WO1998015077A1

WO1998015077A1 - Hierarchical synchronization method

Info

Publication number: WO1998015077A1
Application number: PCT/FI1997/000583
Authority: WO
Inventors: Jukka Kainulainen
Original assignee: Nokia Telecommunications Oy
Priority date: 1996-09-30
Filing date: 1997-09-26
Publication date: 1998-04-09
Also published as: EP0965193A1; FI963909A; CN1232586A; CN1138362C; FI104665B; AU4461297A; FI963909A0

Abstract

The invention relates to hierarchical synchronization method for use in a communications system utilizing a message-based synchronization method. The system comprises a plurality of nodes interconnected by transmission links. According to the method, the nodes interchange signals incorporating synchronization messages which include a synchronization signature capable of indicating the priority level of the transmitted signal in the internal synchronization hierarchy of the system. In order to improve the interference immunity of the synchronization of the system, a portion of the available transmission links are defined as reserve links in terms of synchronization, and the synchronization signature is complemented with information indicating whether the signal has passed via such a reserve link.

Description

Hierarchical synchronization method

Field of the invention

The present invention relates generally to the synchronization of communications networks and particularly to improvement of the interference immunity of network synchronization in a communications network employing message-based synchronization.

Background of the invention In the context of the present invention, the term node is used to denote a junction point of transmission sections in a communications network. Accordingly, a node may be formed by any apparatus or equipment such as a branching unit or a cross-connection device.

In a system using message-based synchronization, the nodes of the system are interconnected by transmission links serving for the information transfer needs of the system. The same links also carry the clock frequency of the transmitting node to the receiving node. Each node selects as the source of its own clock frequency either the frequency from a neighbouring node, the frequency of its own clock source or a frequency brought into the node from an external source through a separate clock input. To force all nodes of a system to operate at the same clock frequency, in most cases the system is preferentially synchronized to a single clock source called the master clock. For this purpose, all the nodes of the system having a direct connection with the selected master clock will be synchronized with said master clock, while more remote nodes connected with said nodes of direct connection, however, not having a direct connection with the master clock, will be synchronized with said nodes situated closer to the master clock. Analogously, the nodes more remote from the master clock will be synchronized to those nodes which are located one link closer to the master clock.

In order to establish a synchronization hierarchy of the above- described scheme within a communications system, the nodes of the system exchange synchronization messages with each other. These messages carry information which permit the individual nodes to select the optimal clock source for their clock synchronization. The nodes of the system are prioritized, and the system tends to synchronize itself with the clock frequency of the node possessing the highest priority level. Normally, a given priority level can be assigned to one node of the system only. The synchronization messages normally contain information on the clock source used by the transmitting node, the priority level of the transmitting node, and a parameter value indicating the quality of the clock signal. Thus, any single node can select the synchronizing source for its own clock signal to be the clock frequency of that neighbouring node whose clock frequency originates from a desired node and is of the highest quality. At the startup of the system, when no incoming synchronization message has yet been processed, each node uses its internal clock as its clock frequency source. As soon as the first incoming synchronization messages have been processed, the node selects the clock frequency of the highest-priority neighbouring node as the source of its clock frequency. When the system has attained a steady state of synchronization after all required synchronization messages are transferred throughout the system, the system runs hierarchically synchronized with the clock frequency of the master source.

In Fig. 1 is shown in steady state a system MS utilizing message- based synchronization. The priorities assigned to the nodes are indicated by numerals marked inside the circles representing the nodes. The smaller the value of the numeral the higher the priority of the node. Synchronization m- essages sent by node n (n = 1-6) are marked with reference symbol MSG_n. Generally, the synchronization messages sent by the individual nodes are different from each other and have a format dependent on the message- based synchronization method used in the system. Distribution of the clock frequency from the master clock (node 1) to the other nodes of the system is indicated with solid lines. While the internode links indicated with dashed lines are not used for system synchronization under normal conditions, they are available during state change situations.

