Departmental Papers (ESE)

Document Type

Conference Paper

Date of this Version

December 2002

Comments

Copyright 2002 IEEE. Reprinted from Proceedings of 41st IEEE Conference on Decision and Control 2002, Volume 2, pages 2366-2371.
Publisher URL: http://ieeexplore.ieee.org/xpl/tocresult.jsp?isNumber=26567&page=14

This material is posted here with permission of the IEEE. Such permission of the IEEE does not in any way imply IEEE endorsement of any of the University of Pennsylvania's products or services. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by writing to pubs-permissions@ieee.org. By choosing to view this document, you agree to all provisions of the copyright laws protecting it.

Abstract

We consider resource allocation in a system which must process and subsequently deliver jobs from N input terminals to N output terminals as per matching constraints. Jobs are stored in a common memory before processing and transfer. The resource allocation objective is to minimize the job loss due to memory overflow. We present optimal scheduling and memory management policies for attaining this objectives for N = 2 and symmetric traffic. We identify certain characteristics of the optimal strategy for N = 2 and asymmetric traffic, and present a near optimal heuristic for this case. We obtain a heavy traffic lower bound for the optimal discounted cost function for the general case of arbitrary N and arbitrary traffic. We use this lower bound to propose a heuristic strategy whose performance is close to the optimal strategy in this case. The policies proposed in this paper substantially outperform the existing strategy for the system, which was designed and proved to be optimal under the assumption of infinite storage.

Share

COinS
 

Date Posted: 24 November 2004

This document has been peer reviewed.