[Dry goods] NTP time synchronization server technology details
[Dry goods] NTP time synchronization server technology details
A.1 Time synchronization principle
The principle of time synchronization is to adjust the internal clock and time of the device according to the received time. After calibrating the time to seconds, the principle of time synchronization is similar to the principle of frequency synchronization on the clock. It regulates both the frequency and phase of the clock. At the same time, the phase of the clock is expressed as a value, that is, the time. Different from frequency synchronization, time synchronization accepts discontinuous time information, and discontinuously regulates the device clock, that is, the adjustment control of the device clock phase-locked loop is periodic, and its period corresponds to the period of time acquisition, and is related to the adjustment method, The accuracy of the clock is related to its stability.
A.2 Time definition
When planning and designing a time synchronization network, the following terms are often mentioned in terms of time concepts: mean solar day, universal time, international atomic time, coordinated universal time, leap second, etc. These terms are explained and defined separately below.
a) Average solar day
People are used to determining the time based on the position of the sun on the celestial sphere, but because the earth's orbit around the sun is an ellipse, the apparent movement speed of the true sun on a weekday is uneven (that is, the true solar time is uneven) ). In order to obtain a time measurement system based on the apparent movement of the real sun on a weekday but overcoming its unevenness, the concept of an average solar day was introduced. The basic unit of mean solar time is the mean solar day. 1 mean solar day is equal to 24 mean solar hours, and 1 mean solar hour is equal to 86,400 mean solar seconds.
b) Universal time (UT0/UT1/UT2)
Greenwich (the seat of the original Greenwich Observatory in the southern suburbs of London, England, which is the starting point of the world's geographic longitude), which starts at midnight at 0 o'clock, is called Universal Time. There is a strict conversion relationship between world time and sidereal time. People get the average world time by observing stars with astronomical telescopes all over the world, and its accuracy can only reach 10-9. Due to the unevenness of the movement of the earth’s poles and the rotation of the earth, the world time obtained at the beginning is also uneven, and we will record it as UT0; people add the pole shift correction to UT0, and the result obtained is recorded as UT1; plus the earth The empirical correction for the seasonal variation of the rotation rate results in UT2.
c) International Atomic Time (TAI)
The atomic time measurement standard officially replaced the astronomical definition of the second in 1967. The new second is defined as: the transition between the two hyperfine energy levels of the ground state of the cesium Cs133 atom at sea level in a zero magnetic field lasts for 9192631770 cycles The time of is an atomic hour, we call it International Atomic Time (TAI), and its stability can reach more than 10-14. In addition, it is stipulated that the starting point of atomic time is at 0:00 (UT) on January 1, 1958, that is, at this moment, atomic time coincides with universal time.
d) Coordinated Universal Time (UTC)
Compared with the world time based on the rotation of the earth, atomic time is a uniform measurement system, which is very important for measuring time intervals. But the universal time reflects the position of the earth in space, and corresponds to the cycle of spring, summer, autumn and winter, day and night,
It is a time we are familiar with and indispensable in daily life. In order to balance these two needs, the Coordinated Universal Time (UTC) system was introduced. UTC is still an atomic time in nature, because its second length is specified to be equal to the second length of atomic time, but at the time, through manual intervention, it is as close as possible to universal time.
e) Leap second
UTC uses atomic time seconds for the second length, but in terms of time, manual intervention is required to make it as close to the universal time as possible. This requires the “leap second operation” of UTC, that is, whenever the difference between UTC and universal time UT1 exceeds close to or exceeds 0.9 seconds, the UTC time at the end of June or December of the current year will be increased by one second or decreased by one second .
A.3 Description of time synchronization network
The time synchronization network is composed of time synchronization device nodes and time synchronization links. Time synchronization can be roughly divided into the following three processes:
a) The acquisition of high-precision UTC time information is currently mainly achieved through satellite receiving systems.
b) Time transfer
Time information is transmitted from the high-level time synchronization device to the low-level time synchronization device and from the time synchronization device to the communication device that needs time synchronization. Depending on the time accuracy required by the equipment, different transmission methods can be used.
c) Time allocation is the method by which equipment obtains time synchronization through appropriate means.
