# S/G Light-tree: Multicast Grooming Architecture for Improved Resource Allocation

- Authors
- Publisher
- Other IFIP Publications
- Publication Date
- Disciplines

## Abstract

Provisioning Lightpaths and Computing Resources for Location-Transparent Scheduled Grid Demands Hong-Ha Nguyen, Mohan Gurusamy Department of Electrical and Computer Engineering National University of Singapore {g0501100, [email protected] Luying Zhou Institute for Infocomm Research, Singapore [email protected] Abstract—In this paper, we define a new problem of provi- sioning lightpaths and computing resources for a set of location- transparent scheduled grid demands in optical grid networks. A location-transparent scheduled grid demand specifies only an amount of computing resources needed in a specified time inter- val to process input data. The network node generating a demand is called a client node. There are several network nodes which have sufficient resources for a demand. These nodes are called resource nodes. An algorithm is used to choose a resource node to reserve a specified amount of computing resources and provision a lightpath between the resource node and the client node. Given a set of location-transparent scheduled grid demands, it is required to provision the best lightpath (i.e. wavelength resources) as well as computing resources available during the specified time interval for each demand so as to optimize a certain objective function. In our work, we develop integer linear programming (ILP) formulations for 2 objective functions: 1) Given a network capacity, maximize the number of demands accepted; 2) Minimize the total number of wavelength-links to honor a given set of demands. Because the ILP algorithms are computationally expensive, we also develop heuristics to deal with large networks. The simulation results show that our heuristics achieve good performance. I. INTRODUCTION There are many large-scale applications which overwhelm the computing system’s capacity of an organization [1]. Hence, it needs the cooperation of computing devices from different organizations to run these applications. The grid can help to solve this problem. The grid, simply defi

## There are no comments yet on this publication. Be the first to share your thoughts.