BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Modeling data network sessions - Resnick\, S (Cornell)
DTSTART:20100111T153000Z
DTEND:20100111T163000Z
UID:TALK22500@talks.cam.ac.uk
CONTACT:Mustapha Amrani
DESCRIPTION:A session is a higher order entity resulting from amalgamating
  packets\, connections\, or groups of connections according to specified b
 ut not unique rules. For example\, using various rules\, the flow of packe
 ts past a sensor can be amalgamated into higher level entities called sess
 ions using a threshold rule based on gaps between packet arrivals. We rapi
 dly review some probability modeling based on sessions before turning to s
 tatistical analysis. Statistical analysis of these sessions based on packe
 ts is complex: session duration (D) and size (S) are jointly heavy tailed 
 but average transmission rate (R=S/D) is sometimes not heavy tailed and ar
 rival times of sessions is not Poisson. By segmenting sessions using a pea
 k rate covariate\, we find conditional on a peak rate decile\, within this
  decile segment session initiations can be modeled as Poisson. For modelin
 g the distribution of (D\,S\,R)\, the conditional extreme value (CEV) mode
 l may be a useful variant. (Joint work at various times with Jan Heffernan
 \, Bikramjit Das\, Luis Lopez-Oliveros\, T. Mikosch\, Bernardo D'Auria\, H
 . Rootzen\, A. Stegemen.) 
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR