WO2007120089A1

WO2007120089A1 - Network synchronisation supervision

Info

Publication number: WO2007120089A1
Application number: PCT/SE2006/050075
Authority: WO
Inventors: Jan Pettersson
Original assignee: Telefonaktiebolaget Lm Ericsson (Publ)
Priority date: 2006-04-13
Filing date: 2006-04-13
Publication date: 2007-10-25
Also published as: TW200740162A

Abstract

The present invention relates to network synchronisation supervision. A clock supervision unit (500) is disclosed, as well as a method of monitoring the synchronisation of a node having a clock (220) and being connected to a network, wherein the clock (220) of the node is adjusted in accordance with a synchronisation reference of a signal (200iii) from the network. A synchronisation deviation of the clock from the synchronisation reference is measured at different points in time, and the resulting synchronisation deviation measurement values are analysed in a manner so that frequency components of the synchronisation deviation may be found.

Description

NETWORK SYNCHRONISATION SUPERVISION

Field of the invention

The present invention relates the supervision of radio communication networks. In particular, the invention relates to the supervision of the synchronisation in such networks.

Background

In mobile radio communications, speech and data can be transmitted between a mobile station and a mobile radio communications network by means of radio waves. On the network side, the radio waves are generated by a radio base station. For a particular transmission session, the generated radio waves arc of a particular radio frequency (which frequency may vary during the session). In order to generate radio waves of an intended frequency, the radio base station is locked to a synchronisation reference having a reference frequency. A synchronisation reference signal may be transmitted on a dedicated synchronisation signal line. However, in many mobile radio communication networks, the synchronisation reference is provided by a transmission network to which the radio base station is connected. The synchronisation reference could for example be extracted by the radio base station from the transmission rate of a transmission line from the transport network, or from the frequency of transmission of a particular frame on the transmission line. etc.

A transport network is in turn generally connected to a primary clock reference, which could e.g. be a national or international Caesium atomic resonance clock, which is used in order to keep the synchronisation reference of the transport network at a constant nominal value. However, a transport network is built of many different components, most of which do not operate in an ideal fashion. The non-ideal performance of some of these components will affect the synchronisation reference provided by the transport network. Hence, the synchronisation reference may deviate from the nominal value.

Deviations from the nominal value of the synchronisation reference lead to deviations from the intended frequency generated by the radio base station. Depending on the magnitude of the deviation, this can lead to serious problems in the mobile radio communications network: an offset in the radio frequency from the intended frequency will cause interference with other radio sessions in the mobile radio communications network or in other networks, the transmission range may decrease, the functionality of the radio base station may deteriorate etc.

Summary A problem to which the present invention relates is how to improve the performance of a mobile radio communication system.

This problem is addressed by a method of monitoring a synchronisation of a node, where the node has a clock and is connected to a network. The clock of the node is adjusted in accordance with a synchronisation reference of a signal from the network. The method comprises measuring a synchronisation deviation of the clock from the synchronisation reference at different points in time, the measuring resulting in synchronisation deviation measurement values. The method further comprises performing an analysis of the synchronisation deviation measurement values etc. in a manner so that frequency components of the synchronisation deviation may be found, and generating analysis results in accordance with the analysis.

The problem is further addressed by a clock supervision unit adapted to monitor a synchronisation of a node, where the node is connected to a network, and the node is adjusted in accordance with a synchronisation reference of a signal from the network. The clock supervision unit comprises a data measuring unit having an input for receiving a clock signal from the clock, and an input for receiving the signal from the network. The data measuring unit is adapted to measure the synchronisation deviation in the clock signal from the synchronisation reference, and to generate synchronisation deviation measurement values. The clock supervision unit further comprises a data analysis unit adapted to perform an analysis of the synchronisation deviation measurement values in a manner so that frequency components of the synchronisation deviation may be found, and to generate an analysis result in accordance with the analysis.

The invention can advantageously be implemented by means of a computer program product for monitoring the synchronisation of a node according to the inventive method. Such computer program product could preferably be stored on memory means arranged to fit in a communication node, such as a radio base station.

By the invention is achieved that abnormal variations in the synchronisation of the node may be detected. Such abnormal variations could for example be due to errors in the synchronisation reference, or to problems in the node. Since variations caused by errors in the node normally occurs with a different frequency than errors due to problems in the node, an indication of the origin of any problems in synchronisation can be obtained.