The basic concept of message-based synchronization is that the system operator defines the synchronizing hierarchy of the nodes by assigning each node a dedicated signature serving to indicate the level of the node in the hierarchy, and the system is allowed to synchronize itself in a self- contained manner with the defined master clock utilizing all existing internode links as required. If the chain of synchronizing links to the master clock is disrupted and a redundant chain of internode links is not available, or the master clock fails, the system will assume synchronization with a node of the next highest level. Such a reaction to a change in the system synchronization takes place via message interchange between the nodes. In the case that the timing information received by a node is disrupted, the synchronization hierarchy is reconstructed from the point of synchronization discontinuity onward (that is, hierarchically outward from the master clock node of the system). Generally, the end result is a hierarchy structure approximating the original situation, however, having the faulty link replaced by a functional link with the rest of the system configuration remaining almost unchanged.

Message-based synchronizing methods of the above-described kind are disclosed, e.g, in US Pat. Nos. 2,986,723 and 4,837,850, from which more details on the this type of synchronization scheme are to be found. Messages used in a prior-art message-based master-slave synchronization method (SOMS) will be elucidated in more detail later in conjunction with the discussion of Figs. 2 and 3. As is evident from the above description, synchronization in the message-based synchronization method is conventionally established over the shortest path from the node running under the master clock to the other nodes of the network. According to this concept, all the links between the nodes are considered equal with regard to synchronization. Thus, any link can be utilized for synchronization purposes provided that it momentarily fulfills the specified quality criteria. Even when a link is known to be prone to fail, it may still be used under normal conditions for synchronization provided that said link at the moment fulfills the specifications defined for synchronization, and moreover, happens to be situated in such a location of the network that the message-based synchronization method can utilize it. If a failure then occurs in such a fault-sensitive link used for synchronization, the message-based synchronization method has to initiate its synchronization recovery routine for the purpose of reconstructing its synchronization hierarchy in at least some areas of the network. Hereby, the quality of synchronization may deteriorate temporarily and uncontrolled bit slips can occur.

In the prior art, these undesirable situations have been circumvented by unconditionally inhibiting the use of the link for synchronization if its failure rate has been found excessively high. This approach, however, is handicapped by the shortcoming that the alternative synchronization paths potentially available in the network cannot be utilized under fault situations, ultimately causing loss of connection with the master clock. Summary of the invention

It is an object of the invention to overcome the above-described drawbacks of conventional techniques and to provide a message-based synchronization method in which links prone to fail during malfunction cannot cause changes in network synchronization and by virtue of which method all links remain available for the synchronizing method as efficiently as possible. The goal of the invention is achieved by the solution specified in the appended independent claims.

The invention is based on the concept of explicitly defining certain links as reserve paths of synchronization, said arrangement being implemented so that the node equipment mark the synchronization signatures received over such reserve links with a special reserve link identifier. (It must be noted herein that said links are defined as reserve links only with regard to synchronization, not with regard to normal data transmission.) Of two separate synchronization signatures, when one is received over a reserve path (that is, a path including a reserve link) and the other over a normal path, yet both signatures originating from the same master clock, the signature received over the normal path is always assigned a higher priority. Each node which synchronizes itself with a signal whose synchronization signature indicates that the signal has passed over a reserve path or link must preserve in its outgoing synchronization signature the information indicating the use of such a reserve path.

By virtue of the arrangement according to the invention, network synchronization can be controlled through inhibiting the synchronization use of certain paths under normal conditions even if the message-based synchronization method would prefer the use of such paths. However, these reserve paths may be utilized in a fault situation provided that no other paths can be established to the master clock source.

Furthermore, since the novel use of a reserve path does not affect the value of the synchronization signature, the arrangement according to the invention permits the mutual prioritization of synchronization sources to be distinctly separated from the use of a reserve path. Resultingly, the behaviour of the synchronization method is always precisely known and real-time information is available on a possible use of a reserve path. Brief description of drawings

In the following, the invention and its preferred embodiments will be examined in greater detail with the help of examples illustrated in the appended drawings in which

Figure 1 shows a communications system using a message-based synchronization when the system is synchronized with the clock frequency of the master source; Figure 2 shows a network using a self-organizing master-slave (SOMS) synchronization scheme in its initial state;

