A memetic algorithm for distributed object allocations

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We propose a memetic algorithm based approach for allocating objects in distributed computing environment. Our allocation objective is to allocate objects to different servers so that the number of inter-server communications can be minimized. Assuming servers with similar processing and storage capabilities - to avoid biased allocation of the objects on servers with higher performance - we formulate the problem of allocating distributed objects as a graph bisection problem. Using simulated data, we use the memetic algorithm to solve the graph bisection problem. We compare the performance of the memetic algorithm with the Kernighan and Lin (KL) heuristic and semidefinite programming (SDP) lower bounds for graph bisection problem. The results of our experiments show that the memetic algorithm performs better or equal to the KL heuristic and has a relative performance gap of between 2-5% from the SDP lower bounds.

Original languageEnglish (US)
Title of host publication2010 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2010
Pages140-144
Number of pages5
DOIs
StatePublished - 2010
Event2010 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2010 - Montreal, QC, Canada
Duration: Jul 6 2010Jul 9 2010

Publication series

NameIEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM

Other

Other2010 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2010
Country/TerritoryCanada
CityMontreal, QC
Period7/6/107/9/10

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering
  • Control and Systems Engineering
  • Computer Science Applications
  • Software

Fingerprint

Dive into the research topics of 'A memetic algorithm for distributed object allocations'. Together they form a unique fingerprint.

Cite this