Research on clock synchronization performance monitoring system based on 1588v2 Technology

be based on 1588v2 Research on clock synchronization performance monitoring system based on

The telecommunication network emphasizes the operation, maintenance and management of the network (OAM), Ben OAM The performance monitoring system is used for IEEE1588v2 Specification defined multicast T-BC Form a time synchronization network .

    The performance monitoring system consists of three levels : Adjacent ne monitoring ( Point by point monitoring )、 Ring network Passive Point monitoring ( Segment monitoring )、 Multicast + Unicast monitoring ( Monitoring the whole process ) The three levels of monitoring are carried out independently , The performance data obtained can confirm each other , As the basic data, it is summarized to the network management system to form OAM Information , Including alarms 、 Fault location 、 Performance analysis and maintenance management .

    Monitoring at each of the three levels involves the need to monitor existing 1588v2 The specification can only be realized by making some small changes or rigid provisions . The three levels of monitoring are carried out independently , It can be developed according to the mature conditions .

One 、 Three levels of time synchronization performance monitoring methods

1.

Performance monitoring method i ：Master-Slave monitoring , Form point by point monitoring on the synchronization path . As shown in the figure ：

chart 1 Master-Slave monitoring

　 Usual Master-Slave Synchronization is based on 1588v2 Multicasting mode of network ,Master-Slave Interaction between PTP package (Sync,Delreq,Delresp) Synchronization .

   Master-Slave Monitoring is to achieve Master-Slave In the process of synchronization ,Master get Slave Time and Master-Slave Time delay between .

   Master-Slave The synchronization mechanism of synchronization is 1588v2 The way of “ Event information ”. standard 1588v2 Specifies the 4 Event information :Sync,Delreq,Pdereq,Pdelresp, this 4 Time stamp positions are reserved in all event packet formats .1588v2 Only... Is specified Sync The delivery of packets carries a timestamp t1, It is not required to carry the corresponding timestamp for the other three event information ! However, the three time messages are not prohibited from carrying time stamps .

    Will be 1588v2 Applied to the discussion of telecommunication network , We all recognize that carrier level networks need OAM.

    In order to make 1588v2 have OAM And the potential for performance monitoring , We suggest that Telecom profile Mandatory in the application of :1588v2 All in 4 Event information must carry an accurate timestamp .

    If there is such a stipulation , stay Master-Slave Monitoring Master Can be obtained from delay_req get Slave Time for ;Master towards Slave issue Pdelreq And Slave Interaction P2P after , Realized Master Yes Slave Performance monitoring of .

   Slave Forward synchronization path and Master The reverse monitoring path is the same , The two algorithms for obtaining time are the same , namely (T4-T1+T2-T3)/2. If there is asymmetry on the path , Then the asymmetric value is the same , So theoretically Master-Slave The monitoring threshold can approach 0. In fact, the alarm threshold can be set smaller ( for example 50ns).

   Master-Slave Monitoring is real-time , Form real-time monitoring performance curve , Based on this, threshold warning and threshold alarm are generated .

    When Slave according to BMC The algorithm switches paths to track others Master after , It will no longer return to the original Master issue Delreq, primary Master Accordingly, the Slave Monitoring of . For the Slave The monitoring of is changed to the one it tracks Master To execute .

2. Performance monitoring method II ：Passive Monitoring mode .

This method is mainly aimed at having P Ring network and mesh network of nodes . The ring network is shown in the figure ：    

    according to BMC A broken loop point will appear on the algorithm loop network P( With loop break point P The node of is called in this article P node ). Pictured above ,P For nodes S The port is synchronized with the West contact , Simultaneous use P Port monitoring east adjacent contact . actually P The node continuously compares the two arcs on the ring " Cumulative time error ", What is given is " Cumulative error of two arcs " The difference between the . Set a difference threshold ( for example 100ns), The difference within the threshold indicates that the whole ring network is normal ; If the difference exceeds the threshold , There may be a problem that deserves attention . Be careful , The number of nodes experienced by the East arc is almost the same as that of the West arc ( The difference between the two is 0 or 1)," In the east 、 The length of the west route is approximately the same " So that the threshold of the monitoring mode can be set smaller .

   Passive The monitoring method is ,P Nodes can be represented by Sync The packet obtains the timestamp of the east adjacent node ,P Mouth to mouth Pdelreq With the other party P2P Interaction , The transfer delay between them is calculated , From this we can get P Arc on the east side of the node " Cumulative time error ". In addition, by means of synchronization ,P Node to obtain the West arc " Cumulative time error ", So we can calculate " Cumulative error of two arcs " The difference between the , Realization Passive Monitoring mode . Because fault location mainly depends on "Master-Slave monitoring " The way ,Passive The monitoring method should be combined with "Master-Slave monitoring " To analyze and judge .

