However, ever more crucial in every sector hit by the recent and sudden wave of technological innovation, clock synchronization remains a topic that is still too neglected. This is demonstrated by the poor performance of the solutions adopted by most organizations, which will become a brake on their competitiveness in the short term. In the energy sector, where intelligent grids cannot count on truly efficient synchronization, and in all areas where the 5G-IoT combination could revolutionize dynamics. Efficiency and business.
What Does Clock Synchronization Mean?
It is above all in computer science and engineering that we talk and have always talked about clock synchronization as the process aimed at coordinating otherwise independent clocks. Suppose it may seem exaggerated to give importance to this alignment in everyday life. In that case, it is not at small scales where management difficulties and problems associated with clock distortion can occur, even more, significant in distributed computing where different computers have to realize the same time. Global.
In a network of sensors, for example, the information taken from each of them is valid and usable only if accompanied by a homogeneous time reference, and the traditional methods of clock synchronization are proving to be increasingly inadequate, making evident the need to search for new solutions, often specific for each context.
Why Synchronization Is Important
Before exploring how clock synchronization technologies and methods are evolving, it is necessary to clarify why this issue will play a central role on a global level now and, above all, in the future. The impact of a failure to develop adequate solutions for wireless networks would undoubtedly be significant. It would affect various sectors that would pay dearly for the sudden inefficiency of these networks and the functions they have entrusted to them
From the agri-food industry, which uses them to monitor soils, bacteria, and production, to health that keeps vital parameters in the presence of pathologies or physical exertion, or to the automotive and, in general, to industrial plants that entrust regulation, control to wireless networks process, database and statistical analysis, without neglecting the military, environmental and domestic sectors, with the growing smart home market.
However, clock synchronization is not a priority only for wireless networks: in all modern computer networks, it is already fundamental because every aspect of management, security, planning, and debugging involves determining when events happen. Time provides the only frame of reference, also essential for:
- Track security breaches: Simply using the network or monitoring its components can become impossible if the timestamps in the logs are not accurate. Time is the critical factor that allows an event on one network node to be associated with a corresponding event on another.
- Reduce the clutter in shared file systems – no matter what machine they’re on, it’s essential that change times are consistent.
- Releasing financial services, from billing to advanced trading which requires very accurate timekeeping by law
- Efficient use of the 5G spectrum: operators can maximize the investment made by being able to use more fractions of the spectrum and serving more customers
What Types Of Clock Synchronization Exist
When we enter into the merits of clock synchronization solutions, we cannot take for granted the meaning of the term time, to be defined with two characteristics: the frequency, as the repetition of events such as the beats of a pendulum, and the phases, because such beats, albeit with equal frequency, could alternate.
In some contexts, the synchronization may concern only the frequency, in others also the phase, and up to this point, an atomic clock would be enough. In most current applications, however, as we will see, clock synchronization in its most complete conception must be considered, which also includes the correspondence of the time of day, which requires the use of GPS or GNSS.
How To Choose The Clock Synchronization Algorithm
Phase, frequency, and time synchronization are crucial for many distributed computing applications. There are many protocols and algorithms proposed to make a network of devices synchronized, including the Network Time Protocol (NTP) devised in the early 1980s to synchronize a computer network and the Precision Time Protocol (PTP) introduced with the IEEE 1588 protocol in 2002 for local systems that require higher accuracy than that achievable with NTP.
There are some standard metrics to be used to evaluate the different synchronization algorithms, bearing in mind that the evaluated performances also depend on the platform and implementation.
- Accuracy: it is linked to the synchronization error of each node but represents a global parameter, the maximum or average value of the instantaneous precision iGuidan, a time interval.
- Complexity: level of work performed by the network to maintain the synchronization service, directly proportional to the number of messages transmitted and to the messages processed by the individual nodes
- Channel occupation: an estimate of the congestion of the communication medium, therefore the degree of utilization of the transmission channel to maintain synchronization
- Dynamic memory: portion used by the implemented algorithm must be limited to allow the node to perform operations for which it was used.
- Scalability: the ability of the algorithm to withstand variations in the number of nodes that require synchronization.
- Dependence on the network topology and robustness to failures: level of independence for the arrangement of the nodes in space and from any failures suffered by the nodes
How To Synchronize A Clock
In a wide range of sectors, from finance to energy services, from defense to the media, passing through manufacturing, clock synchronization represents and will increasingly represent one of the most critical IT services. A very high level of accuracy is required, both for business, regulatory and security reasons. In the evolution of this process, passing from simple synchronization signals to more complex IP protocols, such as PTP, a precision of less than one microsecond has been achieved.
With this solution, the GPS source time information is distributed in complex IP packets via a link shared with all network traffic. It is, therefore, essential to have a well-designed network infrastructure to reduce both traffic congestion and IP packet jitter. A clock and signal management platform could be helpful to improve the synchronization performance further. Still, it is then necessary to monitor the clock synchronization so that it continues to ensure accuracy. In this case, some tricks can be helpful.
- Monitor the quality of the source, be it GPS or GNSS
- Plot the clock and time metrics
- Measure clock performance based on parameters such as jitter, path delay, and offset
- Identify changes and deviations to identify and resolve potential problems promptly.
- Monitor the underlying infrastructure such as switches, SFPs, laser bias current …
- Create and enrich the allowlist of systems that can become grandmaster clocks
- Set the PTP role on the switch ports to discard PTP advertisement messages from poorly configured agent devices.
- Configure “slave only” to prevent it from becoming a master.
Despite all these efforts, it must be admitted that in some areas, primarily in wireless sensor networks, neither NTP or PTP are future-proof. Even the latter, as the number of network nodes increases, encounters problems of scalability and more: this is evident by exploring some particularly current and decisive use cases for the economy of many countries, including Italy.
Clock Synchronization In The 5G-IoT Era
Until now, the 1μs timing accuracy ensured by highly stable GNSS synchronized grandmaster mobile networks supported by atomic clocks associated with a packet network with at least partial on-path PTP support has been sufficient but will be for a bit longer. Localization services today require a time difference between base stations in the order of 100ns to identify a user device with sufficient levels of accuracy.
Mobile radio access networks deployed in a distributed way need precise timing in all sites and end-user applications to ensure high levels of efficiency on edge and in the cloud. Even in the industrial field, where IoT and 5G in the coming months will give more concrete results than ever, they will be able to do so only with carefully synchronized control applications. Clock synchronization must make a significant leap forward but fortunately can be broken down into several steps:
- Improve the performance of boundary clocks and transparent clocks so that they better compensate for the transient delay
- To raise the holdover capabilities of atomic clocks by switching from magnetic cesium to optical cesium
- Switch to multiband GNSS receivers to minimize accuracy problems by compensating for variations in atmospheric delay.
- Combine network-based synchronization with a satellite-based backup
- Use sophisticated synchronization probes to ensure higher synchronization quality
- Create a parallel network using dedicated optical channels to limit synchronization problems on packet networks related to jitter and asymmetric delay.