Hence, by means of the invention, the accuracy of any clocks in a communication system can be improved. This is particularly important in radio based communication systems using radio waves for communication, since the frequency of any generated radio waves are determined by means of a frequency of a clock. Hence, by improving the accuracy of the clocks of a radio communication system, interference between radio links transmitting at neighbouring frequencies can be reduced. Furthermore, reductions in the transmission range due to an unwanted offset in the transmitted frequency can be avoided, as well as any deteriorated performance of the functionality of radio base stations caused by an unwanted offset in frequency.

In one embodiment of the invention, the analysis performed includes calculating the Allan variance of the synchronisation deviation at different time scales. In another embodiment, the analysis performed includes performing a Fourier transform of synchronisation deviation measurements performed at different times.

The result from the performed analysis could advantageously be compared to the analysis result to a predetermined value, and/or to a predetermined pattern, and/or to an analysis result obtained at a different points in time. By such comparison is achieved that any deviation of the analysis result from the expected values may be detected, and information regarding the origin of such deviation may be obtained from the frequency dependence of the analysis result. Furthermore, an alarm could advantageously be issued if the comparison indicates that the synchronisation of the clock is erroneous. The invention could be used for selecting which synchronisation reference to use for the synchronisation of a clock of a node, in the situation where the node receives more than one incoming signal carrying a synchronisation reference. The synchronisation deviation of the clock could be measured and analysed in relation to more than one synchronisation references, and the synchronisation reference exhibiting the least fluctuations could be selected for the synchronisation of the clock.

The invention could advantageously be applied to more than one node in a communications system comprising several nodes. The analysis result obtained for different nodes could then be compared, and additional information regarding the underlying reason for any abnormal variations in the synchronisation of the nodes may be obtained. If the synchronisation reference is disturbed in one geographical area, the synchronisation of nodes located in this area may for instance show a more severe abnormal behaviour than the synchronisation of nodes in other parts of the system.

Brief description of the drawings

For a more complete understanding of the present invention, and the advantages thereof, reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which:

Fig. 1 is a schematic illustration of a mobile radio communications system comprising radio base stations and a transport network.

Fig. 2 is another schematic illustration of a mobile radio communications system.

Fig. 3a illustrates an example of how the Allan variance of a clock may vary with frequency when the clock functions normally.

Fig. 3b is an example of how the Allan variance of a clock may vary with frequency when abnormal behaviour of the clock frequency is caused by disturbances in the transport network. Fig. 3c is an example of how the Allan variance of may vary with frequency when abnormal behaviour of the clock frequency is caused by disturbances in radio base station.

Fig. 4 is a flowchart schematically illustrating an embodiment of the inventive method.

Fig. 5 is a schematic illustration of a clock supervision unit.

Detailed description

Fig. 1 schematically illustrates a mobile radio communications system 100, hereinafter referred to as system 100, in which a mobile station 105 can communicate with a radio base station 110 over a radio link 115. The radio base stations 110 and the mobile stations 105 form the radio part of system 100. The radio base stations 110 of system 100 are connected to a transport network 120, often via a radio network controller or a radio base station (not shown). The transport network 120 is typically connected to a reference clock 125 which is used as a synchronisation reference in order to keep the transmission rate constant. The reference clock 125 is typically a national or international Stratum 1 reference such as a Caesium atomic resonance clock or a Global Positioning System (GPS).

As discussed above, radio base stations 1 10 are commonly designed to use a traffic- carrying signal from the transport network 120 as a synchronisation reference for generating radio waves of an intended frequency/,,,,,,,^. The synchronisation reference could for example be the transmission rate or the frequency of transmission of a particular frame. The radio base station 110 is frequency locked to the incoming signal, and a clock reference is derived by the radio base station 110 from the physical interface used for connection (often via a radio network controller/radio base station) to the transport network 120. Alternatively, a dedicated synchronisation signal may be used. In the following, for purposes of illustration, the invention will be described in terms of a system 100 in which the synchronisation reference is carried by a traffic-carrying signal, although the invention is equally applicable to a system using a dedicated synchronisation signal.

Fig. 2 schematically illustrates how the clock signal 20Oi from the reference clock 125 is transmitted via the transport network 120 to the radio base station 115. A first node 203 of the transport network 120 uses the clock signal 20Oi for generating a transmission signal 200ii having a particular, nominal, synchronisation reference R₁₁₀₁,, of a reference frequency. Transport network 120 comprises a number of further nodes 205a- 205n, such as radio links, routers and switches, which are used for transmission of data over transport network 120. A node 205 typically regenerates the signal 200ii. and thus also regenerates the synchronisation reference, R. In such regeneration of the signal 200ii, deviations in the synchronisation reference. R, may occur, thus causing deviations from the nominal synchronisation reference. R_nom in the synchronisation reference, R. of the output signal 200ii being output from the regenerating node 205. Such deviations could for example be caused by a wrongly configured clock reference regeneration or broken equipment in the transport network 120. Since the output signal 200ii of a node 205. here referred to as node 205x. is assumed by the next node 205. here referred to as 205x+l. to carry the nominal synchronisation reference, i?,,_om, such errors in the synchronisation reference regeneration are transferred to the radio base station 110. Furthermore, errors in the synchronisation reference. R, may be due the fact that part of the transport network 205 uses a different reference clock 125.