Figure 3 shows the network of Fig. 2 in steady state; Figures 4a-4d show a diagram elucidating the method according to the invention by a sequence of four different states of a network using message-based synchronization; Figure 5 shows a flow diagram of the comparison process according to the invention of synchronization signatures; Figure 6a shows an apparatus suited for implementing the method according to the invention in single node of the network; Figure 6b shows an alternative embodiment of node equipment; Figure 7a shows an example of the structure of the synchronization message; and Figure 7b shows an exemplifying embodiment of the transfer of the synchronization signature and the reserve path information in the synchronization message illustrated in Fig. 7a.

Detailed description of the invention

Referring to Fig. 2, therein is illustrated a system utilizing Self- Organizing Master-Slave (SOMS) synchronization (which is a well-known message-based synchronization technique), said system in the illustrated case comprising five nodes (or apparatuses) indicated by reference numerals 1-5 according to their assigned level in the synchronization hierarchy. (Herein, the master node of the network has the smallest-value SOMS address.) The nodes interchange messages containing said SOMS addresses. Thus, the nodes can identify each other by virtue of these address numbers and establish a synchronization hierarchy permitting the entire network to be synchronized with the master clock source node. As mentioned above, the synchronization messages interchanged continuously in the network are dependent on the type of the message- based synchronization method used. Moreover, each transmitting node sends a message with an individual content. In an SOMS-synchronized network, the synchronization message is comprised of three distinct parts: a frame structure, a signature and a checksum. The SOMS signature is the most important part of the SOMS message. The signature is formed by three consecutive numbers D1-D3:

D1 indicates the origin of the synchronizing clock frequency used by the transmitting node, that is, the SOMS address of the node acting as the master clock source node for the transmitting node.

D2 is a connection quality parameter, which typically is given a value proportional to the distance of the receiving node to the node indicated by D1. The distance is expressed as the number of nodes between the source node and the receiving node.

D3 is the SOMS address of the transmitting node. Each node (or device) performs continuous comparison between the incoming SOMS signatures and selects the one with the smallest value. While the fields D1 , D2 and D3 are directly combined in the signature into a single number formed by the direct combination (D1 D2D3) of the fields (in the text, however, a hyphen is used to separate the different parts of the signature, e.g., D1-D2-D3). Thus, the primary criterion in the selection of the smallest-value address will be based on the SOMS address (D1) of the node appearing to the preceding nodes as the master clock source, which means that any node aims to synchronize itself with a signal whose clock frequency source can be traced to a node of the smallest possible address. Then, the entire network in steady state will run synchronized with the same master node (because the master node of the entire network has the smallest-value SOMS address). If two or more of the incoming signals are synchronized with the same master node, the receiving node will select the master clock frequency from the transmitting node via which the path to the master node is shortest (i.e., the value of D2 is smallest). The last selection criterion is based on the SOMS address (D3) of the SOMS-message transmitting node, whereby the smallest value of this address is chosen if the preceding steps of address selection have not been able to rank the incoming signals. After a node has validated one of its neighbouring nodes as a new synchronizing source after comparison of the incoming signal SOMS signatures, the node has to reconstruct its own SOMS signature. The new value of the SOMS signature can be derived from the selected smallest-value SOMS signature as follows: the first field (D1) is left unchanged, the value of the second field (D2) is incremented by one and the third field (D3) is replaced with the SOMS address of the node itself.

Additionally, each node has an internal SOMS signature of the format X-O-X, where X is the SOMS address of the node itself. If none of the SOMS messages in the incoming signals has a value smaller than in the internal SOMS signature of the node, the node will select the internal oscillator of the node, or possibly a synchronizing signal present at the external clock signal input of the node, for synchronizing the clock of the node. Obviously, the internal SOMS signature of the node will then be used in the outgoing SOMS message.

The nodes send the SOMS messages continuously in each direction in order to assure maximally rapid emission of altered synchronization information in the SOMS signatures and to keep neighbouring nodes continuously informed of each others' operating status. Before the SOMS mes- sages can be compared with each other, the incoming SOMS messages must be accepted and the SOMS signatures must be separated therefrom.

