Multi-Stream Noise Predictive Filtering for Accurate Data Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing systems face inaccuracies in data transfer due to noise filtering issues when using multiple sensors, leading to unreliable data processing outcomes.
Innovation Solution
A multi-stream noise predictive filter circuit system that generates interim outputs from non-matrix based combinations of data inputs and previous instances, combined using a summation circuit to yield a branch metric, which is then used for data detection and decoding, employing adaptive coefficient sets and linear computations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple sensors are used to sense data on the storage medium, then measurement precision is improved, but noise filtering capability deteriorates leading to inaccurate results
Solution Approach 1:
The patent divides the noise filtering process into separate stages: individual stream processing with dedicated noise predictive filters for each sensor stream, followed by a combining stage. Each filter operates independently on its respective data stream using non-matrix based combinations, allowing tailored noise reduction for each sensor while preserving the benefits of multiple sensors for improved measurement precision.
Solution Approach 2:
The patent employs adaptive coefficient sets that dynamically adjust filter parameters based on noise characteristics. The noise predictive filters use linear computations with coefficients that can be calibrated to match specific noise conditions, enabling the system to adaptively optimize noise filtering effectiveness while maintaining accurate data sensing across multiple sensor streams.
2Device complexity
If matrix based combination methods are used to process multiple data streams, then noise filtering is simplified, but computational complexity increases
Solution Approach 1:
The patent extracts the noise filtering function from the data combination process by applying dedicated noise predictive filters to each individual data stream before combining them. This separation allows the use of simpler non-matrix based combination methods (such as weighted sums or voting mechanisms) while maintaining effective noise reduction, thereby reducing computational complexity compared to matrix based approaches.
Solution Approach 2:
The patent employs computationally efficient non-matrix based combination operations that require less processing power than matrix multiplications. These simpler combination methods use basic arithmetic operations and logical functions that are faster and less resource-intensive, making them suitable for real-time data processing applications where computational speed is critical.
3Measurement precision
If advanced noise filtering is applied to multiple data streams, then data processing accuracy is improved, but device complexity increases
Solution Approach 1:
The patent segments the noise filtering architecture into modular noise predictive filter circuits, each handling a specific data stream independently. This modular approach allows for standardized, reusable filter blocks that can be replicated for each sensor stream without proportionally increasing overall system complexity. Each module uses non-matrix based combinations and linear computations that are computationally tractable and easier to implement than monolithic advanced filtering approaches.
Data Source
Figure 1a
Figure 1b
Figure 1c~3
AI summary
The present invention is related to systems and methods for branch metric calculation based on multiple data streams (108a, 108b, 108c) in a data processing circuit (100). The branch metric is calculated based on a set of mutiple stream noise predictive filter circuits (125).