DocumentCode :
3100383
Title :
Code Design as an Optimization Problem: from Mixed Integer Programming to an Improved High Performance Randomized GRASP like Algorithm and from This One to an Improved Genetic Algorithm
Author :
da Fonseca, José Barahona
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., New Univ. of Lisbon, Monte de Caparica
fYear :
2006
fDate :
Nov. 28 2006-Dec. 1 2006
Firstpage :
180
Lastpage :
180
Abstract :
We begin to show that the design of optimum codes is a very difficult task by a set of preliminary brute force experiments where we generate all the possible optimum codes of a given length and minimum Hamming distance and then estimate the probability of finding one of these codes filling randomly the matrix that defines the code. Then we develop a novel approach to the code design problem based on the well known optimization technique of Mixed Integer Programming. Unfortunately the GAMS optimization software package limitation of 10 indexes imposes a limit of a maximum length 5 in the code to be designed. We show some results confirmed by the literature with this MIP model. Finally we develop a high performance randomized algorithm that surprisingly has much better runtimes than the MIP model.
Keywords :
encoding; genetic algorithms; integer programming; randomised algorithms; GAMS optimization software package; code design; improved genetic algorithm; minimum Hamming distance; mixed integer programming; optimization problem; optimum codes; randomized GRASP like algorithm; Algorithm design and analysis; Binary codes; Computational intelligence; Design optimization; Error correction codes; Filling; Genetic algorithms; Hamming distance; Linear programming; Software packages;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computational Intelligence for Modelling, Control and Automation, 2006 and International Conference on Intelligent Agents, Web Technologies and Internet Commerce, International Conference on
Conference_Location :
Sydney, NSW
Print_ISBN :
0-7695-2731-0
Type :
conf
DOI :
10.1109/CIMCA.2006.69
Filename :
4052802
Link To Document :
بازگشت