Title :
Parameter optimization for importance sampling in encoded systems
Author :
Melo, Hallyson L M ; Gurjão, Edmar C. ; Albert, Bruno B. ; De Assis, Francisco M.
Author_Institution :
Fed. Univ. of Campina Grande, Campina Grande
Abstract :
In this paper we present a new methodology based on the stochastic gradient descent to optimizing parameters of the biased distribution along importance sampling simulations. A particular aspect of the new technique is that of use of multiple parameters for the simulation of each codeword. An example of application estimating the performance of a encoded system is given.
Keywords :
encoding; gradient methods; importance sampling; optimisation; stochastic processes; biased distribution; codeword; encoded systems; importance sampling simulations; parameter optimization; stochastic gradient descent; Bit error rate; Computational modeling; Decoding; Discrete event simulation; Distribution functions; Error probability; Monte Carlo methods; Optimization methods; Performance analysis; Stochastic systems; Importance Sampling; Optimization; Simulation;
Conference_Titel :
Telecommunications Symposium, 2006 International
Conference_Location :
Fortaleza, Ceara
Print_ISBN :
978-85-89748-04-9
Electronic_ISBN :
978-85-89748-04-9
DOI :
10.1109/ITS.2006.4433249
