TY - GEN
T1 - Real-valued parallel clonal selection algorithm for design optimization in electromagnetics
AU - Bayraktar, Z.
AU - Bossard, J. A.
AU - Wang, X.
AU - Werner, Douglas Henry
PY - 2010
Y1 - 2010
N2 - Clonal Selection Algorithm (CLONALG) is one of the many branches of the Artificial Immune System Algorithms that are used to tackle engineering problems [1]. Inspired by the adaptive immune response of living organisms, CLONALG combines the clonal selection principle with the affinity maturation process, where those antibodies that successfully recognize the invading antigens are selected to proliferate [2]. Through evolution-based maturation of high affinity antibodies, the algorithm progresses towards finding a global solution to the optimization problem at hand. CLONALG has been shown to be very effective on multi-pole problems due to its inherent properties, yet it has not become a popular choice among optimization tools in the electromagnetics community. The primary reason for this is the large number of clones generated at each iteration, which increases the computational burden when used in conjunction with a full-wave electromagnetic solver. In this paper, we demonstrate that a successful parallel implementation of CLONALG would decrease the computational burden of multiple calls to the electromagnetic solver by distributing the work load over multiple processors. The optimization of a multi-layered frequency selective surface (FSS) filter for the X-band will be presented and compared with the more traditional parallel genetic algorithm (GA) optimizer in order to demonstrate the effectiveness of the parallel CLONALG implementation in the area of electromagnetics design.
AB - Clonal Selection Algorithm (CLONALG) is one of the many branches of the Artificial Immune System Algorithms that are used to tackle engineering problems [1]. Inspired by the adaptive immune response of living organisms, CLONALG combines the clonal selection principle with the affinity maturation process, where those antibodies that successfully recognize the invading antigens are selected to proliferate [2]. Through evolution-based maturation of high affinity antibodies, the algorithm progresses towards finding a global solution to the optimization problem at hand. CLONALG has been shown to be very effective on multi-pole problems due to its inherent properties, yet it has not become a popular choice among optimization tools in the electromagnetics community. The primary reason for this is the large number of clones generated at each iteration, which increases the computational burden when used in conjunction with a full-wave electromagnetic solver. In this paper, we demonstrate that a successful parallel implementation of CLONALG would decrease the computational burden of multiple calls to the electromagnetic solver by distributing the work load over multiple processors. The optimization of a multi-layered frequency selective surface (FSS) filter for the X-band will be presented and compared with the more traditional parallel genetic algorithm (GA) optimizer in order to demonstrate the effectiveness of the parallel CLONALG implementation in the area of electromagnetics design.
UR - http://www.scopus.com/inward/record.url?scp=78349296450&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=78349296450&partnerID=8YFLogxK
U2 - 10.1109/APS.2010.5562197
DO - 10.1109/APS.2010.5562197
M3 - Conference contribution
AN - SCOPUS:78349296450
SN - 9781424449682
T3 - 2010 IEEE International Symposium on Antennas and Propagation and CNC-USNC/URSI Radio Science Meeting - Leading the Wave, AP-S/URSI 2010
BT - 2010 IEEE International Symposium on Antennas and Propagation and CNC-USNC/URSI Radio Science Meeting - Leading the Wave, AP-S/URSI 2010
T2 - 2010 IEEE International Symposium on Antennas and Propagation and CNC-USNC/URSI Radio Science Meeting - Leading the Wave, AP-S/URSI 2010
Y2 - 11 July 2010 through 17 July 2010
ER -