Clock Recovery via Frequency Prediction During Network Quiet Periods
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Clock synchronization over packet networks is challenged by packet delay variations (PDV) that cause high jitter and wander, especially during high network traffic, leading to unreliable clock recovery in legacy services.
Innovation Solution
A method that utilizes quiet periods with low network traffic to train a frequency prediction unit, allowing it to predict frequency updates during high traffic periods, using an accept-rejection algorithm to select between predicted and actual values for robust clock recovery, thereby minimizing the impact of packet delay variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If timing packets are used for clock synchronization over packet networks, then cost and availability are improved, but packet delay variations cause high jitter and wander in the recovered reference clock
Solution Approach 1:
The system performs preliminary actions by detecting quiet periods in advance and training the frequency prediction unit during these low-traffic periods. This preliminary training enables the system to predict frequency updates during high-traffic periods without waiting for actual timing data, thus resolving the contradiction between using packet networks (low cost) and maintaining synchronization quality (high reliability).
Solution Approach 2:
The frequency prediction unit acts as an intermediary between the timing recovery circuit and the local oscillator. During high-traffic periods when packet delay variations cause high jitter, the frequency prediction unit interpolates expected frequency updates based on trained models, mediating between the unreliable timing packets and the requirement for stable clock synchronization.
2Reliability
If clock recovery algorithm stops during high traffic periods and enters hold-over mode, then reliability is improved by avoiding poor synchronization, but phase and frequency drift occur due to fixed value updates
Solution Approach 1:
The system maintains continuous clock recovery operation by switching between timing recovery circuit output during quiet periods and frequency prediction unit output during high-traffic periods. This continuous operation eliminates the need to stop and enter hold-over mode, while still maintaining synchronization reliability through intelligent selection based on network conditions.
Solution Approach 2:
The system dynamically switches between two operational modes (timing recovery circuit and frequency prediction unit) based on real-time detection of network traffic conditions. This dynamic adaptation allows the system to maintain both reliability and stability by using the appropriate method for each condition, rather than stopping entirely during high-traffic periods.
3Reliability
If timing packets are filtered to reduce packet delay variation impact, then synchronization quality is improved, but device complexity increases due to additional filtering mechanisms
Solution Approach 1:
The frequency prediction unit performs self-service by automatically detecting quiet periods and training itself using timing packet data during these periods. Once trained, it autonomously predicts frequency updates during high-traffic periods without requiring complex external filtering mechanisms, thus achieving high synchronization quality while minimizing added complexity.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A method of recovering timing information in a packet network, involves detecting quiet periods in the packet network when network packet delay variation (PDV) is low. A frequency prediction unit is trained during the quiet periods to learn output clock variations of a timing recovery unit to permit the frequency prediction unit to predict frequency update values for a local oscillator during non quiet periods taking into account the historical output clock variations during quiet periods. The output of the frequency prediction unit is used as the active frequency update values during non quiet periods.