Low Power implementation of FIR Filter for De-noising the EOG signal

Abstract

In this paper, we propose a modified FIR filter using canonical signed digit representation and compressor techniques for de-noising the Electrooculography signal (EOG). A dipole, formed between cornea and retina in an eye, resulting in voltage difference with occurrence of eye movement, which in turn generates an electrical signal. These electrical signals, called Electrooculography (EOG) signal and are useful in many medical and bioelectrical applications. To remove the noise in EOG signal an enhanced 16-bit FIR filter is used. The proposed FIR architecture uses canonical signed digit based multipliers and 3:2 or 4:2 compressors to achieve greater improvement in power and delay compared to conventional shift-add method of multiplication. Canonic signed digit (CSD) is used in order to reduce the design complexity of the multiplication by representing the filter coefficients in CSD format. Instead of using Carry Look Ahead adders, 4:2 compressors are used. The proposed approach will be implemented on FPGA development kit.