Network Reading Group, Spring 2002

The network reading group discusses network research literature and ideas.

Location: 380 Soda Hall

Date & Time: Thursday 11:00am-12:00pm

Organizer: George Porter gporter@EECS.Berkeley.EDU.

To join the email list: Send mail to majordomo@iceberg.CS.Berkeley.EDU with the line "subscribe netread".

To email people: The alias for the group is netread@iceberg.CS.Berkeley.EDU.

Past meetings of the reading group:

Schedule:
01/31/02 George Porter
  • ``Stability issues in OSPF routing'', Anindya Basu, Jon Riecke (SIGCOMM 2001) ppt presentation
  • ``The Impact of Internet Policy and Topology on Delayed Routing Convergence'', Labovitz, Ahuja, Wattenhofer, Venkatachary (INFOCOMM 2001) ppt presentation
    • 02/07/02 Matthew Caesar
  • Span: An energy-efficient coordination algorithm for topology maintenance in ad hoc wireless networks - Chen, Jamieson, Balakrishnan, Morris (To appear in ACM Wireless networks journal, vol. 8, no. 5, sept 2002) Powerpoint presentation
    • 02/14/02 Karthik Lakshminarayanan
  • Controlling High Bandwidth Aggregates in the Network - Manajan, Bellovin, Floyd, Ioannidis, Paxson, Shenker (under submission to CCR) Powerpoint presentation
    • 02/21/02 Christos Papadimitriou. "Power Laws and the Internet"
    02/28/02 Yan Chen: Practice talk for IPTPS paper (Powerpoint Slides)
    03/07/02 Adam (12)
    03/14/02 Lakshmi: Relationships between Proof Carrying Codes and Dynamic Packet State (11)
    03/21/02 Sridhar: Checksums and Packet Reordering (12) PPT Slides
    03/28/02 No NETREAD -- Spring Break
    04/04/02 Weidong (12)
    04/11/02 Bhaskaran Raman
  • Revealing the problems with 802.11 medium access control protocol in multi-hop wireless ad hoc networks - Shugong Xu and Tarek Saadawi
    • 04/11/02 Bhaskar
    04/18/02 AP Presentation 1 Slides | Presentation 2 Slides
    John Byers, Jeffrey Considine, Michael Mitzenmacher and Stanislav Rost, "Informed Content Devlivery Across Adaptive Overlay Networks," SIGCOMM 2002.

    Ross Anderson, "Why Cryptosystems Fail?," Communications of the ACM, 1994.

    04/25/02 Yan Chen
    05/09/02 Chris Olston (10:30am)

    Title: Approximate Data Caching

    In many data replication environments, exact synchronization between source data objects and cached copies is infeasible due to its demands on network bandwidth. As a result, stale (out-of-date) copies are permitted and synchronization is performed less frequently than source data changes. Overall, less frequent synchronization means decreased network cost but also decreased cache precision. Historically, the fundamental tradeoff between network cost and the precision of cached data has been managed in an ad-hoc way. In this talk I will outline a principled approach to managing and exploiting this tradeoff.

    If network usage is flexible but communication incurs a cost, it is desirable to minimize network cost while still providing guarantees about the precision of cached objects. Our solution, called TRAPP (for Tradeoff in Replication Precision and Performance), caches approximate values with bounded imprecision, rather than stale exact values. TRAPP guarantees to satisfy per-query precision constraints while minimizing network cost.

    Our solution incorporates adaptive algorithms that work well under fluctuating data update rates and query loads. The talk will provide an overview of our work, some empirical results, and an animated demonstration of one of our adaptive algorithms.

    Links of Interest: