PAM Symbol Group Trellis Encoding for Lower-Error SerDes Links
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-speed SerDes channels face significant impairments such as insertion loss, noise, inter-symbol interference, and non-ideal decision feedback equalizer behavior, which hinder achieving stringent symbol error rate targets and necessitate complex receiver architectures, limiting system performance and reliability.
Innovation Solution
Integrate a convolutional trellis encoder with a PAM-modulated symbol alphabet optimized for high-speed SerDes channels, associating specific symbol groups with trellis state transitions to enhance coding gain, reduce symbol error rate, and mitigate DFE-induced error propagation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional PAM modulation is used in high-speed SerDes channels, then the system can operate at high data rates, but the symbol error rate increases due to channel impairments such as insertion loss, noise, and inter-symbol interference
Solution Approach 1:
The patent segments the PAM symbol transmission into groups of symbols that are independently encoded and decoded. Each symbol group is associated with a specific trellis state transition, allowing the receiver to process and correct errors within each group independently, thereby maintaining high data rates while reducing the overall symbol error rate through grouped error correction.
Solution Approach 2:
The patent implements decision feedback equalization (DFE) where the receiver uses feedback from previously decoded symbol groups to correct current symbol decisions. The DFE coefficients are adapted based on channel conditions, creating a feedback mechanism that continuously improves decoding accuracy and reduces symbol error rates in high-speed SerDes channels.
2Reliability
If decision feedback equalization is used to compensate for channel impairments, then symbol error rate decreases, but error propagation occurs due to non-ideal DFE behavior
Solution Approach 1:
The patent divides the symbol stream into discrete groups that are processed independently through the trellis encoder. Each group is associated with a specific trellis state transition, which isolates errors within each group and prevents them from propagating to subsequent groups. This segmentation approach maintains the benefits of DFE while minimizing error propagation effects.
Solution Approach 2:
The patent incorporates forward error correction coding within each symbol group, providing a cushion against errors before they can propagate. The trellis-based FEC code adds redundancy that allows the receiver to detect and correct errors within each group, preventing non-ideal DFE behavior from causing widespread error propagation.
3Reliability
If complex receiver architectures are implemented to address channel impairments, then symbol error rate improves, but device complexity increases
Solution Approach 1:
The patent segments the receiver functionality into modular components: a symbol group detector that identifies trellis state transitions, a FEC decoder that processes each group independently, and a DFE unit that handles equalization. This segmentation allows each component to be optimized and implemented with simpler logic, reducing overall device complexity while maintaining high reliability through coordinated operation of the modular blocks.
4Reliability
If symbol groups are associated with trellis state transitions, then coding gain increases, but the mapping complexity increases
Solution Approach 1:
The patent pre-establishes a lookup table that maps trellis state transitions to corresponding PAM symbol groups. This preliminary action eliminates the need for complex real-time calculations during data transmission. The receiver simply queries the lookup table using the detected trellis state transition, significantly reducing mapping complexity while maintaining high coding gain through optimized symbol group selection.
Data Source
AI summary
A method may include receiving eight bits of information; determining a state transition of a convolutional trellis at least partially based on a two-bit portion of the eight bits of information; determining a group of PAM-modulated symbols at least partially based on the determined state transition of the convolutional trellis, the eight bits of information, and a set of predetermined groups of PAM-modulated symbols pre-associated with the determined state transition of the convolutional trellis; and encoding the eight bits of information into the determined group of PAM-modulated symbols.