Forney-Based Symbol Decisions With Confidence for High-Speed SerDes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-speed communication systems face challenges in achieving low-cost and reliable hard-decision decoding with corresponding soft-decision outputs due to the complexity of existing soft-decision MLSE algorithms, which are cost-prohibitive for high-speed architectures, and Forney-based approaches fail to provide soft-decision estimation or confidence levels for hard-decision symbols.
Innovation Solution
Implementing a Forney-based decision generation component that generates hard-decision estimations for received digital signals and calculates a confidence level using a Forney algorithm, enabling parallel processing in high-speed SerDes receivers through a block-based approach.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If soft-decision MLSE algorithms (e.g., SOVA) are employed to generate confidence levels for hard-decision symbols, then BER performance is improved, but device complexity and implementation cost become prohibitive for high-speed architectures
Solution Approach 1:
The patent replaces the complex soft-decision MLSE algorithm with a simpler, more cost-effective approach using the Forney algorithm combined with a saturation detector. This substitution uses less complex computational objects (simple error value calculations and saturation detection) to achieve the same functional goal of providing confidence levels for FEC decoding, making high-speed implementation feasible.
Solution Approach 2:
The patent changes the computational parameters and approach by using the Forney algorithm to calculate error values and comparing these against saturation thresholds, rather than implementing the full soft-decision MLSE algorithm. This parameter change transforms an intractable computational problem into a manageable one suitable for high-speed serial interfaces.
2Device complexity
If Forney-based decision generation is used to reduce complexity, then device complexity is reduced, but soft-decision estimation or confidence level output is not provided
Solution Approach 1:
The patent merges the Forney algorithm (which provides error detection with reduced complexity) with a saturation detector mechanism (which generates confidence level information). This combination integrates the benefits of both approaches: the computational efficiency of Forney-based methods and the confidence level output capability of soft-decision algorithms, thereby preventing loss of information while maintaining low complexity.
Solution Approach 2:
The saturation detector acts as an intermediary component that takes the error values from the Forney algorithm and transforms them into confidence level information. This intermediary element bridges the gap between the simple Forney error detection and the need for soft-decision confidence levels, enabling both functions to coexist without direct implementation of complex soft-decision MLSE.
3Reliability
If high-order partial response equalization is employed to overcome high insertion loss, then BER performance is improved, but bandwidth requirements increase and device complexity increases
Solution Approach 1:
The patent changes the equalization approach from high-order partial response (which requires narrow bandwidth and complex processing) to a combination of standard PR equalization with enhanced decision generation using saturation detection. This parameter change in the equalization strategy maintains BER performance while reducing bandwidth constraints and simplifying the overall system complexity.
Data Source
AI summary
A receiver including an equalization component to receive a signal comprising a sequence of samples corresponding to symbols and generate an equalized signal with an estimated sequence of symbols corresponding to the signal. The receiver further includes a decision generation component to detect that an aggregate error level associated with the equalized signal exceeds a saturation threshold level. The decision generation component identifies a set of errors including a first error associated with a first symbol having a highest error level and a last error associated with a last symbol. The decision generation component generates, based on the equalized signal, a decision including a sequence of one or more bits that represent each symbol of a first subset of the sequence of symbols and a confidence level corresponding to the decision, where the confidence level is based at least in part on a distance between an error level of each symbol and a level of the first error.