Equalization and the impulsive MOOSE: fast adaptive signal recovery in very heavy tailed noise

Eugene Dubossarsky; Thomas R. Osborn; Sam Reisenfeld

Equalization and the impulsive MOOSE: fast adaptive signal recovery in very heavy tailed noise

Eugene Dubossarsky, Thomas R. Osborn, Sam Reisenfeld

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

Abstract

We combine Weighted Order Statistic (WOS) filters as efficient reconstructors of signals with extremely impulsive noise with the deconvolution capabilities of artificial neurons in a fast, memory efficient way. The Multi-rate Optimising Order Statistics Equalizer (MOOSE) neural architecture combines a FIR WOS Hybrid (FWH) pre-filter and a neural deconvolver with differing sampling rates. These cooperate synergistically after training. In the simulated equalization of a BAM channel with high ISI and additive noise. with Gaussian and frequent, high amplitude impulsive components. the linear MOOSE proved far superior to linear filtering. MOOSE computational complexity is far below that of the comparably performing L1 filter.

Original language	English
Title of host publication	Proceedings of the international workshop on applications of neural networks to telecommunications 3
Editors	Joshua Alspector, Rodney Goodman, Timothy X Brown
Place of Publication	Mahway, NJ
Publisher	Lawrence Erlbaum Associates
Pages	232-240
Number of pages	9
ISBN (Print)	0805829008
Publication status	Published - 1997
Externally published	Yes
Event	3rd International Workshop on the Application of Neural Networks to Telecommunications (IWANNT 97) - MELBOURNE, Australia Duration: 9 Jun 1997 → 11 Jun 1997

Publication series

Name	INTERNATIONAL NEURAL NETWORKS SOCIETY SERIES
Publisher	LAWRENCE ERLBAUM ASSOC PUBL
Volume	3

Conference

Conference	3rd International Workshop on the Application of Neural Networks to Telecommunications (IWANNT 97)
Country/Territory	Australia
City	MELBOURNE
Period	9/06/97 → 11/06/97

Cite this

Dubossarsky, E., Osborn, T. R., & Reisenfeld, S. (1997). Equalization and the impulsive MOOSE: fast adaptive signal recovery in very heavy tailed noise. In J. Alspector, R. Goodman, & T. X. Brown (Eds.), Proceedings of the international workshop on applications of neural networks to telecommunications 3 (pp. 232-240). (INTERNATIONAL NEURAL NETWORKS SOCIETY SERIES; Vol. 3). Lawrence Erlbaum Associates.

Dubossarsky, Eugene ; Osborn, Thomas R. ; Reisenfeld, Sam. / Equalization and the impulsive MOOSE : fast adaptive signal recovery in very heavy tailed noise. Proceedings of the international workshop on applications of neural networks to telecommunications 3. editor / Joshua Alspector ; Rodney Goodman ; Timothy X Brown. Mahway, NJ : Lawrence Erlbaum Associates, 1997. pp. 232-240 (INTERNATIONAL NEURAL NETWORKS SOCIETY SERIES).

@inproceedings{5da4f5d358174f098950a8d9a3fdeff4,

title = "Equalization and the impulsive MOOSE: fast adaptive signal recovery in very heavy tailed noise",

abstract = "We combine Weighted Order Statistic (WOS) filters as efficient reconstructors of signals with extremely impulsive noise with the deconvolution capabilities of artificial neurons in a fast, memory efficient way. The Multi-rate Optimising Order Statistics Equalizer (MOOSE) neural architecture combines a FIR WOS Hybrid (FWH) pre-filter and a neural deconvolver with differing sampling rates. These cooperate synergistically after training. In the simulated equalization of a BAM channel with high ISI and additive noise. with Gaussian and frequent, high amplitude impulsive components. the linear MOOSE proved far superior to linear filtering. MOOSE computational complexity is far below that of the comparably performing L1 filter.",

author = "Eugene Dubossarsky and Osborn, {Thomas R.} and Sam Reisenfeld",

year = "1997",

language = "English",

isbn = "0805829008",

series = "INTERNATIONAL NEURAL NETWORKS SOCIETY SERIES",

publisher = "Lawrence Erlbaum Associates",

pages = "232--240",

editor = "Joshua Alspector and Rodney Goodman and Brown, {Timothy X}",

booktitle = "Proceedings of the international workshop on applications of neural networks to telecommunications 3",

note = "3rd International Workshop on the Application of Neural Networks to Telecommunications (IWANNT 97) ; Conference date: 09-06-1997 Through 11-06-1997",

