CN102497244B - Stimulant clock recovery method - Google Patents

Abstract

An analog clock recovery method provided by the present invention, according to the clock frequency difference and phase difference calculated by the clock adjustment algorithm of the IEEE1588 protocol, simulates the recovery of the clock frequency and phase, so as to realize the clock on the equipment that does not have the hardware clock recovery capability recovery; and the time after the recovery of the local clock is obtained through integer calculations, avoiding the problem of low precision in floating-point calculations, and when determining the time after the recovery of the local clock, the time required to execute the method is accumulated to further improve the time accuracy, thereby providing a When the clock accuracy reaches the microsecond level, the time after the recovery of the local clock is obtained through integer calculation, and the problem of low precision in floating-point calculation is avoided, and the time required to execute the method is accumulated when determining the time after the recovery of the local clock, which improves the time precision.

Description

An Analog Clock Recovery Method

technical field

The invention relates to the field of network synchronous clocks, in particular to an analog clock recovery method.

Background technique

Traditional clock synchronization refers to synchronizing clocks distributed in various places. Simply put, it is to keep the frequency and phase of the clocks the same. Clock synchronization is widely used in many fields, such as industrial control, automatic production, measurement and other industries that require precise time. At present, with the gradual IP-based network, the clock synchronization capability of the packet network has gradually received people's attention. There are two main aspects: first, the packet network can carry TDM (time division multiplexing mode) services, and provide a mechanism for TDM service clock recovery, so that TDM services still meet certain performance indicators after passing through the packet network; second, the packet network can Like a TDM network, it provides a high-precision network reference clock to meet the synchronization requirements of network nodes or terminals. Now there are several protocols that can realize clock synchronization, such as ntp protocol, IEEE1588 protocol, etc. The time precision of the ntp protocol is at the millisecond level, while the time precision of the IEEE1588 protocol is at the nanosecond level. Therefore, the IEEE1588 protocol has been widely used in recent years due to its superiority in precision.

The IEEE1588 protocol itself fully considers the transmission delay and processing time of the message in its design, and relies on hardware to obtain the sending and receiving time stamps of the message, minimize the transmission delay of the message, and calculate the time between devices. The master-slave clock frequency difference f and phase difference p are used to adjust the local clock, that is, the clock adjustment algorithm. Therefore, the IEEE1588 protocol has high requirements for hardware, and it will eventually restore the frequency and phase of the local clock. Generally, the frequency and phase of the local clock of the slave clock hardware device can be modified, but not all hardware devices support the modification of the frequency and phase of the local clock. Therefore, for devices without hardware clock recovery capability, realizing clock synchronization is an urgent problem to be solved.

Contents of the invention

The invention provides a simulated clock recovery method, which enables the IEEE1588 protocol to be implemented on a device without hardware clock recovery capability, and provides a clock precision of microsecond level for other application protocols on the device.

The technical means that the present invention adopts is as follows: a kind of clock recovery method of simulation, comprises:

The parameter acquisition step is to obtain the elapsed time t _pass of the local hardware clock, the time t _{adj_i-1} after the last recovery of the local clock, the frequency recovery value f _{adj_i-1} after the last recovery of the local clock, and the frequency ratio between the local clock and the main clock The obtained adjustment coefficient C, and the clock frequency difference _{f_i} and the phase difference p obtained by the clock adjustment algorithm;

The step of calculating the time t _now after the recovery of the local clock, the time after the recovery of the local clock is the time t _{adj_i-1} after the last recovery of the local clock plus the product of the elapsed time of the local hardware clock t _pass and the adjustment coefficient C , and then the sum of the phase difference p.

Further, the parameter acquisition step also includes:

Obtain the time t _local of the local hardware clock and the hardware time t _hard at the last restoration, and obtain the elapsed time t _pass of the local clock hardware according to (t _local -t _hard ).

Further, the adjustment coefficient C obtained from the frequency ratio of the local clock and the master clock is obtained by (1+f _{adj_i-1} /10 ⁹ ) or (10 ⁹ +f _{adj_i-1} )/10 ⁹ .

Further, the calculated time after the recovery of the local clock also includes:

In the step of updating the frequency offset integer value f _{adj_i} , the frequency recovery value f _{adj_i-1} after the last recovery of the local clock is accumulated with the current clock frequency difference f _{_i} ;

The step of updating the recovered time t _{adj_i} is to save the determined local clock recovery time t _now as the recovered time;

The step of updating the restored hardware time t _hard is to save the local clock time t _local obtained this time as the restored hardware time.

Further, the step of determining the time t _now after the recovery of the local clock also includes the step of accumulating compensation values for executing the algorithm.

Further, the compensation value includes the execution time of the parameter acquisition step, the execution time of determining the time that the local clock should pass, the execution time of the accumulated phase difference p and the time t _{adj_i-1} after the last recovery of the local clock.

Further, a step of recovering the local clock every predetermined time, and calculating the time t _now after recovery.

Further, both f _{adj_i-1} and t _{adj_i-1} are zero during the initial adjustment calculation.