When a given transmission link submits the first time an SOMS message, the SOMS signature of the message is accepted immediately for comparison if the message is faultless. When the incoming signal from the transmission link has an acceptable SOMS signature and the received messages carrying the same, unchanged signature are faultless, the situation remains unchanged. If the received SOMS message is found corrupted, the current SOMS signature is retained valid until three successive faulty SOMS messages have been received. Then, the SOMS signature is ex- eluded from the comparison process. This kind of waiting for three consecutive SOMS messages serves to eliminate temporary disturbances.

If a link fails to provide any SOMS message even when the link is operating otherwise acceptably, the comparison process activates a delay corresponding to three consecutive SOMS messages before the currently selected SOMS signature is invalidated. If the link fails entirely, the SOMS signature is discarded immediately. Equally, when interference imposed on the incoming signal makes it impossible to extract an SOMS signature of sufficient quality for comparisons, the SOMS signature of that link is discarded. Instead, the comparison process is carried out using the default SOMS signature for that incoming link having all the fields (D1 , D2 and D3) set to their maximum values (MAX-MAX-MAX).

When the incoming SOMS message is detected to have a new, changed SOMS signature, the signature is admitted directly into the comparison process provided that the message is faultless. Thus, any changes in the network configuration are treated without delay. Initially, each node uses its internal source of synchronization frequency, whereby the node sends the other nodes its internal SOMS signature of the format X-O-X. This signature is also compared with the other incoming SOMS signatures. If none of the incoming signatures has a smaller value than that of the node's internal signature, the node will continue using its internal synchronization source.

Still referring Fig. 2, the SOMS-synchronized network shown therein is in its initial state when no node (or device) has yet had time to process the incoming SOMS messages. All the nodes give the highest priority to the internal SOMS signature of the node, because no other signatures have been processed so far. In Fig. 2, the incoming SOMS signatures of each node are marked beside the node, and the selected signature is marked inside the outline frame (herein it must be noted that in the initial state shown in Fig. 2, all the nodes use their internal sources of synchronization). Links used for synchronization are drawn with a solid line, and the standby links are indicated by dashed lines (whereby it must be noted that in the initial state illustrated in Fig. 2, all the lines represent standby links).

After the nodes have had some time to process the incoming SOMS messages, node 1 stays synchronized with its internal clock source, nodes 2 and 4 will synchronize with node 1 on the basis of its signature 1-0-1 , node 3 will synchronize with node 2 (signature 2-0-2) and node 5 with node 3 (signature 3-0-3). By the same token, the nodes will rewrite their own new SOMS signatures in the above-described fashion and attach the signature to the outgoing SOMS message. After the network has assumed steady state, its configuration will be as shown in Fig. 3. Herein, all the nodes are synchronized with the master node 1 via the shortest possible path. As mentioned above, one or more of the transmission links of the network may have a higher failure rate due to, e.g., different character of said link or links with regard to other links of the network. The links which are known prone to fail are defined according to the invention as reserve links in synchronization use by means of complementing the synchronization message passing (or passed) via such a link with additional information indicating the passage of the respective synchronization signature via a reserve link.

In Figs. 4a-4d is illustrated the effect of the present method on the synchronization behaviour of the network through a drawing presenting a network of eleven nodes in four different states. In all the illustrated states, the master clock of the network is node 1 as indicated by oblique hatching of the circle representing the node. While the message-based system is in the examples assumed to be SOMS-synchronized, the invention may as well be applied to other message-based synchronization techniques in which the synchronization signature transferred in the network contains information on the master clock identity or equivalent data. Solid lines in similar manner as those of Figs. 2 and 3 are used in the diagrams to indicate connections along which synchronization takes place.