As seen above, it is likely that the synchronisation reference. R, of the input signal 200iii to the radio base station 110 from the transport network 120 deviates from the nominal synchronisation reference. R_nom. Since the radio base station 110 is locked to the synchronisation reference of the incoming signal 200iii. the radio base station 110 will not detect this deviation.

A radio base station 110 is normally equipped with a filter 210 for filtering the incoming signal 200iii to remove unavoidable phase variations introduced by the transport network 120. However, slow variations of the synchronisation reference, R. which occur at a time scale similar to the time constant of the filter or greater, cannot be removed by such filter.

The radio base station 110 of Fig. 2 further comprises a radio transmitter 215, and an internal clock 220. (The clock 220 is referred to as internal in order to distinguish the clock 220 from the reference clock 125. The internal clock 220 may however be implemented exterior to the radio base station 110). The frequency locking of the radio base station 110 to the incoming signal 200iii is of a time constant T, and the internal clock 220 is adjusted in accordance with the synchronisation reference, R, of the incoming signal 200iii at regular time intervals, corresponding to the time constant 7^"(the adjustment normally being limited in amplitude so as not to cause phase jumps of a disturbing magnitude). Thus, changes in the synchronisation reference. R, that occurs at a rate of the order of magnitude of the time constant Toϊ the frequency locking, or slower, will be followed by the radio transmitter 215.

The internal clock 220 is used by the radio transmitter when generating a radio signal 200iv of a particular frequency/ Since the internal clock 220 is locked to the incoming signal 200iii. the frequency difference between the true synchronisation reference R of the incoming signal 200iii and the nominal synchronisation reference, R,_wm, will cause a deviation in the frequency/of the generated radio signal 200iv from the intended frequency

Jmtended-

The performance of the radio part of system 100 is rather sensitive to deviations from the intended frequency/,,,^,,^, and hence to deviations from the nominal value, R_nom, of the synchronisation reference of the incoming signal. R. Deviations in intended frequency. f i_ntended- us small as -0.05 ppm in WCDMA systems, or -0.1 ppm in GSM systems, normally cause problems in the radio part of system 100. The performance of the transport network 120 is, however, generally not affected by such small deviations in the transmission rate. Typically, the transmission over the transport network 120 can function normally even if the deviation in transmission rate is over 10 ppm. Hence, deviations in the synchronisation reference from the nominal value. /?„„,„. which will cause great problems in the radio part of system 100, will generally not be detected in the transport network 120.

Further deviations from the intended frequency f_mlended may be caused by problems in the radio base station 110, such as for example broken equipment, or the ambient temperature differing from the working temperature of the radio base station 1 10.

Problems arising from deviations from the intended frequency /,,,,..,,_Ot0Of signal 200iv on radio link 1 15 would typically involve interference with radio links 1 15 transmitting at neighbouring frequencies, a decrease of the transmission range, deterioration of the functionality of the radio base station 1 10 etc. When a mobile station 105 is in contact with two neighbouring radio base stations 1 10 which arc not synchronised, problems in hand- over will occur. One of the radio base stations 110 will in this situation interpret the a deviation from intended frequency, generated by poor synchronisation, as a Doppler shift, and will hence interpret the situation as if the mobile station 105 was moving. As a consequence, the transmission range will be affected. Furthermore, if the deviation from the intended frequency R_nom is large, the frequency/may fall outside the frequency range allocated to the operator of radio base station 110. resulting in deviations from the standard and intrusion of another operator^'s frequency range.

