Compressed Signal Reconstruction With Overlapping Windows
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current compressed sensing systems face a trade-off between reconstruction window size, signal reconstruction accuracy, and processing complexity, leading to increased latency and computational complexity, making them inadequate for embedded applications.
Innovation Solution
The system employs a method with a compressive sampling module and a signal reconstruction module that uses overlapping measurement windows and discrete cosine transforms to calculate reconstruction windows, reducing latency and complexity while maintaining high reconstruction accuracy, by iteratively calculating and substituting values between overlapping segments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If long reconstruction windows of samples are used to improve signal reconstruction accuracy, then reconstruction quality improves, but processing complexity and latency increase significantly
Solution Approach 1:
The patent divides the long reconstruction window into multiple overlapping segments or blocks. Each segment is processed independently through the reconstruction algorithm, and the results are combined. This segmentation reduces the computational complexity from O(N³) for the entire window to O((N/M)³ × M) = O(N³/M²) for M segments, while the overlapping ensures continuity and accuracy across segment boundaries.
Solution Approach 2:
The patent performs preliminary processing by pre-calculating and storing certain transformation matrices or basis functions that are reused across multiple segments. This preliminary action reduces the repeated computational overhead in each segment processing, thereby lowering overall complexity while maintaining reconstruction accuracy.
2Measurement precision
If long reconstruction windows are used to improve signal reconstruction accuracy, then reconstruction quality improves, but latency increases
Solution Approach 1:
By segmenting the reconstruction window, the patent enables parallel processing of multiple segments. The overlapping segments can be processed concurrently or with minimal sequential dependency, significantly reducing the overall reconstruction latency compared to processing a single long window sequentially.
Solution Approach 2:
The overlapping segments ensure that processing can continue continuously without waiting for complete window accumulation. Each segment can be processed as soon as sufficient overlapping data is available, maintaining continuous useful action and reducing idle time, thereby lowering latency while preserving accuracy through the overlap region.
3Measurement precision
If overlapping measurement windows are used with iterative calculation and value substitution, then reconstruction accuracy improves for shorter window sizes, but computational steps increase
Solution Approach 1:
The patent applies local quality by performing iterative calculation and value substitution only in the overlapping regions of adjacent segments, rather than reprocessing entire segments. This localized refinement focuses computational effort where it is most needed (at segment boundaries) while accepting approximate values in non-overlapping regions, thereby improving accuracy with minimal additional computational steps.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach achieves signal reconstruction accuracy comparable to longer window sizes with reduced latency and computational complexity, eliminating reconstruction artifacts and improving efficiency for biosignal processing.
Implementation Method 1
calculating a corresponding first plurality of reconstruction windows based on said first plurality of measurement windows and calculating a first version of a reconstructed signal based on said first plurality of reconstruction windows
Data Source
AI summary
A system for compressed sensing comprising: a compressive sampling module configured for providing a CS-sampled signal and a signal reconstruction module configured for receiving and allocating a first plurality of measurement windows comprising a number of samples from the CS-sampled signal, calculating a corresponding first plurality of reconstruction windows based on the first plurality of measurement windows and calculating a first version of a reconstructed signal based on the first plurality of reconstruction windows. The signal reconstruction module is also configured to allocate a second plurality measurement windows overlapping in content with the first plurality of measurement windows, calculate a second plurality of reconstruction windows based on the second plurality of measurement windows, calculate a second version of the reconstructed signal based on the second plurality of reconstructed windows and generate a reconstructed signal based on values from the first version and the second version of the reconstructed signal.