In Fig. 4a is shown the initial state in which the synchronization process advances along paths 1®2®4®7®9, (1®)2®5 as well as 1®3®6®8®10 and (1®3®6®)8®11. As an example, the transmission links of the network are assumed to be optical fiber links with the exception of two links (marked with reference symbols X and Z) that are assumed to be radio links which by their character are more sensitive to interference. Hence, the latter links are defined as reserve links for synchronization. On the basis of this categorization of links, synchronization next proceeds as shown in Fig. 4b, that is, using the paths 1®2®4®7®9, (1®)2®5®6®3, (1®2®5®6®)8®10, and (1®2®5®6®)8®11. In this state, link X is not used any more for synchronization. Should a link now fail between, e.g., nodes 6 and 8, the network will assume the synchronization state shown in Fig. 4c in which the reserve link Z is taken in synchronization use, because no other facility is available for transmitting the master clock frequency to nodes 8, 10 and 11. Resulting- ly, synchronization will now take place along paths 1®2®4®7®9®10®8®11 and (1®2)®5®6®3. Should the link between, e.g., nodes 1 and 2 further fail, the synchronization state will assume the configuration shown in Fig. 4d. Herein, also the reserve link X must be taken in synchronization use in order to distribute the master clock frequency throughout the network.

While this kind of definition of reserve links in the illustrated case could not contribute to the synchronization recovery capability of the net- work, it does improve the interference immunity in a significant manner (provided that the links X and Z exhibit a high sensitivity to interference or failure). Hence, if the concept of reserve connections would not be utilized, disturbance on the links X and Z would also cause changes in network synchronization, and thereby, affect the network operation in a wider scale during each change situation. By contrast, when the links X and Z are defined as reserve links, disturbance on these links will not have any effect on the synchronization of the network under normal conditions.

In Fig. 5 is shown a flow diagram illustrating the comparison of synchronization signatures occurring in each node of a network using a message-based SOMS synchronization method. In the process illustrated in the diagram is assumed that a "new" synchronization signature received from one interface is compared with a signature received from some other interface called the "old" synchronization signature in the figure.

In the first step of the process (step 51), the values of the signa- ture parameters D1 are compared. If the value of parameter D1 in the new signature is better (having a smaller value) than in the old signature, a direct decision is made that the new signature is of higher quality, irrespective of whether the new and/or old signature has been received over a reserve link or not. Correspondingly, if the parameter D1 of the old signature is better, a direct decision is made in favour of the old signature ignoring in a similar manner whether the new and/or old signature have been received over a reserve link. By contrast, if the values of parameters D1 are equal, the comparison is made on the reserve link information carried in the synchronizing signatures. Step 52 is first performed to test if the new signature has been received over a reserve link (one or more). If the new signature has passed over a reserve link, the next operation (step 53) tests whether the old signature stems from path containing a reserve link. If the old signature has not been passed via a reserve link but the new signature has been, the old signature is selected as being a better one. Correspondingly, if step 54 finds that the new signature is received over a reserve-link-free path but the old signature not, the new signature is selected as being a better one. Finally, if neither or both of the signatures stem from a path containing a reserve link, the selection process proceeds to step 55 in order to compare the values of parameters D2. After this step, the selection process does not avail of the reserve link information any more. The comparison performed in step 55 makes it possible to select the signature of a better value in parameter D2. In the case that the value of parameter D2 in both signatures is equal, the process proceeds to step 56 in order to compare the parameters D3 and to select the signature of higher quality based on the better value of parameter D3. If both signatures have the same value of parameter D3, the signatures are considered to be of equal quality.

As compared to the comparison process of synchronization signatures used in the conventional SOMS technique, the arrangement according to the invention is different by having the comparison process complemented with a block denoted in the flow diagram with the reference symbol NB, in which the reserve link information carried by the signatures is tested.

As is evident from Fig. 5, the basic strategy is in every case to select from two synchronization signatures, of which both indicate the same master clock, the one in which the additional information indicates that the synchronizing signal has not been passed via a reserve link. By contrast, if two synchronization signatures indicates different synchronization sources, the one having the higher-level source in the synchronization hierarchy is always selected. In the latter case, the selection process ignores whether the signature has passed via a reserve path or not. In this fashion, the synchro- nization is under normal situations always based on the clock source having the highest quality.

The preceding discussion gives the preferred selection criteria for an SOMS-based synchronization method. Obviously, the selection criteria may be varied according to the synchronization method used and its func- tional behaviour.