Based on this, an analog clock recovery method provided by the present invention is used in conjunction with the hardware clock, and the clock frequency difference and phase difference calculated according to the clock adjustment algorithm of the IEEE1588 protocol can be used to restore the analog clock frequency and phase without hardware Implement clock recovery on devices with clock recovery capabilities; and obtain the time after local clock recovery through integer calculations, avoiding the problem of low precision in floating-point calculations, and add up the time required to execute the method when determining the time after local clock recovery, and further The time precision is improved, and then a clock with microsecond precision is provided to meet the application of upper layer protocols, such as service-based OAM protocols.

Description of drawings

FIG. 1 is a schematic flow chart of the simulated clock recovery method of the present invention.

Detailed ways

The principles and features of the present invention are described below in conjunction with the accompanying drawings, and the examples given are only used to explain the present invention, and are not intended to limit the scope of the present invention.

The present invention is carried out based on following design:

In the IEEE1588 network, according to the interactive message, the time difference offset and the path delay delay between the master clock and the slave clock can be finally calculated, and then the current frequency difference between the master clock and the slave clock can be calculated through the existing clock adjustment algorithm f _{_i} and phase difference p, according to the frequency difference fi (unit ppb, one billionth) and phase difference p (unit nanoseconds), first make the frequency of the slave clock consistent with the frequency of the master clock, and then the time of the slave clock It is consistent with the time of the main clock, which means that the TOD time deviation of the two clocks is at the nanosecond level.

The physical quantities available when performing clock recovery include:

f _{_i} , represents the frequency difference (in ppb), and represents the i-th frequency recovery value;

p, represents the phase difference (in nanoseconds);

t _local , indicating the local hardware clock (in nanoseconds);

The physical quantities that need to be calculated include:

t _now , indicating the recovery time (in nanoseconds) adjusted by software simulation;

Intermediate variables include:

t _pass indicates the elapsed time of the local hardware clock;

C, represents the adjustment coefficient C obtained by the frequency ratio of the local clock and the master clock;

f _{adj_i-1} , indicating that the currently adjusted frequency restores an integer value;

t _{adj_i-1} , indicating the adjustment time after the last recovery;

t _hard , indicating the hardware time at the last recovery;

The basic formula for clock recovery is as follows:

t _now =t _{adj_i-1} +t _pass *C+p; (Formula 1)

t _pass =t _local -t _hard (Formula 2)

C=(1+f _{adj_i-1} /10 ⁹ ) or, C=(10 ⁹ +f _{adj_i-1} )/10 ⁹ (Formula 3)

f _{adj_i} = f _{adj_i-1} + f _{_i} ; (Formula 4)

t _adj-i = t _now ; (Formula 5)

t _hard = t _local (Formula 6)

When the computer performs calculations, dividing the integer value of f _{adj_i-1} by 1 billion will result in a decimal, and the floating-point calculation precision of the device is not as high as the integer calculation precision, so the formula can be modified as follows:

t _now ＝t _{adj_i-1} +(t _local -t _hard )*(10 ⁹ +f _{adj_i-1} )/10 ⁹ +p; (Formula 7)

Through the clock recovery formula, the current frequency difference pin phase difference p between the master clock and the slave clock is calculated. Among them, the frequency difference f _{_i} is a signed integer value, and the unit is ppb. If f _{_i} is a positive number, it means that the local clock frequency is slow, and it needs to go f _{_i} ns more than before; otherwise, it means that the frequency of the local clock is fast, and it takes Walk less f _{_i} ns than before. The phase difference p is also a signed number, and the unit is ns. If p is a positive number, it means that the local time is slower than the time of the main clock by pns, and the local time needs to be increased by pns; otherwise, it means that the local time is slower than the time of the main clock by pns. Decrement pns from local time.

Based on the above idea, a kind of analog clock recovery method provided by the present invention includes:

The parameter acquisition step is to obtain the elapsed time t _pass of the local hardware clock, the time t _{adj_i-1} after the last recovery of the local clock, the frequency recovery value f _{adj_i-1} after the last recovery of the local clock, and the frequency ratio between the local clock and the main clock The obtained adjustment coefficient C, and the clock frequency difference _{f_i} and the phase difference p obtained through the clock adjustment algorithm;

The step of calculating the time t _now after the recovery of the local clock, the time after the recovery of the local clock is the time t _{adj_i-1} after the last recovery of the local clock plus the product of the elapsed time of the local hardware clock t _pass and the adjustment coefficient C , and the sum of the phase difference p

As an embodiment of the inventive method, the flow process as shown in Figure 1:

Step S301: Read the time t _local of the local clock, and obtain the time t _adj-i after the last restoration and the hardware time t _hard at the last restoration, and obtain the elapsed time t of the local clock hardware according to (t _local -t _hard ) _pass , and obtain the frequency recovery value f _{adj_i-1} after the last recovery of the local clock, the adjustment coefficient C obtained from the frequency ratio of the local clock and the master clock, and the clock frequency difference f _{_i} and phase difference p obtained through the clock adjustment algorithm; Wherein, the adjustment coefficient C obtained by the frequency ratio between the local clock and the main clock is obtained by (1+f _{ddj_i-1} /10 ⁹ ) or (10 ⁹ +f _{adj_i-1} )/10 ⁹ ;

