Title :
A comparison between recurrent neural network architectures for digital equalization
Author :
Ortiz-Fuentes, Jorge D. ; Forcada, Mikel L.
Author_Institution :
Dept. Llenguatges i Sistemas Inf., Alicante Univ., Spain
Abstract :
This paper shows a comparison between three different first-order recurrent neural network (RNN) architectures (fully recurrent, partially recurrent, and Elman (1990)), trained using the real-time recurrent learning (RTRL) algorithm and the GSM training sequence ratio (26/114) for digital equalization of 2-ary PAM signals. The results show no substantial effect of the particular architecture or the number of units on the overall performance. This is due to the assumption of a suboptimal equalization scheme by the RNNs, because of the learning algorithm. The results are compared to those obtained using a classical (decision-feedback equalizer) approach
Keywords :
adaptive equalisers; cellular radio; digital radio; learning (artificial intelligence); neural net architecture; pulse amplitude modulation; recurrent neural nets; signal reconstruction; 2-ary PAM signals; DFE; Elman neural network; GSM training sequence ratio; adaptive equalizers; decision-feedback equalizer; digital equalization; first-order recurrent neural network; fully recurrent neural network; partially recurrent neural network; performance; real-time recurrent learning algorithm; recurrent neural network architectures; signal reconstruction; suboptimal equalization; Adaptive equalizers; Bandwidth; Distortion; Filters; GSM; Informatics; Intersymbol interference; Neural networks; Propagation delay; Recurrent neural networks;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1997. ICASSP-97., 1997 IEEE International Conference on
Conference_Location :
Munich
Print_ISBN :
0-8186-7919-0
DOI :
10.1109/ICASSP.1997.595494
