Continuous and Parallel Optimization of Dynamic Bandwidth Scheduling in WDM Networks

Pragatheeswaran Angu, Byrav Ramamurthy
December 2010
Cite DOI URL

Abstract

Many scientific and Grid applications require high-speed circuits of guaranteed bandwidth for scheduled transfers. Offline optimization of dynamic scheduled bandwidth demands is an efficient way of finding the near-optimal solution to the bandwidth scheduling problem. In this paper, we propose a continuous and parallel optimization method to address the dynamic and deterministic bandwidth scheduling problem in next generation wavelength-division multiplexing (WDM) networks. In this method a greedy algorithm and genetic algorithm are run in parallel in separate threads and both of them take the Dynamic Scheduled Bandwidth Demand (D-SBD) as their input. The user gets his response only from the greedy algorithm and hence he will get a deterministic answer in a short amount of time. The genetic algorithm takes as one of its inputs the output of the greedy algorithm and does the optimization of the D-SBDs with minimizing blocking probability as its fitness function. The greedy algorithm copies the optimized reservation database of the genetic algorithm at regular intervals. The user submitting a D-SBD request is unaware of the optimization done by the genetic algorithm. This method is evaluated using both trace-driven simulation of real network traffic from the DOE ESnet network and stochastic traffic in ESnet network topology and a 24 node network topology. We also compare our approach with an earlier proposed method called re-optimization at blocking. Adding the genetic algorithm improves the performance of the network (in terms of blocking probability) compared to using only the greedy approach or the re-optimization at blocking method.

Type
Conference paper
Publication
2010 IEEE Global Telecommunications Conference GLOBECOM 2010
Bandwidth Telecommunication Traffic Wavelength Division Multiplexing WDM Networks Probability Telecommunication Network Topology Dynamic Scheduling Greedy Algorithms Optimization 24 Node Network Topology Bioinformatics Blocking Probability Minimization Continuous Optimization D-SBD Databases Deterministic Algorithms Deterministic Bandwidth Scheduling DOE ESnet Network Dynamic Scheduled Bandwidth Demand ESnet Network Topology Genetic Algorithms Genomics High-Speed Circuit Minimization Next Generation Networks Next Generation Wavelength-Division Multiplexing Networks Offline Optimization Parallel Optimization Real Network Traffic Stochastic Processes Stochastic Traffic Trace-Driven Simulation
Byrav Ramamurthy
Byrav Ramamurthy
Professor & PI

My research areas include optical and wireless networks, peer-to-peer networks for multimedia streaming, network security and telecommunications. My research work is supported by the U.S. National Science Foundation, U.S. Department of Energy, U.S. Department of Agriculture, NASA, AT&T Corporation, Agilent Tech., Ciena, HP and OPNET Inc.