International Society for Industrial Process Tomography

FPGA Implementation of LMS and NLMS Adaptive filters for Electrical Impedance Tomography System

Al Amin Saichul IMAN^1*, Marlin R. Baidillah¹, M. Takei¹

¹Grad. School of Science and Engineering, Div. Fundamental Engineering, Dept. Mechanical Eng., Chiba University, Chiba, 263-8522, Japan

^*Email: amin3vdc@gmail.com

ABSTRACT

Elimination of noise signal is desired in order to achieve high sensitivity detection in Field Programmable Gate Array (FPGA)-based electrical impedance tomography (EIT) system. In order to eliminate the noise signal, an adaptive filter has been implemented on FPGA-based EIT system. Two adaptive filter, i.e. Least Mean Square (LMS) and Normalized Least Mean Square (NLMS) algorithms, are evaluated in the experiments through the reconstructed image performances and the power spectrum density (PSD). The PSD of a time series describes the distribution of power of adaptive filter performance into frequency components composing that signal. The result showed that the ADC performances of NLMS has PSD = -147 dBV²/Hz and effective number of bit (ENOB) = 11.2 bit, meanwhile LMS has PSD = -139 dBV²/Hz and ENOB = 9 bit. This higher ADC performances of NLMS as compared with LMS leads to the better image reconstruction performances in terms of amplitude response (AR), position error (PE), resolution (RES), and shape deformation (SD). This study contributes that the implementation of adaptive filter for FPGA-based EIT system enhances the sensitivity detection of EIT system.

Keywords >Electrical impedance tomography, power spectra density, adaptive filter, Zynq, FPGA.

Industrial Application General