Network timing synchronization systems

• US 20100118895A1
• Filed: 09/21/2009
• Published: 05/13/2010
• Est. Priority Date: 09/22/2008
• Status: Active Grant

• Current Assignee: Codrut Radu Radulescu
• Original Assignee: Codrut Radu Radulescu
• Inventors: Radulescu, Codrut Radu
• Granted Patent: US 8,416,812 B2
• Time in Patent Office: Days
• US Class Current: 370/503
• CPC Class Codes: G04G 7/00 Synchronisation radio-controlled time-pieces G04R
• CPC Class Codes: H04J 3/0667 Bidirectional timestamps, e.g. NTP or PTP for compensation of clock drift and for compensation of propagation delays arrangements for monitoring round trip delays in packet switching networks H04L43/0864

Abstract

A method and means synchronize timing of a follower system to a reference system. A Hierarchical CFF function (“HCFF”) is applied to a set of Correction Factor Functions types (“CFFs”) or a set of other HCFF. Each CFF type uses the same input data set specific to that type and generates at least one Correction Factor Solution (“CFS”) for each of the CFF, wherein the CFS consists of only CF or the CFS consists of both i) CF and ii) a SACF. The HCFF takes as input a set of CFS and generates at least one CFS, wherein the CFS consist of only the CF, or the CFS consists of both the i) CF and ii) a SACF.

Claims

1. A method for synchronizing timing of a follower system to a reference system, comprising the steps:

applying a Hierarchical CFF function (“

HCFF”

) to a set of Correction Factor Functions types (“

CFFs”

) or a set of other HCFF, wherein

there are several types of CFF and HCFF according to the aspect of the timing they process and according to the process they use

each CFF type uses the same input data set specific to that type and generates at least one Correction Factor Solution (“

CFS”

) for each of the CFF, wherein the CFS consists of only CF or the CFS consists of both CF and a SACF, and

the HCFF takes as input a set of CFS and generates at least one CFS, wherein the CFS consist of only the CF, or the CFS consists of both the CF and a SACF; and

using said CFS to synchronize timing.

8. A method for synchronizing timing of a follower system to a reference system as recited by claim 1, wherein the HCFF input data comprises a set of CSF_k, and wherein using the CFS_k to synchronize timing comprises the steps of:

generating SACF_k=0 if SACF_k<

SACF_kHT, wherein SACF_kHT is a hierarchical threshold for the CFF_k; and

generating

$C<br><br><br><br>F=\underset{k}{\sum <br><br>}<br><br>C<br><br><br><br>F_k\star <br><br>S<br><br><br><br>A<br><br><br><br>F<br><br><br><br>C_k/\underset{k}{\sum <br><br>}<br><br>S<br><br><br><br>A<br><br><br><br>F<br><br><br><br>C_k$

 

generating wherein k indicates each input CFS.

11. A method for synchronizing timing of a follower system to a reference system as recited by claim 8, comprising the steps:

SACF_kHT=Max((1+(SACF_i+SACF_e)/2),min_HT),

SACF_i is the SACF generated by a unidirectional CFF on a link L from the ingress TSPs and

SACF_e is the egress SACF generated by said CFF on the link L from egress TSPs; and

generating SACF as the average of the SACF_k used in the generation of CF.

12. A method for synchronizing timing of a follower system to a reference system, as recited by claim 8, wherein the SACF is direct proportional to the ratio between the number of TSPs inside the SV and the number of TSP in the extended vicinity (“

EV”

).

14. A method for synchronizing timing of a follower system to a reference system as recited by claim 1, wherein the CFF functions take their input TSPs from links to a plurality of time reference sources.

15. A method for synchronizing timing of a follower system to a reference system as recited by claim 1 wherein the synchronization of the system comprises:

a virtual oscillator that adjusts time stamps TS(n+1) with a CF(n) provided by a control process by using the formula;

V_TS(n+1)=V_TS(n)+(TS(n+1)−

TS(n))*CF(n), wherein TS(n), TS(n+1) are generated from a fixed, free running oscillator, the control process would have adjusted an adjustable oscillator by the same CF(n), and which adjusted time stamp V_TS(n+1) is relative to a previous adjusted V_TS(n) from a time stamp TS(n); and

V_TS (n+1) is further passed to the timing synchronization algorithm (claim 1) as if said time stamps were taken with an adjustable oscillator.

