Alignment Detection Circuit for Multi-Lane Network Interfaces
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for alignment detection in high-speed Ethernet interfaces, such as those defined in IEEE 802.3ba and 802.3bj standards, face challenges with high silicon area and power consumption when using parallel correlators for quick alignment, or slow alignment with low silicon and power costs when using a single correlator.
Innovation Solution
An alignment detection circuit comprising a buffer, candidate selection circuit, and correlator circuit that identifies candidate data blocks with symmetry indicative of alignment markers, allowing for efficient detection and location of alignment markers with reduced latency and complexity, using a single correlator and pre-screening to filter out false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a large bank of parallel correlators is used for alignment detection, then alignment speed is improved (around 200 μs), but silicon area and power consumption increase significantly
Solution Approach 1:
The alignment detection process is segmented into two distinct stages: a correlation stage that processes multiple candidate positions in parallel using a single correlator, and a verification stage that validates potential alignment markers. This segmentation allows the system to achieve fast alignment speed without requiring a large bank of parallel correlators, as the correlation stage efficiently narrows down candidates before verification is needed.
Solution Approach 2:
The system performs preliminary correlation operations at multiple candidate positions simultaneously using a single correlator, then uses the results to identify and verify alignment markers. This preliminary action reduces the search space before final verification, achieving fast alignment without the silicon area cost of parallel correlator banks.
2Speed
If a large bank of parallel correlators is used for alignment detection, then alignment speed is improved (around 200 μs), but power consumption increases significantly
Solution Approach 1:
The alignment detection process is segmented into two distinct stages: a correlation stage that processes multiple candidate positions in parallel using a single correlator, and a verification stage that validates potential alignment markers. This segmentation allows the system to achieve fast alignment speed without requiring a large bank of parallel correlators, as the correlation stage efficiently narrows down candidates before verification is needed.
Solution Approach 2:
The system performs preliminary correlation operations at multiple candidate positions simultaneously using a single correlator, then uses the results to identify and verify alignment markers. This preliminary action reduces the search space before final verification, achieving fast alignment without the power consumption cost of parallel correlator banks.
3Area of stationary object
If a single correlator is swept across the data stream for alignment detection, then silicon area and power consumption are reduced, but alignment speed decreases (on the order of 10 ms worst case)
Solution Approach 1:
The system performs preliminary correlation operations at multiple candidate positions simultaneously using a single correlator, then uses the results to identify and verify alignment markers. This preliminary action reduces the search space before final verification, achieving fast alignment without the silicon area cost of parallel correlator banks.
Solution Approach 2:
Instead of sweeping through data sequentially in one dimension, the system evaluates multiple candidate positions in parallel across different positions in the data stream. This dimensional approach to correlation allows simultaneous processing of multiple locations, dramatically improving alignment speed while using a single correlator.
4Use of energy by stationary object
If a single correlator is swept across the data stream for alignment detection, then silicon area and power consumption are reduced, but alignment speed decreases (on the order of 10 ms worst case)
Solution Approach 1:
The system performs preliminary correlation operations at multiple candidate positions simultaneously using a single correlator, then uses the results to identify and verify alignment markers. This preliminary action reduces the search space before final verification, achieving fast alignment without the silicon area cost of parallel correlator banks.
Solution Approach 2:
Instead of sweeping through data sequentially in one dimension, the system evaluates multiple candidate positions in parallel across different positions in the data stream. This dimensional approach to correlation allows simultaneous processing of multiple locations, dramatically improving alignment speed while using a single correlator.
Data Source
AI summary
In an example implementation, an alignment detection circuit includes a buffer, a candidate selection circuit, and a correlator circuit. The buffer is configured to receive a data stream from a data lane, the data stream including alignment markers delineating data frames, each of the alignment markers having a predefined bit pattern. The candidate selection circuit is configured to identify candidate data blocks in successive data blocks of the data stream provided by the buffer, each of the candidate blocks having a measure of symmetry satisfying a threshold metric indicative of the predefined bit pattern. The correlator circuit is configured to search for at least one of the alignment markers in each of the candidate blocks and adjust alignment of the data stream in the buffer in response to locating the at least one alignment marker.