Appendix B Introduction to Time Interface (informative appendix)
At present, the more common time interfaces in the world include several methods such as 1PPS+ToD, DCLS, IRIG-B, NTP, PTP, serial port ASCII string, etc. These methods will be briefly introduced below.
B.1 1PPS＋ToD
The second pulse signal does not contain time information, but its rising edge marks the accurate start of each second. It is usually used for local testing and can also be used for intra-office time distribution, with an accuracy of 100ns. ToD interface usually adopts RS232/RS422 serial communication port to encode time information. However, because there is no unified standard for the ToD interface, devices from different manufacturers cannot be interoperable, so the application range of this method is relatively small.
B.2 DCLS
DCLS is another transmission pattern of IRIG-B code. It uses DC potential to carry symbol information, which is equivalent to the envelope of IRIG-B modulation code. IRIG-DCLS technology is transmitted through leased dedicated lines. The comparison between IRIG-B common mode and IRIG-DCLS mode is shown in Figure 2.
Figure 2 Comparison of IRIG B and DCLS
B.3 IRIG-B
IRIG encoding originated from recording time information for tapes, with obvious analog technology colors. It has been widely used as a time transfer standard since the 1950s.
IRIG-B uses 1KHz sine wave as the carrier frequency for amplitude modulation, and encodes the nearest 1 second. The contents included in the IRIG-B frame include: day, hour, minute, second and control information, etc. The maximum channel bandwidth occupied by it is 3KHz. Therefore, it can be transmitted in the building with an ordinary twisted pair cable, or can be transmitted over a long distance on the analog telephone network. In the 1990s, in order to meet the needs of the year in the turn of the century, IEEE 1344-1995 stipulated a new format of IRIG-B time code, requiring the code to also include the year, and other aspects remained unchanged.
B.4 NTP
There are three main protocols for time synchronization in computer networks: Time Protocol, Daytime Protocol, and Network Time Protocol (NTP). There is also a simple network time protocol SNTP (Simple Network Time Protocol). SNTP is a simplified version of the NTP protocol with simpler functions.
Among the above-mentioned network time protocols, the NTP protocol (RFC 1305) is the most complex and can achieve the highest time accuracy. It is also the most widely used time protocol currently.
B.5 PTP（IEEE 1588）
The NTP protocol (that is, the Network Time Protocol) is implemented by software, while the PTP protocol uses both software and hardware. The two cooperate with each other to achieve more accurate time synchronization.
Similar to the NTP protocol, the slave clock sends a message with a timestamp to the master clock, the master clock receives and responds, and records the received local time and the local time of the sending time in the response message; the slave clock according to four times Poke, calculate the total delay between the local and the opposite end; assuming that the round-trip delay is equal, the difference between the local time and the remote time can be calculated from the clock to adjust the local time until it is synchronized with the opposite end.
B.6 Serial port ASCII string
Through the RS232/RS422 serial communication port, the time information is encoded in an ASCII code string. The baud rate is generally 9600bps, the accuracy is not high, and the 1PPS signal is usually used at the same time.
Currently, the most widely used time synchronization technologies are mainly DCLS and NTP. Among them, the time accuracy obtained by using DCLS technology is high, but this method cannot monitor the change of transmission delay in real time, and is only suitable for occasions where the transmission link is relatively unchanged; the time accuracy obtained by using NTP is low, but this method can be monitored in real time The change in transmission delay can also make full use of existing network resources, and the networking is simple and easy to maintain.
Appendix C Application considerations of time synchronization equipment (informative appendix)
In the application of time synchronization equipment, appropriate time synchronization network technology should be adopted according to the different needs of time services and the construction cost of time synchronization network. At the current stage, it is advisable to adopt a variety of means to build a time synchronization network, such as using DCLS, NTP and other technologies to provide ordinary precision time services, and PTP technology to provide high-precision time services. For a long period of time, it will be a situation where multiple methods coexist. With the development of technology, a unified time and frequency synchronization network formed by a single technology (such as PTP technology) may appear in the future to meet various time synchronization and frequency synchronization requirements.
