A Faster and More Accurate Iterative Threshold Algorithm for Signal Reconstruction in Compressed Sensing.

Area of Science:

• Signal Processing
• Compressed Sensing
• Algorithm Optimization

Background:

• Compressed sensing (CS) enables signal reconstruction from undersampled data.
• The Fast Iterative Soft Threshold Algorithm (FISTA) is a common CS reconstruction method.
• Existing algorithms like FISTA face limitations in accuracy and efficiency for demanding applications.

Purpose of the Study:

• To develop an improved algorithm, FIPITA, for more accurate and efficient CS signal reconstruction.
• To address the limitations of FISTA by introducing novel mechanisms.
• To evaluate the performance of FIPITA against existing methods.

Main Methods:

• Proposed the Fast Iterative Parametric Improved Threshold Algorithm (FIPITA).
• Incorporated a modified threshold function to minimize reconstruction error.
• Implemented a restart adjustment mechanism to accelerate convergence.
• Optimized gradient parameters through impact assessment.

Main Results:

• FIPITA demonstrated a significant reduction in residual error compared to FISTA, RadaFISTA, and RestartFISTA (e.g., 6.35% lower residual rate in one case).
• FIPITA required substantially fewer iterations to reach minimum error (e.g., approximately 102 fewer iterations than FISTA).
• Simulation experiments on one-dimensional signals validated FIPITA's superior performance.

Conclusions:

• FIPITA offers enhanced accuracy and efficiency in compressed sensing signal reconstruction.
• The modified threshold function and restart mechanism are key contributors to FIPITA's improved performance.
• FIPITA represents a promising advancement for CS applications requiring high-fidelity signal recovery.