System on Chip Implementation of Adaptive Moving Average Based Multiple-Model Kalman Filter for Denoising Fiber Optic Gyroscope Signal

Abstract

This paper proposes a combination of adaptive moving average process with multiple model kalman filter to efficiently denoise a digital Fiber Optic Gyroscope (FOG) signal. This algorithm has two phases i) Identification of transition of signal in a single frame of the signal ii) Filter the signal using a multiple model kalman filter. The transition locations are identified by comparing sample variance with a threshold value. Two different kalman filters are used to denoise the signal, one in the vicinity of transition region and other for non transition region. The performance of the algorithm is compared with adaptive moving average filter, standard kalman filter, standard multiple model kalman filter. Simulation results reveal that the proposed adaptive moving average based multiple model kalman filter (AMAMMKF) efficiently denoises the signal both in the transition and non-transition region. This paper also focuses on the system on chip (SoC) implementation of the proposed AMAMMKF algorithm in Virtex 5 FX70T1136-1 field programmable gate array (FPGA).