Optical Self-Similar Cluster Switching (OSCS) – A Novel Optical Switching Scheme by Detecting Self-Similar Traffic

This paper proposes a novel framework for bandwidth provisioning based on detecting self-similar traffic. The method is called Optical Self-similar Cluster Switching (OSCS). The objective of OSCS is to detect potential characteristics of self-similar traffic in communication networks such that network resources can be statistically multiplexed in presence of self-similarity in data-dominant traffic. In the paper, the concept of a self-similar cluster in a traffic stream is first defined by identifying two properties serving as bases for the study. It is followed by a detailed description of the proposed strategies for data burst classification and assignment under the OSCS framework. In fact, the fundamental principle of OSCS is to utilize the partial predictable nature of a self-similar cluster to compensate the unpredictable or high-variability nature of self-similar traffic, which is a root reason of network performance deterioration. Based on both simulation and analysis conducted for verifying the proposed framework, the results reveal that the objective of OSCS is perfectly realized by compensating non-predictability of traffic self-similarity, where self-similar clusters are set up for a partial prediction of the burst arrival. Furthermore, network resources have been statistically multiplexed by OSCS under self-similar traffic, and the flooding effect in networks caused by self-similar traffic has been successfully smoothed out by OSCS.