.
• Incentives: Deploying new experiments requires platform operator support.
• Trust: Platforms supporting custom measurements limit their availability to vetted researchers for security reasons.

The Design

In response to the difficulties, we proposed PacketLab. PacketLab addresses the aforementioned problems with two design choices:

1. Exporting the network (BSD-socket-like) interface of endpoints (i.e. vantage points) to experimenters via a common mechanism - the PacketLab protocol.
2. Introducing the use of cryptographic certificates with program-based restrictions for endpoint access control.

Exporting a common interface helps experiment compatibility via the notion of write once, run anywhere. Moreover, exporting a common interface reduces platform operator burden for new experiments as no effort other than interface maintenance is required. The relatively-simple BSD-socket-like interface also reduces the platform operator effort in supporting the PacketLab interface. The cryptographic certificate design enables a scalable trust architecture as operators can use PacketLab experiment privilege provisioning capabilities to customize restrictions.

The People

PacketLab is currently being developed by a collaboration of students and researchers from the Center for Applied Data Analysis (CAIDA), the University of California at San-Diego (UCSD), and the University of Illinois at Urbana-Champaign (UIUC).

Active Members

Kirill Levchenko (UIUC)	kc claffy (CAIDA)	Bradley Huffaker (CAIDA)
Ricky Mok (CAIDA)	Tzu-Bin Yan (UIUC)	Zesen "Jason" Zhang (UCSD)
Jingxuan Wu (UIUC)	Patrick Marschoun (UIUC)

Past Members

Lamya Alowain (UIUC)	Ansh Shah (UIUC)	Vinay Pillai (UCSD)
Yuxuan "Michael" Chen (UIUC)	Anthea Chen (UIUC)	Tabish Shaik (UIUC)
Richard Masser-Frye (UCSD)