Please use this identifier to cite or link to this item: http://localhost:8080/xmlui/handle/123456789/1250
Full metadata record
DC FieldValueLanguage
dc.contributor.authorR. Chandrasekharam, V.V. Vinod-
dc.contributor.authorS. Subramanian-
dc.date.accessioned2024-10-29T09:49:24Z-
dc.date.available2024-10-29T09:49:24Z-
dc.date.issued1994-
dc.identifier.urihttp://localhost:8080/xmlui/handle/123456789/1250-
dc.description.abstractThe embedding of a complete graph in a minimum sized hypercube is an important problem which models the classical state encoding problem of Finite State Machines (FSMs). As this problem is an NP-hard optimization problem, acceptable final solutions are generally obtained by employing heuristic methods or Simulated Annealing (SA). In this paper the efficacy of a Genetic Algorithm (GA) for this problem is studied. This study includes a comparison of three different crossover methods of GA along with their implementation details and their suitability for this embedding problem. The experimental results on a number of MCNC benchmark FSMs indicate the superiority of GA in finding a better (near optimal) solution than a heuristic solution. These results experimentally establish the time efficiency of GA over SA for this embedding problem.en_US
dc.description.sponsorshipNITWen_US
dc.language.isoenen_US
dc.publisherElsevier Science Ltden_US
dc.subjectGenetic algorithmen_US
dc.subjecthypercubeen_US
dc.titleGenetic algorithm for embedding a complete graph in a hypercube with a VLSI applicationen_US
dc.typeArticleen_US
Appears in Collections:Computer Science and Engineering

Files in This Item:
File Description SizeFormat 
0165-60742990100-7.pdf1.06 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.