Quantization for distributed estimation using neural networks

Authors:
Vasileios Megalooikonomou;Yaacov Yesha
Affiliations:
Department of Computer and Information Sciences, Temple University, 1805 N. Broad Street, 303 Wachman Hall, Philadelphia, PA;Department of Computer Science and Electrical Engineering, University of Maryland, Baltimore County, 1000 Hilltop Circle, Baltimore, MD
Venue:
Information Sciences—Applications: An International Journal
Year:
2002

Citing 4
Cited 0

Next century challenges: scalable coordination in sensor networks

MobiCom '99 Proceedings of the 5th annual ACM/IEEE international conference on Mobile computing and networking
Directed diffusion: a scalable and robust communication paradigm for sensor networks

MobiCom '00 Proceedings of the 6th annual international conference on Mobile computing and networking
MEMS-based integrated-circuit mass-storage systems

Communications of the ACM
Smart Dust: Communicating with a Cubic-Millimeter Computer

Computer

Quantified Score

Hi-index	0.00

Visualization

Abstract

We propose a neural network approach for the problem of quantizer design for a distributed estimation system with communication constraints in the case where the global observation model is unknown and one must rely on a training set. Our method applies a variation of the Cyclic Generalized Lloyd Algorithm (CGLA) on every point of the training set and then uses a neural network for each quantizer to represent the training points and their associated codewords. The codeword of every training point is initialized using a previously proposed regression tree approach. Simulation results show that there is an improvement of the proposed approach over using regression trees only.