Team for Research in
Ubiquitous Secure Technology

Citation
Michael K. Reiter, Asad Samar. "Self-service, with applications to distributed classifier construction". Talk or presentation, 27, April, 2006; Poster given at Trust NSF Site Visit.

Abstract
Motivated by the desire to efficiently construct global network traffic models to aid in traffic characterization and intrusion detection, we introduce a service implementation strategy called "self service" that scales by harnessing client resources in the implementation of the service. As its name indicates, self-service operates by migrating relevant portions of service state to each client, which performs its operations itself. As such, self-service is best suited for services whose state can be decomposed into small objects, of which client operations typically involve only a few. We will give an overview of the main algorithms that contribute to this strategy, including those for implementing serializable or strictly serializable operations, for surviving client disconnections, and for optimizing operation latency. We also describe the preliminary performance of an application that motivated this work, i.e., a service for constructing global network traffic models to aid in traffic characterization and intrusion detection.

Electronic downloads

• reiter_SelfService_TrustApril06Poster.ppt · application/vnd.ms-powerpoint · 291 kbytes

• HTML

  Michael K. Reiter, Asad Samar. <a
  href="http://www.truststc.org/pubs/88.html"
  ><i>Self-service, with applications to distributed
  classifier construction</i></a>, Talk or
  presentation,  27, April, 2006; Poster given at Trust NSF
  Site Visit.

• Plain text

  Michael K. Reiter, Asad Samar. "Self-service, with
  applications to distributed classifier construction".
  Talk or presentation,  27, April, 2006; Poster given at
  Trust NSF Site Visit.

• BibTeX

  @presentation{ReiterSamar06_SelfserviceWithApplicationsToDistributedClassifier,
      author = {Michael K. Reiter, Asad Samar},
      title = {Self-service, with applications to distributed
                classifier construction},
      day = {27},
      month = {April},
      year = {2006},
      note = {Poster given at Trust NSF Site Visit.},
      abstract = {Motivated by the desire to efficiently construct
                global network traffic models to aid in traffic
                characterization and intrusion detection, we
                introduce a service implementation strategy called
                "self service" that scales by harnessing client
                resources in the implementation of the service. As
                its name indicates, self-service operates by
                migrating relevant portions of service state to
                each client, which performs its operations itself.
                As such, self-service is best suited for services
                whose state can be decomposed into small objects,
                of which client operations typically involve only
                a few. We will give an overview of the main
                algorithms that contribute to this strategy,
                including those for implementing serializable or
                strictly serializable operations, for surviving
                client disconnections, and for optimizing
                operation latency. We also describe the
                preliminary performance of an application that
                motivated this work, i.e., a service for
                constructing global network traffic models to aid
                in traffic characterization and intrusion
                detection.},
      URL = {http://www.truststc.org/pubs/88.html}
  }
  

Posted by Christopher Brooks on 4 May 2006.
Groups: trust
For additional information, see the Publications FAQ or contact webmaster at www truststc org.