Low-power and Low-cost Implementation of SVMs for Smart Sensors

Author

Boni, Andrea ; Gasparini, Leonardo ; Pianegiani, F. ; Petri, Dario

Author_Institution

Dipt. di Informatica e Telecomunicazioni, Universita degli Studi di Trento, Italy

Volume

1

fYear

2005

fDate

16-19 May 2005

Firstpage

603

Lastpage

607

Abstract

In this paper, we propose an efficient implementation of SVMs on a low-power and low-cost 8-bit microcontroller that can be applied to design smart sensors, sensor networks and in the area of pervasive computing, where intelligent data analysis is required, such as pattern classification, signal estimation and so on. A new model selection algorithm to extract the optimal number of free parameters and an optimized implementation which exploits the CORDIC algorithm are detailed and discussed with examples and simulation results.

Keywords

intelligent sensors; low-power electronics; microcontrollers; support vector machines; 8 bit; CORDIC algorithm; SVM; intelligent data analysis; low-cost implementation; low-power implementation; microcontroller; optimized implementation; pattern classification; signal estimation; smart sensors; support vector machines; Computational modeling; Data analysis; Data mining; Estimation; Intelligent networks; Intelligent sensors; Microcontrollers; Pattern classification; Pervasive computing; Signal design; Support Vector Machines (SVMs); model selection; smart sensors;

fLanguage

English

Publisher

ieee

Conference_Titel

Instrumentation and Measurement Technology Conference, 2005. IMTC 2005. Proceedings of the IEEE

Conference_Location

Ottawa, Ont.

Print_ISBN

0-7803-8879-8

Type

conf

DOI

10.1109/IMTC.2005.1604187

Filename

1604187