Path Metric Calculation for High-Speed Maximum Likelihood Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-speed maximum likelihood detectors face challenges in reducing the complexity and area required for path metric computation, particularly in systems like E2PR4 detectors, due to the need for simultaneous computation of numerous branch metrics, which is time and area consuming.
Innovation Solution
The method simplifies path metric calculation by utilizing relations between branch metrics of paths on the trellis with common initial or final states, allowing for the reduction of independent calculations and area needed for detector chips, using components that are independent of received data or paths to pre-calculate parts of the path metric difference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If simultaneous computation of numerous branch metrics is performed for high-speed detection, then detection speed is improved, but computational complexity and area consumption increase
Solution Approach 1:
The patent segments the path metric computation into two distinct components: a first component that is independent of the specific paths and received data, and a second component that depends on paths and received data. This segmentation allows the first component to be pre-calculated and stored, while only the second component requires simultaneous computation during detection, thereby reducing the computational complexity and area consumption while maintaining high detection speed.
2Speed
If simultaneous computation of numerous branch metrics is performed for high-speed detection, then detection speed is improved, but area consumption increases
Solution Approach 1:
The patent performs preliminary computation and storage of the first component of the path metric difference, which is independent of specific paths and received data. By pre-calculating and storing this component in a lookup table or memory structure, the system eliminates the need to re-compute it during simultaneous detection operations, thereby reducing the area consumption for computational hardware while maintaining high detection speed.
3Device complexity
If path metric calculation is simplified by using relations between branch metrics, then computational complexity is reduced, but calculation accuracy may be affected
Solution Approach 1:
The patent creates a computational model that copies the structure of the path metric calculation while separating it into two components. The first component is copied and stored in advance, and the second component is computed during detection. This copying approach maintains the mathematical accuracy of the original path metric calculation while reducing computational complexity, as the copied first component is reused across multiple detection operations without loss of precision.
Data Source
AI summary
A calculator for use in a maximum likelihood detector, including: a receiver for receiving convolution encoded data which may include noise; first calculator for calculating a first component of a first path metric difference between two possible sequences of states corresponding to the convolution encoded data, the two sequences each having a length equal to a constraint length of the convolution encoded data, and the two sequences starting at a same state and ending at a same state, adapted to calculate the first component using the convolution-encoded data and using convolution encoding parameters of the convolution-encoded data, wherein the first component is independent of the two sequences; and second calculator for calculating a second component of the first path metric difference using the two sequences, wherein the second component is independent of the convolution encoded data; and using the first and second components to obtain the first path metric difference.