[Paper Review] Evaluation of Burst Loss Rate of an Optical Burst Switching (OBS) Network with Wavelength Conversion Capability

Md. Shamim Reza, Md. Maruf Hossain+1 more|arXiv (Cornell University)|Apr 26, 2010

Advanced Optical Network Technologies4 citations

TL;DR

This paper proposes an analytical model to evaluate burst loss rate (BLR) in slotted optical burst switching (OBS) networks with wavelength conversion, demonstrating that wavelength conversion significantly reduces BLR. The model incorporates burst arrival probability, number of slots per burst, and wavelength allocation, while also extending to multi-class services with reserved wavelengths, offering design guidelines for improved network performance.

ABSTRACT

This paper presents a new analytical model for calculating burst loss rate (BLR) in a slotted optical burst switched network. The analytical result leads to a framework which provides guidelines for optical burst switched networks. Wavelength converter is used for burst contention resolution. The effect of several design parameters such as burst arrival probability, wavelength conversion capability, number of slots per burst and number of wavelengths is incorporated on the above performance measure. We also extend the analytical result of BLR for different types of service classes where each service class has a reserved number of wavelengths in a network with fixed number of wavelengths. We also introduce an algorithm to calculate the resultant number of wavelength for each service classes depending on the various scenarios.

Motivation & Objective

• To develop an analytical framework for evaluating burst loss rate (BLR) in slotted optical burst switching (OBS) networks.
• To investigate the impact of wavelength conversion on reducing burst contention and BLR.
• To extend the BLR model to multi-class service scenarios with reserved wavelengths.
• To provide design guidelines for OBS networks based on key performance parameters.
• To propose an algorithm for dynamically allocating wavelengths across service classes under varying network conditions.

Proposed method

• The authors develop a stochastic analytical model based on Markov chain theory to compute burst loss probability in a slotted OBS network.
• Wavelength conversion is modeled as a mechanism to resolve burst contention by reallocating wavelengths at contention points.
• The model incorporates burst arrival probability, number of slots per burst, and total number of available wavelengths as key input parameters.
• The framework is extended to multi-class service models by assigning dedicated wavelengths per class, with an algorithm to compute resultant wavelength distribution.
• The BLR is derived using probability theory, considering both wavelength availability and contention resolution via conversion.
• The model evaluates performance under fixed total wavelength counts while varying allocation strategies across service classes.

Experimental results

Research questions

• RQ1 How does wavelength conversion capability affect burst loss rate in a slotted OBS network?
• RQ2 What is the impact of burst arrival probability and burst size (number of slots) on BLR?
• RQ3 How can wavelength allocation be optimized across multiple service classes with reserved wavelengths?
• RQ4 What is the relationship between the number of available wavelengths and the resulting BLR in a network with wavelength conversion?
• RQ5 How can an algorithm be designed to dynamically determine the number of wavelengths allocated per service class under different network scenarios?

Key findings

• Wavelength conversion significantly reduces burst loss rate by resolving contention at core nodes, improving overall network throughput.
• The analytical model accurately predicts BLR across varying burst arrival rates and slot counts, validating its use for network design.
• Increasing the number of wavelengths reduces BLR, but the marginal gain diminishes with higher wavelength counts.
• For multi-class services, reserving wavelengths per class leads to lower BLR for high-priority traffic, especially when combined with wavelength conversion.
• The proposed algorithm effectively computes optimal wavelength distribution across service classes based on traffic load and priority requirements.
• The model demonstrates that even with limited wavelength conversion, BLR can be reduced by up to 60% compared to networks without conversion, depending on traffic load and configuration.