Title :
A genotype-to-phenotype mapping for microstructured polymer optical fibres
Author_Institution :
Sch. of Math. & Stat., Univ. of Sydney, Sydney, NSW, Australia
Abstract :
Although glass fibres are standard in long-distance telecommunications; customised polymer microstructured optical fibres play a more significant role in many diverse new short distance applications. Our prototyping process involves drilling an array of holes in a cylindrical preform. That preform is subsequently heated and pulled into a narrow fibre. The size and position of the holes create an effective refractive index profile, which in turn determines the optical transmission properties of the fibre. In this paper, a new variable-length genotype is introduced which controls the coordinates of the centres of ´potential´ holes. The genotype-to-phenotype mapping carefully determines which holes are ´activated´ and the final radius of each hole, consistent with manufacturing constraints. Two manufacturing constraints are: a minimum spacing between adjacent holes and that the drill bits are only available in a discrete range of radii. A cross-over operator is designed that works with variable-length genotypes and its effect on the distribution of genome lengths is explored in detail. An implementation of NSGA-II is used to perform a multi-objective optimisation with four objectives. One of these objectives (wanting a parabolic index profile) is the same as in our previous work. Two are new: minimising the deformability of the design and minimising the detrimental effects of surface roughness. The final objective is not optical but related to the efficiency of the GA and is to minimise the number of inactive genes. The behaviour of the genetic algorithm and a number of interesting designs are discussed.
Keywords :
holey fibres; optical fibre fabrication; optical polymers; NSGA-II; cross-over operator; genetic algorithm; genotype-to-phenotype mapping; glass fibres; long-distance telecommunications; microstructured polymer optical fibres; multi-objective optimisation; optical transmission properties; parabolic index; polymer microstructured optical fibres; refractive index; surface roughness; Bioinformatics; Biological cells; Genomics; Optical refraction; Optical variables control; Preforms;
Conference_Titel :
Evolutionary Computation (CEC), 2011 IEEE Congress on
Conference_Location :
New Orleans, LA
Print_ISBN :
978-1-4244-7834-7
DOI :
10.1109/CEC.2011.5949643