Mark Allman's Research Activities http://www.icir.org/mallman/research Mark Allman's Research Activites Appropriate Byte Counting http://www.icir.org/mallman/research/proj-abc.html 1123771897 Appropriate Byte Counting

For a while I have been working on a slight change in the way TCP increases it congestion window called appropriate byte counting. The scheme has performance and security benefits (at the cost of increasing burstiness a bit). A few references:

(Posted: Sun Oct 20, 2002; Updated: Thu Feb 6, 2003) ]]> Thu Aug 11 10:51:37 EDT 2005 Bursting in TCP http://www.icir.org/mallman/research/proj-bursting.html 1137074552 Bursting in TCP

Ethan Blanton and I have been working on both assessing the amount and the impact of micro-bursting on TCP performance in the wild, as well as avaluating ways to cope with micro-bursting within TCP.

Papers

(Posted: Fri Feb 25, 2005; Updated: Wed Jan 11, 2006) ]]> Thu Jan 12 09:02:32 EST 2006 Energy-Aware Network Architecture http://www.icir.org/mallman/research/proj-energy-arch.html 1216662508 Energy-Aware Network Architecture

The current network architecture often assumes hosts are either connected or disconnected and often assumes connectivity for smooth operation. We are thinking about a new network architecture that allows hosts to be "selectively connected" to the network such that they can power down to conserve power, but do so without losing their standing or presence in the network.

People

Papers

Misc.

  • Project poster presented at the November 2007 FIND PI meeting.

(Posted: Thu Nov 15, 2007; Updated: Mon Jul 21, 2008) ]]> Mon Jul 21 13:48:28 EDT 2008 Enterprise Measurement http://www.icir.org/mallman/research/proj-enterprise.html 1165933611 Enterprise Measurement

We are attempting to look at traffic characteristics inside enterprise networks to (i) assess how traffic characteristics differ from characteristics found in wide-area networks and (ii) to provide models of "background traffic" for the community (and for security researchers in particular, for testing defense mechanisms).

Papers:

We have also made available over 100 hours of anonymized enterprise packet traces available from the main project web page, which is here.

(Posted: Thu Aug 11, 2005; Updated: Tue Dec 12, 2006) ]]> Tue Dec 12 09:26:51 EST 2006 Explicit Error Transport Notification http://www.icir.org/mallman/research/proj-eten.html 1123771945 Explicit Error Transport Notification

I have worked with a number of folks recently on the problem of making TCP robust to non-congestion segment loss. Recently I have been working with Wes Eddy and Shawn Ostermann on using aggregate feedback from the network to help make TCP robust to links with non-negligable BERs. References:

Papers, Reports, Etc.

Talks

  • Mark Allman. A(nother) Technique for Improving Transport Protocol Performance in Lossy Networks. ICIR Wednesday Lunch, April 2004. Slides.

  • Mark Allman. Improving Performance of Internet Protocols Over Wireless Networks. Kent State University CS Seminar, November 2003. (Previously given at Case Western Reserve University's EECS seminar series.) Slides.

  • Wesley Eddy, Shawn Ostermann, Mark Allman. New Techniques for Making TCP Robust to Corruption-Based Loss. Third Space Internet Workshop, June 2003. Talk abstract. Slides. Wes' talk notes.

(Posted: Mon Feb 24, 2003; Last Updated: Tue Sep 21, 2004) ]]> Thu Aug 11 10:52:25 EDT 2005 Thwarting Attacks Through Cross-Organization Information Sharing http://www.icir.org/mallman/research/proj-history.html 1216662881 Thwarting Attacks Through Cross-Organization Information Sharing

We have been working on several ideas to use practical cross-organizational information sharing to find attackers and mitigate their impact.

People:

Papers:

Talks:

Also see the CCIED center's web pages.

(Posted: Mon May 23, 2005; Updated: Mon Jul 21, 2008) ]]> Mon Jul 21 13:54:41 EDT 2008 Increasing TCP's Initial Congestion Window http://www.icir.org/mallman/research/proj-initwin.html 1123771890 Increasing TCP's Initial Congestion Window

With several collaborators I have been investigating the impact of using a larger initial congestion window in TCP for some time now. The proposal is now a standards track RFC. Some pointers to our work:

(Posted: Fri Oct 25, 2002) ]]> Thu Aug 11 10:51:30 EDT 2005 Jump Start http://www.icir.org/mallman/research/proj-js.html 1192451144 Jump Start

We are considering the notion that congestion control does not require a startup phase (ala traditional slow start).

People

Papers

Talks

This is related to the Quick-Start work, which is focused on faster startup with the consent of all the nodes along the path. Work on that front can be found here.

(Posted: Tue Dec 12, 2006; Updated: Wed Feb 7, 2007) ]]> Mon Oct 15 08:25:44 EDT 2007 Internet Measurement http://www.icir.org/mallman/research/proj-meas.html 1216662638 Internet Measurement

One of my activities is to develop techniques to measure (and, hopefully better understand) networks.

Papers:

Other References:

In addition to Internet measurement I have also been working on measuring enterprise networks. There is a seperate page covering those activities.

(Posted: Wed Jun 11, 2003; Updated: Mon Jul 21, 2008) ]]> Mon Jul 21 13:50:38 EDT 2008 Opportunistic Personas http://www.icir.org/mallman/research/proj-personas.html 1186598744 Opportunistic Personas

We have started to think in terms of informal crytpo and loosening the way we think about coupling authentication and identity. By doing this we believe we can achieve better security in a number of cases, even if not ideal security.

