Top Banner
Bandwidth Allocation in Self-Sizing Networks André C. Drummond Prof. Nelson L. S. da Fonseca
5

Bandwidth Allocation in Self-Sizing Networks André C. Drummond Prof. Nelson L. S. da Fonseca.

Jan 15, 2016

Download

Documents

Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Bandwidth Allocation in Self-Sizing Networks André C. Drummond Prof. Nelson L. S. da Fonseca.

Bandwidth Allocation in Self-Sizing Networks

Bandwidth Allocation in Self-Sizing Networks

André C. Drummond

Prof. Nelson L. S. da Fonseca

Page 2: Bandwidth Allocation in Self-Sizing Networks André C. Drummond Prof. Nelson L. S. da Fonseca.

Measurement Based Equivalent Bandwidth Estimators Evaluation

Measurement Based Equivalent Bandwidth Estimators Evaluation

FEP – Fractal Brownian Motion Envelope Process based Estimator:

GA – Gaussian Approximation Allocation Estimator:

Page 3: Bandwidth Allocation in Self-Sizing Networks André C. Drummond Prof. Nelson L. S. da Fonseca.

Numeric Results – FEP vs GA Estimative

Numeric Results – FEP vs GA Estimative

Voice (Exponential)

● 400 fonts● 12 Mbps● Hurst 0.6

Data (Pareto)

● 15 Mbps● Hurst 0.75

Video (Real traces)

● 180 fonts● 75 Mbps● Hurst 0.9

FEP is robust and better adapts to different network scenario!

Page 4: Bandwidth Allocation in Self-Sizing Networks André C. Drummond Prof. Nelson L. S. da Fonseca.

Bandwidth Allocation in Self-Sizing Networks

Bandwidth Allocation in Self-Sizing Networks

The Band Partitioning Problem inherits some uncertainties:● Bandwidth estimators uncertainties;● Network traffic fluctuations;

A Multi-Objective Fuzzy Optimization Model was proposed.

Page 5: Bandwidth Allocation in Self-Sizing Networks André C. Drummond Prof. Nelson L. S. da Fonseca.

Numeric Results – Classic vs Fuzzy ModelNumeric Results – Classic vs Fuzzy Model

Fuzzy model packet loss values are 54% lower.

Fuzzy model delay values are 33% lower.

Fuzzy model allocates more 9% of link capacity.

The Fuzzy Model provides more flexibility to cope with sudden increases on traffic!