FPGA-Based Weighted DTW Framework with Hybrid Gait Symmetry Index for Real-Time Wearable Gait Classification

Area of Science:

• Rehabilitation Engineering
• Biomedical Signal Processing
• Wearable Computing

Background:

• Gait symmetry analysis is crucial for assessing mobility impairments and neurological disorders.
• Existing methods struggle with gait variations and irregular walking patterns, limiting clinical utility.
• Need for robust, real-time gait analysis systems for wearable applications.

Purpose of the Study:

• To develop a robust, low-latency, wearable edge-computing framework for real-time gait analysis.
• To integrate gait symmetry variability, statistical features, and Dynamic Time Warping (DTW) for enhanced gait classification.
• To implement the system on an FPGA for efficient, energy-saving computation.

Main Methods:

• A hybrid feature vector combining DTW similarity, gait symmetry index (GSI), and statistical gait descriptors was used.
• Classification of gait patterns into five categories: normal, slow, medium, fast, and abnormal.
• Wearable edge-computing platform with an NI myRIO, IMU, FPGA, and ARM processor for real-time processing and communication.

Main Results:

• The FPGA-based architecture achieved an end-to-end processing latency of approximately 4 ms at 100 MHz.
• Real-time signal preprocessing, feature extraction, and gait classification were performed onboard.
• Successful transmission of real-time gait data to a remote workstation via WiFi.

Conclusions:

• The proposed FPGA-based gait analysis system offers low-latency, energy-efficient, and real-time performance.
• The framework enhances robustness against gait variations and irregular walking patterns.
• This technology is well-suited for wearable rehabilitation systems, assistive devices, and continuous mobility monitoring.