Papers:

(Posted: Wed Aug 08, 2007; Updated: Wed Aug 08, 2007) ]]> Wed Aug 8 14:45:44 EDT 2007 Personal Namespaces http://www.icir.org/mallman/research/proj-pnames.html 1195591019 Personal Namespaces

This project involves developing an over-arching naming system for the Internet that provides an indirection layer to current naming schemes. Each person could then have their own namespace to populate (or, multiple namespaces for multiple purposes).

Papers

(Posted: Tue Nov 20, 2007; Updated: Tue Oct 23, 2007) ]]> Tue Nov 20 15:36:59 EST 2007 Quick-Start http://www.icir.org/mallman/research/proj-quickstart.html 1172842236 Quick-Start

With Sally Floyd, Amit Jain and Pasi Sarolahti I am working on a method to make TCP startup more rapidly (with the consent of the nodes on the path).

Papers:

More information and references can be found on the Quick-Start web page.

Also, this work is related to the Jump Start work, which takes the approach of wondering whether a startup phase is needed at all. That work can be found here.

(Posted: Thu May 26, 2005; Updated: Sat Jan 27, 2007) ]]> Fri Mar 2 08:30:36 EST 2007 Assessing RED's Biases http://www.icir.org/mallman/research/proj-redbias.html 1123771909 Assessing RED's Biases

Wes Eddy (Ohio University) and I have been working on quantifying the biases in RED queueing that occur depending on the units used to calculate the queue size and the marking probability (whether in terms of packets or in terms of the size of the packets).

At SIGCOMM 2002 Wes presented a poster detailing our work comparing RED byte and packet modes:

  • The abstract of the poster was given to attendees.
    The abstract also appears in ACM Computer Communication Review, July 2002.
  • Also, the poster is available here.
  • Wes' pitch from SIGCOMM was videotaped and is available here.

A paper detailing our results:

(Posted: Mon Aug 26, 2002; Updated: Mon Apr 28, 2003) ]]> Thu Aug 11 10:51:49 EDT 2005 Making TCP Robust to Packet Reordering http://www.icir.org/mallman/research/proj-reorder.html 1157165524 Making TCP Robust to Packet Reordering

Several colleagues and I are thinking about new ways to make TCP robust to packet reordering in networks.

People: Ethan Blanton, Sumitha Bhandarkar, A. L. Narasimha Reddy

Papers

Talks

  • Mark Allman. On Packet Reordering and TCP Performance. NASA GRC 5610 Research Retreat. May 2002. Slides available.

(Posted: Tue Sep 24, 2002; Updated: Fri Sep 1, 2006) ]]> Fri Sep 1 22:52:04 EDT 2006 Reducing RTOs in TCP http://www.icir.org/mallman/research/proj-rtoreduce.html 1222876063 Reducing RTOs in TCP

We have been working on evaluating and designing mechanisms that will reduce TCP's reliance on the retransmission timer for repairing segment loss, as well as correcting the RTO estimate when it fails. A few references:

Papers

Talks

  • Josh Blanton. Comparing RTO Reduction Algorithms in TCP, NASA Glenn Research Center, August 2003. Slides.

(Posted: Mon Feb 24, 2003; Updated: Wed Oct 1, 2008) ]]> Wed Oct 1 11:47:43 EDT 2008 Use of TCP SACK Information http://www.icir.org/mallman/research/proj-sack.html 1123771937 Use of TCP SACK Information

With a few collaborators I have been working on specifying the use of SACK and DSACK information.

Documents:

Talks:

  • Mark Allman. Status report given at TSV-WG meeting, IETF-55 Atlanta, GA.

(Posted: Sun Oct 20, 2002; Updated: Tue Feb 24, 2004) ]]> Thu Aug 11 10:52:17 EDT 2005 Sharing Internet Measurement Data http://www.icir.org/mallman/research/proj-simr.html 1216662677 Sharing Internet Measurement Data

With a number of collaborators I have been thinking about ways in which Internet measurement data can be shared among the community in a sane way, as well as actually taking some concrete steps and releasing some data.

Papers

Workshop Reports

Talks

Data

  • As part of the LBNL/ICSI tracing project we have released 11GB of anonymized packet headers from over 100 hours of tracing a medium size enterprise network.

(Posted: Mon Jun 7, 2004; Updated: Mon Jul 21, 2008) ]]> Mon Jul 21 13:51:17 EDT 2008 Urban Sensing Systems http://www.icir.org/mallman/research/proj-urban.html 1165937867 Urban Sensing Systems

With UCLA we have started working on the topic of personal, social and urban-scale sensing applications.

Papers:

Also see the project's web page here.

(Posted: Tue Dec 12, 2006; Updated: Tue Dec 12, 2006) ]]> Tue Dec 12 10:37:47 EST 2006 Evaluating XML-RPC http://www.icir.org/mallman/research/proj-xmlrpc.html 1123771917 Evaluating XML-RPC

I have been looking at XML-RPC from various angles (performance, coding complexity, etc.). In general, I think it would be quite nice if more folks looked at systems like this from more angles (so, programming languages folks, networking folks, etc.). References:

(Posted: Wed Nov 6, 2002; Updated: Mon Apr 28, 2003) ]]> Thu Aug 11 10:51:57 EDT 2005