16. The method for synchronizing timing of a follower system to a reference system as recited by claim 1, wherein the timing to be synchronized is UTC, comprises the steps:

determining an offset (offset_UTC), between a Reference UTC and Follower node UTC local time counter (F_CTR_UTC) by finding symmetrical propagation delay packets in between the directions and the associated Teons, which is done by determining local minimums of the difference between the arrival and departure time, for both directions, minPD_d=ABS(A_d(x)−

D_d(x)), wherein d is i for ingress or e for egress;

and finding a minimum of (ABS(minPD_i−

minPD_e))/2=offset_UTC(ix,iy,ex,ey) wherein ex,ey indicate the egress TSP (eTSP) and ix,iy indicate ingress TSP (iTSP) corresponding to said minimum of the pair combination of local minimums from each direction, which eTSP and iTSP define a roundtrip Teon; and

determining the UTC CFS from the round trip Teon that provides the solution for timing correction.

17. A method for synchronizing timing of a follower system to a reference system as recited by claim 1, further providing a secondary follower timing synchronization from a secondary reference time node comprising the steps:

time stamping within a fixed delay interval the time counter (time stamp=RNCOm) and a secondary physical or virtual time counter (time stamp=kNCOm);

time stamping with the time counter reception of secondary events (RNCOn);

determining through interpolation secondary time stamps (kNCOn) associated with the secondary follower timing from the RNCOn time stamps, by using the formula kNCOn=kNCOm+(kNCOm+1−

kNCOm)*(RNCOn−

RNCOm)/(RNCOm+1−

RNCOm);

synchronizing the secondary follower with the CFS computed from kNCOn time stamps.

2. A method for synchronizing timing of a follower system to a reference system as recited by 1, wherein the input data of a CFF function comprises a set of timestamp pairs (“

TSP”

); and

wherein the CFF comprises the steps of;

making a set of Teons, wherein

each Teon includes a plurality of departure time and arrival time TSPs,

each of said TSP is associated with a communication event,

all the communication events occur within a synchronization update interval, and

are separated by a defined minimum time interval;

extracting at least one raw correction factor from at least one of the Teons; and

using said at least one raw correction factors (“

CFn”

) to synchronize the timing.

3. A method for synchronizing timing of a follower system to a reference system as recited by claim 2, wherein additional Teons are made from newly received time stamp pairs and processed as they arrive.

4. A method for synchronizing timing of a follower system to a reference system as recited by claim 2, wherein extraction of a CFn from a Teon comprises:

calculating CFn(x,y)=((A(x)−

A(y))−

(D(x)−

D(y))/(A(x)−

A(y)), wherein

D(x) is the departure TS and A(x) is the arrival time of a TSPx included in the Teon; and

D(y) is the departure TS and A(y) is the departure TS of a different TSP included in the same Teon;

5. A method for synchronizing timing of a follower system to a reference system as recited claim 4, wherein

any CFn(x,y), CFn(v,w) pair included in the Teon and used to synchronize timing further satisfies the relation |CFn(x,y)−

CFn(v,w)|=AT0_SACFn<

AT0 wherein

any of a TSPk part of the Teon is at least once part of said relation, wherein k is any of x,y,v,w and AT0 is proportional with the precision sought for the synchronization.

6. A method for synchronizing timing of a follower system to a reference system as recited by claim 2, wherein extraction of a CFn from a Teon consists of:

calculating CFn(x,y)=(D(x)−

D(y))/(A(x)−

A(y)), wherein

D(x) is the departure TS and A(x) is the arrival time of a TSPx included in the Teon; and

D(y) is the departure TS and A(y) is the arrival TS of a different TSP included in the same Teon.

7. A method for synchronizing timing of a follower system to a reference system as recited by claim 6, wherein

any CFn(x,y), CFn(v,w) pair included in a Teon and used to synchronize timing, further satisfies the relation |CFn(x,y)−

CFn(v,w)|=AT1_SACFn=<

AT1 wherein

any of the TSPk part of the Teon is at least once part of said relation, wherein k is any of x,y,v,w and AT1 is proportional with the precision sought for the synchronization.

9. A method for synchronizing timing of a follower system to a reference system as recited by claim 2, wherein the step of using said at least one raw correction factor (“

CFn”

) to synchronize the timing, further comprises generating a self assessed confidence factor (“

SACF”

) for each CF proportional with the probability of CF being correct.

10. A method for synchronizing timing of a follower system to a reference system as recited by claim 2, wherein the step of synchronizing the timing comprises the steps of:

adding a number SACF_kn=ATx−

ATx_SACF_n to a bin SACF(CF_k)=SACF_k of a histogram where SACF(CF_k)=SACF(CF_k)+SACF_kn, which SACF(CF_k) bin covers an interval that includes the CFn value, and wherein

x is 1 or 2 according with the function type that generated CF_k, ATx_SACF_n was determined during calculation of CF_k, and were said addition is performed for every said CF_k included in the SUI cycle;

identifying a range of bins that provide the highest sum of the bin content for the bins in the range, and said range of bins being the solution vicinity (“

SV”

