A VLSI Implementation of Independent Component Analysis for Biomedical Signal Separation Using CORDIC Engine

Area of Science:

• Signal Processing
• Integrated Circuit Design
• Biomedical Engineering

Background:

• Independent Component Analysis (ICA) is crucial for separating mixed signals.
• Super-Gaussian signals present unique separation challenges.
• Efficient hardware implementation of ICA algorithms is needed for real-time applications.

Purpose of the Study:

• To design and implement a Very Large Scale Integration (VLSI) chip for an extended InfoMax independent component analysis (ICA) algorithm.
• To enable the separation of super-Gaussian source signals using a hardware-efficient approach.
• To validate the chip's performance in separating mixed medical signals.

Main Methods:

• Utilized a time-sharing matrix multiplication array (MMA) to reduce circuit area for matrix operations.
• Employed the Coordinate Rotation Digital Computer (CORDIC) algorithm for hyperbolic function calculations (sinh(θ), cosh(θ)).
• Implemented a divider using CORDIC's linear coordinate rotation for tanh(θ) calculation.

Main Results:

• The proposed ICA chip was implemented using TSMC 90-nm CMOS technology.
• Achieved an operation frequency of 100 MHz with a gate count of 90.8K.
• Demonstrated successful separation of mixed medical signals into independent sources.

Conclusions:

• The developed VLSI chip offers an efficient hardware solution for ICA.
• The design effectively separates super-Gaussian signals, including complex medical data.
• This work contributes to advancing signal processing capabilities in biomedical applications.