In present mobile radio communication systems 100. deviations in the transmitted frequency/from the nominal value f, mended cannot be discovered until the symptoms caused by such deviations are detectable (unless a stable reference oscillator is brought to the site of the radio base station 110 and the frequency of the transmitted signal 200i v is actually measured). When problems in the frequency generation of a radio base station 1 10 have been detected by way of detecting the symptoms of such problems, a reference oscillator is normally brought to the site of the radio base station 110 and measurements are performed in order to detect the reason for the problem. To perform site visits is expensive. Since the problems experienced are often not caused by errors in the radio base station, but due to errors in the incoming signal 200iii generated by the transport network 120, the site visits could often be avoided altogether had there been an alternative way of detecting errors in the incoming signal 200iii. Furthermore, if errors in the incoming signal 200iii to the radio base station 1 10 could be detected before such errors were large enough to cause serious problems to the radio signal generation, the performance of the system 100 would be greatly improved.

The synchronisation reference, R, of the incoming signal 200iii is used the in order to determine the duration of one second, or a corresponding representation of time/frequency, of the internal clock 220.

According to the invention, the synchronisation deviation of the internal clock 220 from the clock represented by the incoming signal 200iii is determined and analysed. This deviation will hereinafter be referred to as the deviation y of the internal clock. The deviation y of the internal clock can e.g. be determined by measuring the deviation in the duration of a second in the internal clock 220 from the duration of a second according to the incoming signal 200iii at different points in time. The deviation^ of the internal clock 220 at time /,-_/ can then be expressed as:

mmi ng e ^" +"^{" L} Λ^XUD 'i i'nn⁺cc^Ioommiinngg ■ ^ ^ where D[_{ni ernat} is the duration of a second according to the internal clock 220 at time t,.

£⁾ ",'„'C„O„ _m nui_nn g „ is ^tne duration of a second accordin ^og to incoming ^vj sig ^VJnal 200iii at /,. AD i¹n⁺t e_/nα/ is the change in the duration of a second according to the internal clock 220 which occurs between the two points in time /, and /,_+/, and AD m' co min g is the chan ^*g-e in the duration of a second according to the incoming signal 200iii which occurs between the same points in time.

The deviation y of the internal clock 220 can alternatively be determined by measuring other representations of the deviation than the duralion of a second, such as the deviation in the expected transmission rate according to the internal clock 220 from the actual transmission rate of the incoming signal 200iii. or the phase of the internal clock 220 as adjusted to the incoming signal 200iii. It is often efficient to express the deviation y in terms of Unit Intervals (UI). i.e. how many clock cycles the deviation corresponds to, since the Unit Interval (UI) is a unit commonly used in the maintenance of transport networks. In a system 100. the deviation y can advantageously be determined by use of a Jitter/Wander mask as described in ITU-T G.823. G.824 or G.825.

The measurements of the deviation y can then be analysed in order to delect any deviations in the synchronisation reference R of signal 200ili from the nominal value R_nOm- and/or any internal problems of the radio base station 110. In a non-ideal system, the deviation y of the internal clock will vary with time, and the deviation y will be composed of a number of frequency components. The analysis of the deviation measurements y can advantageously involve analysing how the magnitude of the deviation varies with deviation frequency.

The measurements of the deviation y can preferably be performed at regular intervals. Δt, and be stored in a memory. Such memory could e.g. be in the form of an array having n values, or in the form of a file. The memory could advantageously operate according to the "first in. first out" principle, so that the n most recently measured deviation values are always stored in the array.

According to one embodiment of the invention, the Allan variance. σ_y. of the sampled measurement values is calculated. The Allan variance. σ_y. also known as the Allan deviation, is defined as follows for a set of M values of the deviation y:

where τ is the time between two consecutive deviation measurements of the set of M values, and y, is the i^th deviation value of the set.

The Allan variance. σ_y, is a non-classical statistic which is used to estimate the stability of an oscillator by estimating the frequency fluctuations of the device over a given time interval. Traditionally, the deviation^, of an oscillator is determined in relation to a reference oscillator of known, constant, frequency. However, as further discussed below, the Allan variance can advantageously be used in analysing deviations from a true frequency of an oscillator (e.g. an internal clock 220) also when the frequency of the reference oscillator (e.g. the incoming signal 200iii) varies in an unknown manner. The Allan variance is particularly useful in determining deviations caused by periodical errors in the oscillator or in the reference oscillator. By analysing the measured deviation values y in terms of the Allan variance, periodical errors may be detected even if the mean value of the oscillation frequency does not change.

The Allan variance, σ_y. can be calculated for different sets of measured deviation values having different values of τ. Based on an array of n deviation measurement values where the time difference between two consecutive measurement values is Δt. σ_y can be calculated for n-1 sets having n-\ different values of τ. where Mean be larger for the sets

having smaller number of τ. and M=I for all sets having values of τ > - -At .

