Compressed Domain Threshold Filter Using Steering Vector
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computing systems face inefficiencies in processing compressed domain data, as they often require evaluating entire data samples, leading to increased power consumption and hardware requirements, especially when only a subset of the data contains useful information.
Innovation Solution
A filtering accelerator that employs a threshold filtering method using a steering vector to filter out unnecessary compressed data samples without full evaluation, calculating a covariance matrix to update the steering vector, and sorting elements to optimize the filtering process, thereby reducing power, time, and hardware needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If entire data samples are evaluated, then data processing completeness is improved, but power consumption and hardware requirements increase
Solution Approach 1:
The patent applies partial action by evaluating only the most significant components of compressed data samples using a steering vector derived from the covariance matrix, rather than processing the entire data sample. This allows the system to make filtering decisions based on partial evaluation, reducing power consumption while maintaining sufficient processing completeness for effective data analysis.
Solution Approach 2:
The system performs preliminary action by pre-calculating the covariance matrix and deriving a steering vector before actual data filtering occurs. This steering vector serves as a threshold criterion that enables rapid decision-making during data processing, avoiding the need for full evaluation of every sample and thereby reducing real-time power consumption requirements.
2Reliability
If entire data samples are evaluated, then data processing completeness is improved, but hardware requirements increase
Solution Approach 1:
The patent extracts the essential information needed for filtering by computing a steering vector from the covariance matrix, which captures the most significant characteristics of the data distribution. This extracted steering vector serves as a simplified threshold criterion that replaces the need for complex full-sample evaluation hardware, reducing hardware requirements while maintaining processing effectiveness.
Solution Approach 2:
The system transforms the complex task of evaluating entire compressed data samples into a simpler parameter comparison by converting data characteristics into a steering vector representation. This parameter transformation allows filtering decisions to be made using straightforward threshold comparisons rather than complex sample evaluations, reducing hardware complexity requirements.
3Measurement precision
If full evaluation of compressed data samples is performed, then processing accuracy is improved, but processing time increases
Solution Approach 1:
The patent applies partial action by evaluating only the most significant components of compressed data samples using a steering vector, rather than processing the entire data sample. This allows the system to make filtering decisions based on partial evaluation, reducing processing time while maintaining sufficient accuracy for effective data analysis.
Solution Approach 2:
The system performs preliminary action by pre-calculating the covariance matrix and deriving a steering vector before actual data filtering occurs. This steering vector serves as a threshold criterion that enables rapid decision-making during data processing, avoiding the need for full evaluation of every sample and thereby reducing processing time while maintaining accuracy.
Data Source
AI summary
In an embodiment, a processor includes a compression domain threshold filter coupled to a plurality of cores. The compression domain threshold filter is to: receive a sample vector of compressed data to be filtered; calculate, based at least on a first subset of the elements of the sample vector, an estimated upper bound value of a dot product of the sample vector and a steering vector; determine whether the estimated upper bound value of the dot product satisfies a filter threshold value; and in response to a determination that the estimated upper bound value of the dot product does not satisfy the filter threshold value, discard the sample vector without completion of a calculation of the dot product of the sample vector and the steering vector. Other embodiments are described and claimed.