Figs. 6a and 6b illustrate functional block diagrams of means suited for implementing the above-described method in a single node of a network. In its generalized configuration, the node may be comprised, e.g., of a plurality of parallel interface units IU1 , IU2 IUN, each communicating with at least one neighbouring node and of a control unit CU, common to all the interface units, whereby the control unit performs the decision-making related to node synchronization. The control unit and the individual interface units can communicate with each other via, e.g., an internal bus CBUS of the node equipment.

As an example, the figures show the system node communicating with the neighbouring nodes over two incoming connections A, and A₂, each of these terminated at its own interface unit. Typically, the connections are formed by 2 Mbit/s PCM digital signals compatible with ITU CCITT Recommendations G.703 and G.704 or SDH signals compatible with ITU CCITT Recommendations G.708 and G.709. In the signals of these connections it is possible to transfer the synchronization messages in various manners of which one exemplifying embodiment is explained later in the text. Each interface unit IU may have one or more interfaces through which the node is respectively connected to one or more neighbouring nodes. In general, a node can be characterized as comprising N interface units with M interfaces in total (M³N).

In Figs. 6a and 6b, the reference symbols specific to an interface or interface unit are indicated by a subindex, while parts common to all interface units are without a subindex.

The incoming signal of each connection is passed to a signal transmit/receive block 13_j (i=1 ,2,...), wherein the actual signal processing takes place. From block 13,, the synchronization message extracted therein is further passed to a synchronization message transmit/receive block 16, directly connected thereto. Among other operations, the synchronization message transmit/receive blocks 16, check the integrity of the synchroniza- tion message and pass the validated message further to a centralized decision-making block 20 of the node via the common bus CBUS. The signal transmit/receive blocks also monitor the quality of the receive signal and store the results in interface-specific fault databases 14,. Thus, each synchronization message transmit/receive block can retrieve fault data from its dedicated fault database. Conventional methods are applied in the signal transmit/receive blocks for monitoring faults or changes in a signal passed over a transmission link.

The decision-making block 20 of the control unit CU stores the synchronization signatures received from the interfaces in memory area 21 , performs comparison of the stored synchronization signatures, and on the basis of the comparison, selects the signature with the highest priority as the synchronization source for its own clock. On the basis of received signatures, the decision-making block forms in memory area 22 a priority list in which the available synchronization sources are sorted is descending order according to their priorities resolved from the contents of their synchronization signa- tures so that highest in the list is placed the clock frequency source which is currently selected for the synchronization of the node and whose signature acts as the template on which the outgoing signature of the node is formed. From the interface units, the decision-making block also obtains fault information related to the incoming signals, either as a synchronization message or as separate fault data.

The decision-making block also maintains the currently valid value of outgoing synchronization signature in memory area 24, wherefrom the signature is distributed to the interface-specific synchronization message transmit/receive blocks 16,. An incoming or outgoing link can be defined in the node as a synchronization reserve link in two different ways depending on the storage location of the reserve link information.

In Fig. 6a is shown the first alternative arrangement in which each interface unit stores the information defining which ones of its interfaces are connected to synchronization reserve links. The reserve link definitions entered by the operator are stored separately for each interface (memory areas 17_;), adapted in conjunction with the synchronization message transmit/receive blocks. When the synchronization message transmit block receives from the decision-making block a new outgoing synchronization signature, it complements the signature with the reserve path data. Similarly, when the synchronization message receive block receives a message from the network and recognizes a change in the synchronization signature of the message, it complements the signature with the reserve path information (in the case that this information is lacking) prior to forwarding the signature to the decision-making block of the node.

In Fig. 6b is shown another alternative arrangement in which the operator-entered definitions of synchronization reserve links are stored in a centralized manner in one location (memory area 23 of the decision-making block) for common use in the node. Now, when the decision-making block distributes the outgoing synchronization message to the interface units of the node, the block adds the reserve link information to the messages submitted to the interfaces defined as serving the synchronization reserve links. Similarly, when the decision-making block gets a received synchronization signature from an interface unit, the block can immediately identify a signature received over a reserve link, thus being able to add the reserve link information prior to the comparison of synchronization signatures if so required.

