A neural network tunable filter for multi-tone detection

Author

Rao, Sathyanarayan S. ; Sethuraman, Sriram

Author_Institution

Dept. of Electr. Eng., Villanova Univ., PA, USA

fYear

1992

fDate

11-14 Oct 1992

Firstpage

789

Abstract

A neural network that tunes to a band of frequencies depending on its inputs is proposed as a preprocessor for multiple sinusoid detection and estimation in additive noise. The multilayer perceptron network is trained off-line using the standard backpropagation algorithm. The authors provide a logical development of the problem and discuss the advantages of the proposed scheme. Potential applications in communications are cited, and Monte Carlo simulation results are presented

Keywords

digital filters; feedforward neural nets; signal detection; signal processing; tuning; Monte Carlo simulation; additive noise; backpropagation algorithm; communications; multilayer perceptron network; multiple sinusoid detection; multiple sinusoid estimation; multitone detection; neural network tunable filter; preprocessor; signal processing; Additive noise; Band pass filters; Data preprocessing; Filter bank; Frequency estimation; Frequency shift keying; Neural networks; Noise reduction; Signal processing; Signal to noise ratio;

fLanguage

English

Publisher

ieee

Conference_Titel

Military Communications Conference, 1992. MILCOM '92, Conference Record. Communications - Fusing Command, Control and Intelligence., IEEE

Conference_Location

San Diego, CA

Print_ISBN

0-7803-0585-X

Type

conf

DOI

10.1109/MILCOM.1992.243998

Filename

243998