); and

calculating the solution with the formula

$C<br><br><br><br>F=\underset{k}{\sum <br><br>}<br><br>C<br><br><br><br>F_k\star <br><br>S<br><br><br><br>A<br><br><br><br>F<br><br><br><br>C_k/\underset{k}{\sum <br><br>}<br><br>S<br><br><br><br>A<br><br><br><br>F<br><br><br><br>C_k$

 

wherein CF_k is the bin value and SACF_k is the bin content;

13. A method for synchronizing timing of a follower system to a reference system as recited by claim 10, wherein the bins of the histogram dynamically change their width, by shrinking when the network load decreases and enlarging during periods of network congestion, and such dynamic change is inversely proportional to changes of SACF_kHT, by increasing the size of the bins such that the relation SACF_kHT>

min_HT is maintained until the maximum bin size is reached, wherein the maximum bin size depends on the total number of bins and the CF range to be covered.

18. A system that measure the environmental conditions (“

EC”

) comprising

a device transmitting messages (“

TxSensor”

) containing the departure time stamp of said massages wherein the time stamp is a reading of a counter at the time the message is transmitted and said counter has sufficient stages to prevent wraparound in between transmission time, and the clock is provided by a local oscillator which frequency fluctuates mostly due to the environmental condition that is measured;

a device (“

RxBase”

) that receives said time stamped messages and time stamps the arrival time of said messages with a high stability clock wherein the time stamp is the reading of a counter driven by the high stability clock at the moment said packet is received and said counter has sufficient stages to prevent wraparound in between receive events;

means to determine the frequency offset between the TxSensor and the RxBase; and

means to extract the EC by interpreting the frequency offset.

19. A method to correct the frequency of a free running oscillator comprising the steps of:

sampling the environmental conditions (“

EC”

) at a rate sufficient to capture measurable changes that will determine modifications of the frequency;

calculating an interpolated value CF(EC) corresponding to the environmental conditions sample EC from a CF(EEC(k−

1)) and a CF(EEC(k)) wherein;

CF(EEC(k−

1)) and CF(EEC(k)) are elements of a multidimensional array of correction factors CF indexed on equivalent environmental conditions EEC (ECOCA),

EEC(k−

1) is the closest index in the ECOCA array that is smaller than EC,

EEC(k) is the closest index in the ECOCA array that is greater than EC, and

t(n) is the time of measurement of the EC; and

applying the CF(EC) correction to the oscillator.

20. A method to correct the frequency of a free funning oscillator as recited by claim 19, further comprising the steps of:

sampling the environmental conditions EC(k) during a synchronization update interval (“

SUI”

) cycle, at a rate sufficient to capture measurable changes that will determine modifications of the frequency;

calculating an equivalent environment condition (EEC(SUI)) over said SUI wherein EEC is a constant EC under which the oscillator will generate the same number of oscillations as counted during the measured SUI cycle under fluctuating EC;

Interpolating the CF(EEC(m)) from the CF(EEC(SUI)) that is provided by the synchronization methodology at the end of the current SUI cycle with the CF(EEC(previous_SUI)) that was provided by the synchronization methodology at the end of the previous SUI cycle wherein,

EEC(m) is the closest index in the table to both the EEC(SUI) and EEC(previous_SUI)); and

integrating the interpolated CF(EEC(m)) into the content of location indexed by EEC(m) of the ECOCA.

21. An apparatus for network simulation, comprising:

a high stability frequency oscillator;

a logic unit that is configured to perform the functions of at least one Timing Reference Node (“

Master”

) running a timing synchronization protocol (“

TSP”

);

a means to time stamp the TSP packets;

a means to provide sequences of propagation delay values (“

PD”

) equivalent with PD experienced by the TSP packets for the network configurations that are emulated; and

a means to modify said time stamp of the TSP packets generated by each Master wherein the modification is the addition or the subtraction of a PD associated with the propagation path between each reference node and an equipment under test.

22. An apparatus for network simulation as recited by claim 21, wherein the means to provide the sequenced PD values, uses a method that comprises the steps of:

generating PD from one or more reference files, and combinations thereof and defined by formula;

PD(n)=SUM_k(f_k(PD(n+g_k))), wherein

each file includes records of PD(n) data measured over certain network topologies,

f_k is an function that modifies the amplitude of PDV and g_k is a function that modifies the phase of the reference file consisting of PDV0(n) records, and

k is a natural number, f and g are real functions including identity and null function, as in f_k(x)=x g_k=0.

23. A telecommunication protocol for timing synchronization with identical message structure for both directions, comprising:

an arrival time stamp A(n) and an associated sequence ID from a protocol packet received from the opposite end;

a departure time stamp information D(m) of the previously transmitted packet; and

a departure sequence ID, wherein the sequence ID of the packets transmitted by the follower node are only odd numbers and the sequence ID transmitted by the reference node are only even numbers.