Typically, the signal transmitted from another node of the communications system to the signal-processing blocks of the interface units IU is, e.g., a 2048 kbit/s signal conforming with ITU CCITT Recommendations G.703/G.704, the frame of the signal comprising 32 time slots (TS0-TS31) and each multiframe being formed by 16 frames. In the frame structure of this signal, the synchronization message can be transferred so that, e.g., the synchronization message reserves two bits in some time slot of the frame structure, advantageously from the bits of time slot TSO (it must be noted herein that while the frame alignment signal occupies the bits of time slot TSO in every other frame, the remaining every other frames have bits 4-8 reserved for national use, whereby they can be utilized for transferring the synchronization message). If the synchronization message transfer is implemented using the bits of time slot TSO, maximally three bits will still remain for other use such as the service channel. Obviously, the bits needed for the synchronization message may also be reserved from some other time slot, but this arrangement bears the penalty of stealing the required transmission capacity from the capacity reserved for the payload.

After the two bits discussed above are reserved for the synchron- ization message from a suitable time slot of the frame structure, the message is sent in the selected channel in a "piecewise" manner (2 bits in each frame).

The generalized structure of the synchronization message can be such as, e.g., shown in Fig. 7a comprising eight consecutive bytes. In serial transmission, the actual message begins from the first zero following the string of eight consecutive ones (the messages are sent successively without delay). After the first byte, the most significant bit (bit 8) of each byte is zero in order to prevent the message byte from ever comprising eight ones, which would cause confusion of the message byte with the start byte. In the first byte of the actual message, six bits (bits 2-7) are reserved for header information and the last bit (x) for user data. The next five bytes carry user data in bits 1-7 with bit 8 being zero. Bits 1-7 of the last byte contain the check sum of the message.

In this type of synchronization message, the SOMS signature fields D1-D2-D3 and the reserve path information can be transferred using a method such as that shown in Fig. 7b. Field D3 is transferred in, e.g., bytes 2-4, field D2 in bytes 4 and 5, and field D1 in bytes 6 and 7. The reserve path information Ww may be transferred in, e.g., bits 2 and 3 of the fourth byte, whereby the following bit combinations could be used, for instance:

00: synchronization signature not passed via a reserve path, 01 : synchronization signature passed via a level 1 reserve path,

10: synchronization signature passed via a level 2 reserve path,

11 : synchronization signature passed via a level 3 reserve path.

In the case that no need exists for categorizing different levels for the reserve paths, either of the fourth-byte bits can be used alone (e.g., 0: synchronization signature not passed via a reserve path, 1 : synchronization signature passed via a reserve path).

In the case that the prioritization of reserve paths is adopted, the signature comparison is extended to include the priority level of the reserve path, too, whereby from two signatures originating from the same master node but passed over two different reserve paths is selected the one having the highest priority level. For instance, in the above-described SOMS synchronization method, the priority information may be considered to comprise a parameter of type D1.5, that is, a parameter half-way between parameters D1 and D2, whereby this intermediate parameter is used in the comparison process after the comparison of parameters D1 but prior to the comparison of parameters D2.

Advantageously, the length of the transmit buffer in the node is made equal to the length of the message (8 bytes), thus permitting under interference-free operation the receiving node to find the start of the mes- sage always at the same point of the buffer, whereby there is no need to initiate a buffer content scan for finding the start point of each message separately.

In the case that the communications system employing the method according to the invention is formed by, e.g., an SDH network in which the signals present at the input ports of the node conform to ITU

CCITT Recommendations G.707, G.708 and G.709, the synchronization message can be transferred in that part of the STM-N signal frame (N=1 ,4,16,...) which is specifically reserved for the synchronization information (header part of the STM-1 frame).