   Passive The monitoring method is real-time monitoring , Form the monitoring performance curve of the whole ring network . The basic feature is :① One P Nodes can monitor the entire ring , The monitoring range is large .② utilize 1588v2 Within the regulations PTP package , The monitoring method is simple and feasible .③ When there are temporary works on the ring network ( for example , Add node 、 Optical cable cutting, etc ) When the ring network changes ,Passive The monitoring method can be determined according to the pre project 、 After the performance monitoring data obtained, analyze whether the asymmetric value of the ring network becomes larger ? Whether it is necessary to carry out asymmetric compensation operation .

3.

Performance monitoring method III ： Whole process performance monitoring .

The whole process performance monitoring mode is from the time server to the BTS entrance , As shown in the figure ：

chart 2 Passive Monitoring mode

chart 3 Whole process performance monitoring

    The monitoring is performed by the end transport network element directly connected to the base station , This ne can be called “ The end node ”. stay “ The end node ” Get time synchronization through multicast , For performance monitoring , This end node should be directed to GM Initiate unicast request , Establish another unicast path with the time server , Compare the time obtained by unicast with that obtained by multicast .

    Unicast sessions are initiated and terminated by the end node according to the set schedule , Long term maintenance is not required , Multicast links always exist .

    For multicast , All network elements of the transport network that multicast passes through are 1588v2 Of BC Network element ; For the need of this monitoring method , The ne that unicast path passes through is 1588v2 Of TC Network element . namely , The network is required to have the following functions ：

（1）" The end node " It supports both multicast and unicast , Except multicast , Logically, it also needs to establish another unicast path with the time server ;

（2） Midway nodes support both multicast and unicast , Pair multicast PTP, Node implementations BC function ; Yes unicast PTP, The node is TC;

（3） The time server supports both multicast and unicast . Each time server needs to establish unicast sessions with multiple end nodes , Therefore, the time server is required to have a certain processing capacity so that it can accept multiple unicast requests , For example, thousands of unicast sessions can be accepted and maintained .

    All in all , The whole network is required to support both multicast and unicast .

    because BC The network element clock is deeply involved in multicast PTP, and TC The clock of the ne only shallowly intervenes in unicast PTP, Changes in the clock quality of the network element have a great impact on multicast and a small impact on unicast . It can be considered that , Even multicast PTP And unicast PTP Walking in the same path , End node about multicast - Unicast data comparison is also feasible , There is no need to select different physical paths for unicast .

    When the end node multicast - Unicast difference is within the threshold （ for example 1000ns）, It can be considered that the end-to-end whole process time meets the index , Otherwise, it needs special analysis and attention .

Two 、 Application suggestions of three time performance monitoring methods

    For monitoring purposes , Performance monitoring method i （Master-Slave monitoring ） And method 2 （Passive Node monitoring ） All require interaction between network elements P2P; The third method is to “ Add unicast ” To achieve monitoring purposes . Here's the thing , there P2P For monitoring only , It does not mean that the network synchronization adopts P2P-TC The way .

    The three monitoring methods are a progressive superposition relationship ,Master-Slave Monitoring is point by point monitoring ,Passive Node monitoring is equivalent to segment monitoring , Multicast and unicast cooperation monitoring belongs to the whole process performance monitoring , Thus, a performance monitoring system for the whole network is formed .

3、 ... and 、 Yes IEEE1588v2 Suggestions for modification of the standard

As mentioned above , To realize the time synchronization performance monitoring system proposed in this paper , Need to update existing 1588v2 The standard has some explicit provisions as shown in the following table ：

surface 1 1588v2 Suggestions for modification

Four 、“ probe ” To achieve performance monitoring

    At present, some manufacturers can realize the return network through small box equipment similar to the probe PTP Monitoring of transmission performance , Place the device at the network end node that needs to be monitored , Real time monitoring of... From upstream PTP Time deviation of signal , Through arbitrary IP The channel will monitor the results 、 The alarm information is uploaded to the management system ( It can be done by PC Of web How to achieve ), So as to realize the time synchronization performance monitoring of the return network .CUCC This implementation has been studied , Some suggestions are put forward to solve the problems existing in relevant manufacturers to make them more suitable CUCC Monitoring requirements for network time synchronization performance , It will be deployed in the near future 1588v2 Further actual verification and test work in the current network of the technology . This is right 1588v2 The subsequent maintenance and management after the technology scale deployment is convenient 、 Rapid and convenient fault location is of positive significance .