Quantitative analysis of Elastic Optical Networks using Probabilistic and Support Vector Regression Algorithms

Abstract

Optical signals are transferred at a very high speed in Elastic Optical Network. The impairment of path between optical nodes leads to physical challenges of allocating paths. In this work an intuitive interpretation of paths and its blocking probability has been done in both cases of directed and undirected models. The purpose of Structured Probability Routing Estimation in Optical Networks (SPREON) is to describe the best probability path with route estimation. Significance in data transfer is achieved by considering direct and indirect interactions of the nodes reducing the blocking probability. Unstructured Probability Routing Estimation in Optical Networks (UPREON) group node based on position for data transfer and estimates the routes among the peer nodes. The third work denotes Support Vector Regression Routing Estimation in Optical Networks (SVRREON) is to convert the feature space of resources available between communicating nodes into a linear classifier using Support Vector machine. Thus the non linear process of data transfer is being transformed in linear classification and optical routing and feasible routes are estimated. Results are compared independently across the model and also with existing work for delay, blocking probability metrics.