A reserve path may be activated in a variety of different manners. One straightforward approach is to take a reserve path into use as shown in the diagram of Fig. 5. Alternatively, the use of a reserve path may be permitted only after a preset delay during which the node has not had a reserve- link-free path to the master clock source. This technique takes into account the possibility that due to network delays, a synchronization signature origi- nating from the master clock source may be received in failure or change situations only after a certain delay when transmitted over another path. Such a time-delayed decision-making technique is capable of preventing unnecessary signature hunting, thus inhibiting the node from at once switching to use a signature received via a reserve path and then immediately back to the higher-quality signature now passed over another path.

In the node, a transmission link may be defined as a reserve link in conjunction with either reception or transmission. Further, the reserve path definition may be made at both ends of the link, or alternatively, only at one end. If the reserve path definition is made only at one end of the link, the definition is made in conjunction with both transmission and reception. The latter alternative assumes that the link serves as a synchronization reserve link for both directions of transmission. This convention arises from the fact that a transmission link can be used as a synchronization reserve link in one direction only. More generally, a link is used for synchronization purposes only in one direction as is evident from the foregoing discussion of the potential applications of the method.

To a person versed in the art it is obvious that the invention is not limited by exemplifying embodiments described above with reference to the appended drawings, but rather, may be varied within the scope and inventive spirit of the appended claims and the above-given examples. For instance, the selection criteria applied to a synchronization signature passed over a reserve path may be varied in the discussed manner according to the synchronization method of the communications system and its specific properties.

Claims

1. A hierarchical synchronization method for a message-based communications system comprising a plurality of nodes interconnected by transmission links, in which method the nodes interchange signals incorpo- rating synchronization messages which include a synchronization signature capable of indicating the priority of the transmitted signal in the internal synchronization hierarchy of the communications system, characterized in that a portion of the available transmission links are defined as reserve paths in terms of synchronization, and the synchronization signature is complemented with additional information (Ww) indicating whether the synchronization signature has passed via such a reserve path.

2. A method as defined in claim 1, in which method the transmitted synchronization signature contains information on the originating source of synchronization, characterized in that, from two synchronization signatures indicating the same originating synchronization source, is always preferably selected the one in which the information added to its synchronization signature indicates that the synchronizing signal has not passed over a reserve link.

3. A method as defined in claim 1, characterized in that, from two synchronization signatures indicating different originating synchronization sources, is always preferably selected the one having its originating synchronization source categorized higher in the synchronization hierarchy.

4. A method as defined in claim 1, characterized in that said reserve connections are categorized in different levels by using more than one bit in the added reserve path information.

5. A method as defined in claim 1, characterized in that the addition of said information is performed at both end nodes of a link defined as a reserve link.

6. A method as defined in claim 1, characterized in that the addition of said information is performed only at one end node of the link defined as a reserve link.

7. A method as defined in claim 6, characterized in that the addition of said information is performed in said end node for both transmission directions of the link defined as a reserve link.

8. A method as defined in claim 2, characterized in that the node at the loss of synchronization connection to the master clock source of the network waits for the lapse of a preset delay time prior to selecting its synchronization source from a signal having its synchronization signature passed over a reserve connection.

9. Node equipment for a communications system using a message-based synchronization scheme, said system including a plurality of nodes interconnected by transmission links, said node equipment comprising a centralized control unit (CU) for decision-making of synchronization source selection in the node and interface units (IU1-IUN) whose interfaces permit the node to communicate with other nodes of the network, said nodes interchanging signals including synchronization messages which contain a synchronization signature capable of indicating the priority of the transmitted signal in the internal synchronization hierarchy of the system, said synchro- nization signature being formed in said control unit (CU), characterized in that said node equipment includes - memory elements (17-, 17₂; 23) by means of which a portion of the node interfaces are defined as reserve connections in terms of synchronization, and -means (16a, 16₂; 20) for adding reserve link information to synchronization signatures to be passed via said interfaces.

10. Node equipment as defined in claim 9, characterized in that said memory elements and means for adding the reserve link information are arranged in a distributed manner in conjunction with the inter- faces corresponding to the reserve links.

11. Node equipment as defined in claim 9, characterized in that said memory elements and means for adding the reserve link information are arranged in a centralized manner in conjunction with the control unit (CU) of the node equipment.