Optimized Precision Time Source and Time Synchronization Simulation

GPS time sources have been broadly accepted but are easily blocked, spoofed, and jammed. For data centers, precious time sources and corresponding time synchronization are expected. Existing time synchronization algorithms, such as Berkeley’s algorithm and the Network Time Protocol (NTP), do not perform well when unstable network delays. Developing a new time synchronization that balances time accuracy and energy consumption is critical, especially in large data centers.We propose a 1024-node simulation with computing nodes, network traffic, and router delays in this reach. We innovate a new approach to identifying stable time sources. The preliminary results show that the new idea could effectively decrease instability from 63.34 microseconds to 21.92 microseconds (reduced by 65.39%) for a 13-minute duration between two connected nodes, outperforming the NTP algorithm.Thirdly, we put forth a novel algorithm for time synchronization. Compared to NTP, the novel algorithm demonstrates a 37.96% reduction in the average time error, an 18.30% improvement in energy efficiency, and a 1.54% increase in total energy consumption. Furthermore, experimental tests demonstrate that the new algorithm reduces the average time error by 10.70%. The new algorithm is anticipated to be a foundation for more advanced time synchronization algorithms. Additionally, our research could be utilized as a fallback solution in data centers when GPS time and atomic sources are unavailable.