C.1 Provide ordinary precision time service
Competition in the communications operation market is becoming increasingly fierce, and service quality is one of the most important factors to win the competition. In order to resolve billing disputes and improve network operation efficiency, major telecom operators have successively started the construction of time synchronization networks and the transformation of time synchronization access of business equipment and other methods and methods. Through several years of hard work, the urgent common precision time synchronization problems of some billing systems and network management systems have been initially solved. According to the current time synchronization equipment technology, it is relatively easy to provide ordinary precision time synchronization services, but in the specific promotion and application, some problems are found:
(1) Since the concept of time synchronization was put forward relatively late, only a few new equipment from a few equipment manufacturers have a time input interface, and most communication equipment does not have a time input interface, so the equipment hardware is usually modified to increase the time input interface. , Or adopt the method of modifying the equipment software to increase the network management system to regulate the time function of the network element. Some operators have implemented hardware transformation schemes for some equipment in some areas. However, in the face of many network element equipment from different manufacturers on the Internet, the workload, cost, and risks of this hardware transformation scheme are very huge and cannot be large. Area, large-scale promotion and use. Some operators implement the necessary software upgrades to the network management system to automatically adjust the time of the network element equipment, and develop a special time synchronization program in the operating system of the network management system, and connect to a time synchronization device through the DCN network , So as to obtain time synchronization. The workload, cost, and risk of this software transformation scheme are relatively small, but it must rely on the network management system and the network topology of the DCN network, as well as the deployment and ownership management of time synchronization equipment, and there are some non-technical problems.
(2) For IP-based networks and a large number of computer-based equipment and application systems, there are indeed a large number of ordinary precision time synchronization requirements. Usually, a dedicated NTP server has been considered during the construction process, but the NTP server is usually not included The time synchronization network established by the operator.
Therefore, there is still a lot of demand for time synchronization equipment in the application and promotion of ordinary precision time services. The time synchronization network needs to fully consider the time synchronization requirements of various business networks from its own networking, coverage and management. Develop various business network access time synchronization plans, and incorporate the time synchronization of the business network into the time synchronization network.
C.2 Provide high-precision time services
At present, in the communication field, the demand for high-precision time synchronization mainly comes from CDMA base stations and TD-SCDMA base stations. The existing solution is to configure GPS, that is, configure GPS on each base station device. The biggest problem with using GPS is safety. The U.S. government has never given any promises or guarantees on the quality and duration of GPS signals, and the U.S. government has the ability to severely degrade GPS signals in specific areas. This is a massive use. The most severe security hazard faced after GPS. On the other hand, due to the influence of the GPS receiver’s own anti-jamming technology, installed geographical environment, electromagnetic interference and human operation, there is also the possibility of degradation of GPS receivers, and there are currently not many methods available for GPS Degraded monitoring is also a potential problem affecting network quality. In addition, a large number of GPS receivers are also used in CDMA/TD-SCDMA base stations, and there are some problems such as relatively strict installation conditions and high failure rates in engineering and operation and maintenance. Therefore, backup technical means other than GPS must be considered. In view of the security issues and application issues of GPS, two technologies are currently considered
Technical means to ensure the safety and reliability of CDMA network and TD-SCDMA network: adopt Beidou radio frequency simulation or Beidou external receiver, or through ground high-precision transmission technology.
Although time synchronization devices (including independent and functional modules attached to other communication devices) can provide high-precision time output interfaces, they still need to face the technical difficulties of high-precision time transmission on ground links. Including high-precision time and frequency signal delay, damage, monitoring and other complex technical issues. So far, no country has been able to rely on ground links to transmit high-precision time synchronization, and no mature commercial related products have been seen on the market. Therefore, research is still needed to provide high-precision time services for time synchronization equipment