When the deviation y of the internal clock is represented by the phase of the internal clock 220. the Allan variance can be expressed as follows for a set of N phase measurements:

where y, is a measurement of the phase of the internal clock 220 as adjusted to the phase of the incoming signal 200iii at the time of the measurement.

The Allan variance σ_v, when calculated for a number of different values of r. gives valuable information about the frequency components of the variations in the deviation of the internal clock 220. The different values of rcorrespond to different frequency components of the variations. As is further illustrated in Fig. 3 below, any abnormal behaviour can easily be detected when σ, is plotted as a function of τ. or 1/τ .

As can be seen from expression (1). the measured values of the deviation y of the internal clock 220 will be affected by variations of the internal clock 220, as well as by variations of the synchronisation reference R of the incoming signal 200iii. Variations in the internal clock 220 can for example be caused by thermal noise, ageing, changes in the ambient temperature of the radio base station 110. broken equipment of the radio transmitter 215, etc.

The expected Allan variance σ_v caused by the normal variations in the internal clock 220. including variations caused by thermal noise and ageing, can be determined. Hence, if the measured Allan variance. σ'"^eas , differs from the expected Allan variance. _σ ^c*^Pected there may be reason for concern.

In Fig. 3. the Allan variance of a sample radio base station 1 10 is plotted as a function of log(l/r) in three different situations: Fig 3a illustrates the expected Allan variance of the sample radio base station 1 10 when there are no problems with the synchronisation reference or the internal clock 220 of the radio base station 110: Fig. 3b shows an example of how the Allan variance, σ_y(τ), may vary with 1/rwhen there is a problem in the transport network 120 causing the synchronisation reference to deviate from the nominal value R_nom, and Fig. 3c is an example of how the Allan variance of the sample radio base station 110 may vary with 1/rwhen abnormal behaviour of the clock frequency is caused by disturbances in the radio base station 110 itself. The examples given in Fig. 3 are only illustrative, and the Allan variance may show other dependencies of the deviation frequency 1/r. However, the important point is that the time scale of deviations from the expected Allan variance caused by fluctuations in the synchronisation reference R of the incoming signal 200iii. and the time scale of deviations due to internal problems in the radio base station 110 itself, are often different. Furthermore, different problems in the radio base station 1 10 give rise to deviations from the expected Allan variance at different values of r. and the same applies to different reasons for fluctuations in the synchronisation reference R. Hence, by analysing the Allan variance as a function of ror 1/r. the cause for any synchronisation problems of the radio base station 1 10 can often be derived.

In detecting any deviations from the expected Allan variance. _σ-^^χPtc"^>' (^) , the measured

values of the Allan variance (τ"^m'^s (T) could for example be compared to a threshold

value σ "" ' . which threshold value could preferably be a function of r. Alternatively, the

shape of the curve of σ"'^ιas , when plotted as a function of ror 1/r. could be compared to an expected shape pattern having a predetermined dependency of r, so that deviations in the shape of the curve of <τ"'^eαi are identified in the comparison, rather than deviations in the magnitude of the Allan variance.

Similarly, comparisons could be made between sets of measured values of σ'"^eas which

are measured at different occasions. Hence. σ"^ieas as measured on one occasion could be

compared to σ'"^e"^s as measured the previous hour, the day before, a week ago. etc.. in order to detect deviations in the synchronisation of the internal clock 220 with the reference clock 125. An alarm could advantageously be issued if <j'"^e"^s exceeds (7_y"^{es ι} , if σ"^mii deviates

from the expected shape pattern, or if σ"'^eas as obtained at one occasion differs from

(J™^eas obtained at the different occasion.

Fig. 4 is a flowchart schematically illustrating an embodiment of the inventive method, in which an incoming signal 200iii. carrying a synchronisation reference R. is received by a radio base station 1 10. In step 405, the deviation v of internal clock 220 is measured at a point in time t and the measurement value is added to a memory in which several measurement values of the deviation y, measured at different times /. may be stored. In step 410. it is determined whether it is time to perform an analysis of the deviation values >^• stored in the memory. If not, step 405 is re-entered, and a new measurement value of the deviation y is sampled and added to the memory, at a new point in time /. If it is found in step 410 that it is time to perform an analysis, step 415 is entered. Step 405 is preferably also re-entered, so that the sampling of measurement values y continues. In step 415. the Allan variance σ"'^eM of the deviation measurement values y stored in the memory is calculated, preferably for a number of different values of τ. In step 420, the result of this calculation is compared to predetermined values of the Allan variance (T^ ^{res i} . and/or to an expected shape pattern of the Allan variance as a function of τ. and/or to values of the Allan variance <j"^was obtained from a different set of measurement values of the deviation y sampled at different points in time /. In step 425, it is checked whether the comparison made in step 425 has revealed that the Allan variance deviates from the expected Allan variance. If so. the step 430 is entered, in which an alarm is issued. Alternatively, steps 420-430 could be omitted, and the result of the analysis performed in step 415 could be presented to an operator of system 100. Step 410 could be omitted if a new analysis is desired each time a new measurement value of the deviation y is obtained.

