PTP Boundary Clock Timestamp Adjustment via Pipeline Delay Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The packet processing pipeline in the master device is not configured to operate on the originTimestamp data field in the Sync message, making it impossible to store a timestamp representing the current time when the Sync message leaves the master device.
Innovation Solution
A network device performs operations to generate and send a Sync message that accommodates the limitations of the packet processing pipeline by storing a HW timestamp (T1′) in the originTimestamp data field and adjusting the correctionField data field to account for pipeline processing time, allowing the slave device to compute the correct time when the Sync message left the master device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the packet processing pipeline is used to send Sync messages, then message transmission efficiency is improved, but the ability to store and operate on the originTimestamp data field is lost
Solution Approach 1:
The patent introduces an intermediary mechanism where the kernel process acts as a mediator between the packet processing pipeline and the PTP synchronization function. The kernel process captures the timestamp (T1′) at the point where the Sync message enters the pipeline, stores it in the originTimestamp field, and then allows the message to proceed through the optimized pipeline without requiring the pipeline itself to have timestamping capability. This resolves the contradiction by providing timestamping functionality through an intermediary rather than requiring the pipeline to directly support it.
2Measurement precision
If the HW timestamp T1′ is stored in the originTimestamp data field, then the timestamp represents the time when the Sync message enters the pipeline, but it does not represent the time when the message leaves the master device
Solution Approach 1:
The patent implements a feedback mechanism where the kernel process not only stores the entry timestamp T1′ in the originTimestamp field but also calculates the pipeline processing duration and communicates this information back to adjust the correctionField. The slave device uses this feedback information along with its own reception timestamp to compute the actual departure time T1, thereby recovering the lost information about when the message actually left the master device despite using the entry timestamp for storage.
Solution Approach 2:
The patent changes the parameter representation by storing T1′ (entry timestamp) in the originTimestamp field while simultaneously modifying the correctionField to account for pipeline processing time. This parameter transformation allows the system to work with available hardware timestamps while still achieving accurate departure time calculation through mathematical adjustment, effectively converting an incomplete parameter set into a complete time synchronization solution.
Data Source
AI summary
A Sync message to start off a 1-step synchronization sequence includes a timestamp (T1′) that represents a time when the Sync message entered the packet processing pipeline of a network device to be processed for egress to a slave device. The packet processing pipeline stores a timestamp (Tduration) in the Sync message as part of processing the Sync message. The timestamp Tduration represents the amount of time spent in the packet processing pipeline. The slave device that receive Sync message computes T1 by summing the timestamps T1′ and Tduration stored in the received Sync. message, where T1 represents the time when the Sync message left the network device.