Back to search

University of the Aegean

Μοντελοποίηση, πρόβλεψη και έλεγχος της κίνησης σε δίκτυα υπολογιστών με χρήση τεχνικών στατιστικής επεξεργασίας σημάτων

Abstract

dc:description

The aim of this dissertation is to model the traffic-rate series produced by a computer network node (hardware and its associated software) connected within a Greek Internet Service Provider (ISP) during a time interval in which network administrators have no access to traffic information. Typically, such intervals occur in industrial computer-network-nodes (routers) every 5 or 15 minutes according to their specifications. Providing information to the administrators more frequently may cause significant delay and affect adversely the computer-network-node‟s performance. The proposed model predicts the evolution of the number of users and their bandwidth demand within this 5-minute (15-minute) interval. The model is useful because it can indicate to the administrators‟ possible early corrective actions, thus improving overall Quality of Service (QoS). The modelling methodology is based on the concept of the mean reverting stochastic process with some modifications to accommodate for the behavior of users and computer network nodes within each time interval. The process is controlled adaptively by a proportional controller by regulating (varying) the capacity of computer-network-node. Moreover, another methodology is tested which models the evolution of users and their demand within the 5-minute (15-minute) interval. This is based on a first-order linear stochastic differential equation driven by random disturbances and an adaptive proportional integral derivative (PID) controller, used to control the stochastic process towards its estimate at the end of the time-interval. Finally, we present a methodology to determine adaptive estimates of computer-network-node‟s traffic-rate demand and its evolution during the 5-minute interval based on an autoregressive AR(n) model which uses a window-size of past real-time router traffic-rate data, with coefficients determined by a Kalman Filter. The evolution within the 5-minute interval to the AR(n) estimate is based on the modified mean reverting stochastic process introduced above with parametrically adapted coefficients, the mean of the process being the AR(n) estimate. For all the aforementioned models, simulations in MATLAB/Simulink environment were performed which verify their validity in terms of the underlying statistical characteristics which are similar to those of the original real computer-network-node traffic-rate data.

Degree

thesis:*
Grantor dc:publisher
University of the Aegean
Year dc:date
2021

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Giannopoulos, Iordanis
  • Γιαννόπουλος, Ιορδάνης

Subjects

dc:subject × 34

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/48994

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Giannopoulos, Iordanis; Γιαννόπουλος, Ιορδάνης. Μοντελοποίηση, πρόβλεψη και έλεγχος της κίνησης σε δίκτυα υπολογιστών με χρήση τεχνικών στατιστικής επεξεργασίας σημάτων. University of the Aegean, 2021. http://hdl.handle.net/10442/hedi/48994