As is further discussed below, the inventive method may advantageously be used for selection of the most stable synchronisation reference R in a situation when a radio base station 1 10 receives more than one incoming signal 200iii carrying a synchronisation reference R. In this scenario, steps 405-415 could be performed for each synchronisation reference R. A comparison between the different values of the Allan variance obtained in relation to the different synchronisation references R could then reveal which synchronisation reference R would be the best to use as the synchronisation reference for the internal clock 220.

Fig. 5 schematically illustrates a clock supervision unit 500 according to the invention. Clock supervision unit 500 is adapted to receive the incoming signal 200iii and an internal clock signal 505 from the internal clock 220. The clock supervision unit 500 comprises a measuring unit 510 adapted to measure the deviation y of the internal clock 220 as described above, and a memory 515 for storing measured values of the deviation y.

Furthermore, the clock supervision unit 500 comprises a data analysis unit 520 adapted to perform analyses of the collected deviation data in terms of Allan variance as described above. The clock supervision unit 500 is preferably implemented in the radio base station 1 10. but could, in part or in full, be implemented in other nodes, such as a Base Station Controller (BSC) or a Radio Network Controller (RNC).

The clock supervision unit 500 can advantageously be implemented by use of suitable computer hardware and software.

The clock supervision unit 500 could advantageously be implemented in an operation & maintenance system of system 100. which could preferably include a performance management system for presenting the analysis generated by analysis unit to an operator of system 100.

The clock supervision unit 500 preferably further comprises an alarm unit 525, adapted to issue an alarm if the analysis performed by the data analysis unit 520 indicates an error in the frequency of the internal clock 220. Such alarm could for instance be in the form of the transmission of an alarm signal 530 to an operation & maintenance centre of the system 100. The alarm unit 525 could for example be adapted to issue an alarm if the measured Allan variance. σ'"^eai (r) . differs from the expected Allan variance. σ"^pt£/t (r) . by more than a predetermined value at a particular value of r(the predetermined value preferably being a function of r); if the r-dependence of σ""'" deviates from the expected pattern: and/or if a set of σ'"^aas (r) deviates from a set measured at a previous occasion.

The alarm signal 530 could be a simple indication of the fact that a deviation has been detected, or it could carry information about the frequency of the detected deviation.

The memory 515 can for instance be in the form of an array comprising n values, which are continuously updated to comprise the n most recently measured values of the deviation y. The number of stored measurement values, n, could for example be selected so that a calculation of σ_y can be performed for a value of rof the order of days. In order to allow for flexibility of the analysis of the Allan variance, n could advantageously be selectable. The time elapsed between two deviation measurements, Δi, could advantageously be in the order of magnitude of the time constant T of the frequency locking of the internal clock 220 to the frequency of the incoming signal 200iii. In a typical third generation mobile radio system 100. T is in the order of minutes. However, more frequent measurements of the deviation y may also be of interest. In one embodiment of the invention, the number of values used for the calculation of σ_}( τ) could be selectable, in order to allow for adjustment of the analysis to the circumstances which are presently of interest. The time elapsed between two deviation measurements. At, could advantageously also be selectable. The measured deviation values y can be single measurements, or values averaged over a time period shorter than or equal to Δt.

The data analysis unit 520 could be adapted to continuously update the analysis of the measured data when a new deviation measurement is entered into memory 515. Alternatively, the data analysis unit 520 performs the analysis at regular intervals, or on demand. The data analysis unit 520 could be implemented in a manner so that an operator of the clock supervision unit 500 could select whether the analysis performed by the data analysis unit 520 is performed continuously, at regular intervals, at specific times or on demand.

When the frequencies of the internal clocks 220 of several radio base stations 1 10 in a system 100 are analysed in the manner described above, a comparison between the performance of the different internal clocks 220 can advantageously be performed. Such analysis can provide further information regarding whether any detected deviation from the expected behaviour of an internal clock. 220 is caused by local problems in the radio base station 1 10. or by problems in the transport network 120 - if several internal clocks 220 of radio base stations 1 10 located in a similar geographical neighbourhood display similar deviations in the performance, this is a strong indication of a problem in the transport 5 network 120. Information regarding the results from the analysis of the deviation of an internal clock could be transmitted in a signalling message to a central node in system 100, such a Radio Network Controller or a Base Station Controller.

