Cubic spline interpolation with overlapped window and data reuse for on-line Hilbert Huang transform biomedical...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Computer Engineering

Background:

• Real-time analysis of non-linear, non-stationary biomedical signals is crucial for wearable and implantable sensors.
• The Hilbert-Huang Transform (HHT) is a key technique for this analysis.
• Cubic Spline Interpolation (CSI) is computationally intensive within HHT, causing latency issues.

Purpose of the Study:

• To develop an on-chip HHT implementation that meets real-time requirements.
• To reduce the computational load of CSI in HHT processors.
• To enable accurate signal analysis on resource-constrained wearable and implantable devices.

Main Methods:

• Proposed an on-line processing method for CSI using small, overlapped data windows.
• Implemented data reuse strategies between consecutive windows.
• Focused on optimizing the CSI component of the HHT algorithm for hardware.

Main Results:

• Achieved significant computational savings in CSI: 58% reduction in multiplications and 73% in divisions.
• Maintained interpolation accuracy despite using smaller, overlapping data windows.
• Demonstrated feasibility of real-time HHT on-chip processing.

Conclusions:

• The proposed on-line CSI method effectively reduces latency and computational cost for on-chip HHT.
• This approach enhances the practicality of HHT for real-time biomedical signal analysis in wearable and implantable sensors.
• Optimized HHT implementation enables efficient processing of complex biomedical data.