Step S302: Calculate the time after the recovery of the local clock according to Formula 1 or 7. After obtaining the elapsed time t _pass of the local clock hardware and the adjustment coefficient C, calculate the second value that the local clock should pass during this period, that is, (t _local -t _hard )*(1+f _{adj_i-1} /10 ⁹ ) or ( t _local -t _hard )*(10 ⁹ +f _{adj_i-1} )/10 ⁹ ; secondly, add this period of time to the time after the last recovery, so as to get the time after the local clock recovery; finally add this The phase value adjusted twice is used to obtain the final local time; preferably, formula 7 is used to calculate the time after the local clock is recovered, which avoids the time error caused by the floating-point calculation accuracy of the device being not as high as the integer calculation accuracy.

Step S303: Update the frequency recovery value, that is, the previous accumulated frequency recovery value is accumulated with the current frequency recovery value, indicating that from the current time, the local clock should run at the recovered frequency, that is, f _{adi_i} = f _{adj_i-1} + f _{_i} ;

Step S304: Update the time after the last recovery, that is, save the time calculated this time as the last recovery time, t _adj-i =t _now

Step S305: Updating the hardware time at the time of the last recovery, that is, saving the time calculated this time as the last recovery time, which is used for the next calculation of the clock recovery t _hard =t _local .

In this embodiment, considering the time accuracy, the analog method is used to restore the clock, and the execution time of the method needs to be considered. Preferably, in step S302, when calculating the time after the recovery of the local clock, it is necessary to accumulate the execution time of the parameter acquisition step, the execution time of determining the time that the local clock should pass, and the accumulated phase difference p and the time t _local of the local clock The compensation value of the execution time is used to improve the time accuracy. The determination of the compensation value can be obtained through running analysis. According to the experiment, the compensation value is preferably 90 microseconds.

Preferably, the local clock recovery is performed every predetermined time, and the step of calculating the recovery time t _now is performed. Wherein, f _{adj_i-1} and t _{adj_i-1} are both zero in the initial adjustment calculation.

When obtaining the time after the recovery of the local clock, the values of the frequency difference f and the phase difference p are set to 0, and the running time at the recovered frequency can be obtained according to the above method flow.

With the technical means of the present invention, according to experimental tests, the frequency difference between the master and slave clocks can finally be adjusted to 7ppb.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the present invention. within the scope of protection.

Claims (6)

1. a clock recovery method for simulation, comprising:

Parameter acquiring step, obtains local hardware clock lapse of time t _pass, local clock is last after recovering time t _{adj_i-1}, local clock is last after recovering frequency retrieval value f _{adj_i-1}, the adjustment coefficient C that obtained by the frequency ratio of local clock and master clock and obtain the poor f of clock frequency by clock adjustment algorithm _{_ i}with phase difference p;

Calculate local clock and recover rear time t _nowstep, the time t after the time after described local clock recovers, to be that local clock is last recovered _{adj_i-1}add local hardware clock lapse of time t _passwith the product of described adjustment coefficient C, then with phase difference p sum;

Described parameter acquiring step also comprises:

Obtain the time t of local hardware clock _localhardware time t while recovery with the last time _hard, according to (t _local-t _hard) obtain local clock hardware lapse of time t _pass;

The adjustment coefficient C being obtained by the frequency ratio of local clock and master clock is by (1+f _{adj_i-1}/ 10 ⁹) or (10 ⁹+ f _{adj_i-1})/10 ⁹obtain.

2. method according to claim 1, is characterized in that, calculates local clock and recovers also to include after the rear time:

Renewal frequency skew integer value f _{adj_i}step, the frequency retrieval value f that local clock is last after recovering _{adj_i-1}, with the poor f of this clock frequency _{_ i}cumulative;

Upgrade the time t after recovering _{adj_i}step, time t after recovering by this definite local clock _nowsave as the time after recovery;

Upgrade the hardware time t while recovery _hardstep, by the local clock time t of this acquisition _localsave as the hardware time after recovery.

3. method according to claim 1, is characterized in that, the time t after described definite local clock recovers _nowstep in also comprise the cumulative step of carrying out this algorithm compensation value.

4. method according to claim 3, is characterized in that, the time t that described offset comprises the parameter acquiring step time of implementation, determine time of implementation and the cumulative phase difference p of the time that local clock should pass by and local clock is last after recovering _{adj_i-1}time of implementation.

5. according to the method described in claim 1 to 4 any one, it is characterized in that, carry out local clock recovery every predetermined time, calculate and recover rear time t _nowstep.

6. method according to claim 1, is characterized in that, f when first adjustment is calculated _{adj_i-1}and t _{adj_i-1}be zero.