The teachings of the invention could also be applied to mobile radio communication [0 systems 100 in which one or more radio base stations 1 10 are connected to an additional reference oscillator of known frequency. Such additional oscillator could e.g. be a GPS device or a Rubidium reference, which could be a portable or stationary oscillator. The additional oscillator could be used to provide accurate measurements of any deviations of the synchronisation reference of the transport network from the nominal synchronisation 15 reference. R_nom. These accurate measurements could be used for calibration of the threshold value/pattern used in the comparison of σ'"^eas (r) to Q-^e^^ecle (^)

In the above, the invention has been described in terms of calculating the Allan variance of the deviation y of the internal clock 220 from the synchronisation reference R of the

.0 incoming signal 200iii. However, in other embodiments of the invention, other analyses of the measured values of the deviation y of the internal clock may be used in order to detect problems in the synchronisation of the internal clock 220. Such analysis could for example be a Fourier transform (fast or discrete) of the measured deviation values. In performing a Fourier transform, the direction of the deviation may be obtained as well as the amplitude

25 of the frequency components, so that the analysis result shows whether the internal clock 220 lags or is faster than the clock represented by the synchronisation reference at a particular time. Any other analysis by which information of the frequency components of the synchronisation deviation may be found could alternatively be used - for instance, in a simple case of sinusoidal variations, an analysis of the deviation y as a function of lime

30 could be sufficient to identify frequency components of the deviation. Although the invention has been described in terms of the supervision of the synchronisation of a radio base station 1 10 in a mobile radio communications system 100. the invention may also be applied to the supervision of the synchronisation of other network nodes which are connected to a transport network 120 in which an internal clock

5 220 is adjusted in accordance with a synchronisation reference R of a signal 200iii received by the node. In a mobile radio communication system 100. examples of such network nodes are Base Station Controllers (BSC), Mobile services Switching Centres (MSC). Radio Network Controllers (RNC). Media Gateways (MGW) etc. However, the invention is equally applicable to nodes in landline communication networks.

10

When a network node having an internal clock 220 and being connected so as to receive more than one incoming signal 200iii carrying a synchronisation reference, there will be more than one synchronisation reference R which could be used as a synchronisation reference for synchronisation of the internal clock 220. The synchronisation references

[5 received by the node will be referred to as synchronisation references R_a. Ri,, etc. The present invention could then advantageously be used in determining which of the synchronisation references R_a. R_b, etc. should be used for synchronisation of the internal clock 220. In determining this, the synchronisation deviation y of the internal clock 220 could be measured in relation to more than one of the synchronisation references R₁₁. R_b,

.0 etc., resulting in more than one set of synchronisation deviation measurement values

iVβ ' ^α '— ' ^a J- Iy* ' >£ '•••' .VH _' ^ctc- ^^{n ana}ly^s'^s as described above could then be performed on the more than one set of measurement values, and by comparing the results from these analyses, the synchronisation reference R providing the best synchronisation of the internal clock 220 could be identified and selected as the synchronisation reference R to

25 which the internal clock 220 should be locked. Once the synchronisation reference R to be used has been selected, the synchronisation deviation y of the internal clock in relation to the selected synchronisation reference could be monitored as described above. The process of determining which of synchronisation references R₁₁. R_b. etc. should be used could be repeated at regular intervals, on demand, or when indications are given that the selected

30 synchronisation reference R does not provide adequate synchronisation of the internal clock 220. One skilled in the art will appreciate that the present invention is not limited to the embodiments disclosed in the accompanying drawings and the foregoing detailed description, which are presented for purposes of illustration only, but it can be implemented in a number of different ways, and it is defined by the following claims.

Claims

1. A method of monitoring the synchronisation of a node ( I K)), the node having a clock (220) and being connected to a network (120). wherein the clock (220) of the node is 5 adjusted in accordance with a synchronisation reference of a signal (200iii) from the network, the method comprising measuring (405) a synchronisation deviation of the clock from the synchronisation reference at different points in time, the measuring resulting in a set of synchronisation deviation measurement values;

10 performing (415) an analysis of the synchronisation deviation measurement values in a manner so that frequency components of the synchronisation deviation may be found; and generating analysis results in accordance with the analysis.