}

Dubossarsky, E, Osborn, TR & Reisenfeld, S 1997, Equalization and the impulsive MOOSE: fast adaptive signal recovery in very heavy tailed noise. in J Alspector, R Goodman & TX Brown (eds), Proceedings of the international workshop on applications of neural networks to telecommunications 3. INTERNATIONAL NEURAL NETWORKS SOCIETY SERIES, vol. 3, Lawrence Erlbaum Associates, Mahway, NJ, pp. 232-240, 3rd International Workshop on the Application of Neural Networks to Telecommunications (IWANNT 97), MELBOURNE, Australia, 9/06/97.

Equalization and the impulsive MOOSE: fast adaptive signal recovery in very heavy tailed noise. / Dubossarsky, Eugene; Osborn, Thomas R.; Reisenfeld, Sam.
Proceedings of the international workshop on applications of neural networks to telecommunications 3. ed. / Joshua Alspector; Rodney Goodman; Timothy X Brown. Mahway, NJ: Lawrence Erlbaum Associates, 1997. p. 232-240 (INTERNATIONAL NEURAL NETWORKS SOCIETY SERIES; Vol. 3).

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

TY - GEN

T1 - Equalization and the impulsive MOOSE

T2 - 3rd International Workshop on the Application of Neural Networks to Telecommunications (IWANNT 97)

AU - Dubossarsky, Eugene

AU - Osborn, Thomas R.

AU - Reisenfeld, Sam

PY - 1997

Y1 - 1997

N2 - We combine Weighted Order Statistic (WOS) filters as efficient reconstructors of signals with extremely impulsive noise with the deconvolution capabilities of artificial neurons in a fast, memory efficient way. The Multi-rate Optimising Order Statistics Equalizer (MOOSE) neural architecture combines a FIR WOS Hybrid (FWH) pre-filter and a neural deconvolver with differing sampling rates. These cooperate synergistically after training. In the simulated equalization of a BAM channel with high ISI and additive noise. with Gaussian and frequent, high amplitude impulsive components. the linear MOOSE proved far superior to linear filtering. MOOSE computational complexity is far below that of the comparably performing L1 filter.

AB - We combine Weighted Order Statistic (WOS) filters as efficient reconstructors of signals with extremely impulsive noise with the deconvolution capabilities of artificial neurons in a fast, memory efficient way. The Multi-rate Optimising Order Statistics Equalizer (MOOSE) neural architecture combines a FIR WOS Hybrid (FWH) pre-filter and a neural deconvolver with differing sampling rates. These cooperate synergistically after training. In the simulated equalization of a BAM channel with high ISI and additive noise. with Gaussian and frequent, high amplitude impulsive components. the linear MOOSE proved far superior to linear filtering. MOOSE computational complexity is far below that of the comparably performing L1 filter.

M3 - Conference proceeding contribution

SN - 0805829008

T3 - INTERNATIONAL NEURAL NETWORKS SOCIETY SERIES

SP - 232

EP - 240

BT - Proceedings of the international workshop on applications of neural networks to telecommunications 3

A2 - Alspector, Joshua

A2 - Goodman, Rodney

A2 - Brown, Timothy X

PB - Lawrence Erlbaum Associates

CY - Mahway, NJ

Y2 - 9 June 1997 through 11 June 1997

ER -

Dubossarsky E, Osborn TR, Reisenfeld S. Equalization and the impulsive MOOSE: fast adaptive signal recovery in very heavy tailed noise. In Alspector J, Goodman R, Brown TX, editors, Proceedings of the international workshop on applications of neural networks to telecommunications 3. Mahway, NJ: Lawrence Erlbaum Associates. 1997. p. 232-240. (INTERNATIONAL NEURAL NETWORKS SOCIETY SERIES).

Equalization and the impulsive MOOSE: fast adaptive signal recovery in very heavy tailed noise

Abstract

Publication series

Conference

Fingerprint

Cite this