Simulación de la onda elástica anisotrópica en 3d mediante paralelización en gpu usando cuda para aplicaciones rfid con tecnología de onda acústica superficial (saw).

Rfid ("radio frequency identification") tags based on saw technology ("surface acoustic wave"), also known as rfid saw tags, have positioned themselves as a very attractive option for rfid technology due, above all, to their passive nature and without a chip, the large number of identification codes they can store and the hostile environment in which they can operate. due to the above, the demand for numerical methods capable of simulating this type of labels has increased, due to the need to have efficient numerical tools that help optimize the design of the labels. there are four numerical methods widely used to support the design of the saw rfid tags, however, three of them are analytical, while the other is semi-analytical, therefore,they must make use of restrictive simplifications that limit the versatility of the design. no full wave method has been previously proposed for the simulation of the saw rfid tags, only the fdtd procedure ("finitedifference time-domain") developed by the author of this proposal [1]. however, a limiting factor found in the development of this procedure was the high computational complexity of the simulated problem, considering that when discretizing the computational domain in 3d, cell numbers of the order of 109 were obtained. this has led to propose an extension of the developed method, in order to incorporate high performance computing techniques (hpc, for the acronym in english of "high performance computing") and parallel computing.this project intends to incorporate the mentioned techniques through the use of the nvidia parallel calculation architecture, called cuda, which makes use of the gpu's great processing power ("graphics processor unit").

To develop a fdtd procedure for the simulation of the propagation phenomenon of the 3d anisotropic elastic wave that takes place in the rfid saw tags, by means of parallelization in gpu, using the cuda parallel calculation architecture.

Article isi wos or scopus accepted. counseling in the graduate work of the student jhonatan amado of the master in modeling and computational science of the department of basic sciences of the university of medellin. it differs in the 3d full wave simulation of the rfid saw tag, which allows greater precision in the results and the consideration of all physical phenomena present. the advantage that would offer would be the possibility of simulating a rfid saw tag in 3d with reasonable resources and without applying limiting assumptions.