15 2. The method of claim 1 , wherein the performing of the analysis includes calculating the Allan variance of the synchronisation deviation for at least two subsets of the set of synchronisation deviation measurement values, the two subsets having different time intervals between two consecutive measurement values. >0

3. The method of claim 1. wherein the performing of the analysis includes performing a Fourier transform of the measurement values.

.5 4. The method of any one of claims 1-3. further comprising comparing (420) the analysis result to a predetermined value.

5. The method of any one of claims 1-4. further comprising comparing (420) the analysis result to a predetermined pattern. 50

6. The method of any one of claims 1-5. wherein the steps of measuring, performing an analysis and generating analysis results are repeatedly performed: and wherein a first analysis result obtained at a first point in time is compared (420) to a second analysis result obtained at a different point in time. 5

7. The method of any one of claims 4-6. wherein the method further comprises issuing (430) an alarm is if the comparison indicates that the synchronisation of the node is erroneous.

10 8. A method of supervising a synchronisation of nodes ( 1 10) in a communications system ( 100). the method comprising performing the steps of anyone of claims 1-7 for at least two nodes in the communications system; and comparing the analysis result of the at least two nodes. 15

9. A clock supervision unit (500) adapted to monitor the synchronisation of a node (1 10) having a clock (220) and being connected to a network ( 120). wherein the clock (220) of the node is adjusted in accordance with a synchronisation reference of a signal (200iii) from the network, the clock supervision unit comprising:

>0 a data measuring unit (510) having: an first input for receiving a clock signal (505) from the clock; a second input for receiving the signal from the network; the data measuring unit being adapted to: measure the synchronisation deviation in the clock signal from the 15 synchronisation reference: generate a set of synchronisation deviation measurement values; and the clock supervision unit further comprising: a data analysis unit (500) adapted to perform an analysis of the synchronisation deviation measurement values in a manner so that frequency components of the 50 synchronisation deviation may be found, and to generate an analysis result in response to the analysis.

10. The clock supervision unit (500) of claim 9. wherein the data analysis unit is adapted to perform a Fourier transform of the synchronisation deviation measurement values.

1 1. The clock supervision unit (500) of claim 9. wherein

5 the data analysis unit is adapted to calculate the Allan variance of the synchronisation deviation for at least two subsets of the set. the two subsets having different time intervals between two consecutive measurement values.

12. The clock supervision unil (500) of any one of claims 9- 1 1, wherein

[0 the data analysis unit is adapted to compare the analysis result to a predetermined value, and/or to a predetermined pattern, and/or to an analysis result obtained at a different point in time; and the data analysis unit is adapted to generate a comparison result in accordance with said comparison.

15

13. The clock supervision unit (500) of any one of claims 9-12. further comprising an alarm unit (525) adapted to issue an alarm if the comparison result indicates the synchronisation of the node is erroneous.

50 14. The clock supervision unit (500) of any one of claims 9-14. wherein the data measuring unit (510) further comprises a third input for receiving a second signal from the network, the second signal carrying a second synchronisation reference: the data measuring unit is further adapted to measure the synchronisation deviation in the clock signal from the second synchronisation reference and to generate a second set 15 of synchronisation deviation measurement values in response to the measuring; the data analysis unit (500) is further adapted to perform a second analysis of the second set of synchronisation deviation measurement values, and to generate a second analysis result in response to the analysis: and wherein the data analysis unit is further adapted to comparing the first analysis result to the i0 second analysis result, and to select a synchronisation reference according to which the clock should be adjusted in response to the comparison.

15. A radio base station (110) comprising a clock supervision unit of any one of claims 9- 14.

16. A communications system (100) comprising a clock supervision unit according to any one of claims 9-14.

17. The communication system of claim 16. wherein the communication system comprises at least two radio base stations (100) comprising a clock supervision unit (500) according to any one of claims 9-14. and wherein the communication system comprises means for comparing the analysis result from a first of the at least two radio base stations to the analysis result from another of said at least two radio base stations.

18. A computer program product for monitoring a synchronisation of a node (1 10). the node having a clock (220) and being connected to a network (120). wherein the clock (220) of the node is adjusted in accordance with a synchronisation reference of a signal (200iii) from the network, the computer program product comprising computer program code operable to. when run on computer means, execute the steps of: measuring a synchronisation deviation of the clock from the synchronisation reference at different points in time and generating synchronisation deviation measurement values; and performing an analysis of the synchronisation deviation measurement values in a manner so that frequency components of the synchronisation deviation may be found: and generating an analysis result in accordance with the step of performing an analysis.

19. Memory means arranged to fit in a radio base station ( 1 10), the memory means having stored thereupon the computer program product of claim 18.