Patent Application
20230136924
This application claims priority to European Patent Application No. 21206554.4 filed Nov. 4, 2021, the disclosure of which is hereby incorporated by reference in its entirety.
The invention relates to a method of operating a bidirectional optical transmission link and to a corresponding optical transmission link.
In the optical access networks, the demand for higher data rate keeps growing. At present, this is primarily caused by mobile fronthaul. Optical transceivers based on intensity modulation (IM) and direct detection (DD) are widely deployed in these systems due to the advantage of low costs. Coherent optical transceivers are not yet an option in a medium term for such applications due to their complexity and corresponding high costs.
Therefore, there is a high need for optical transmission systems using IM/DD, for example DWDM systems, to increase the data rate per wavelength while maintaining the reach. However, an increase in the data rate when using non-return-to-zero (NRZ) coded optical signals suffers from both bandwidth limitation of system components and chromatic dispersion of the optical path. The latter restriction is especially true when standard optical fibers are used and the optical transmission signals lie in the optical C-band, i.e. in the wavelength range between 1530 nm and 1565 nm.
In order to increase the data transmission rate to 25 Gb/s and beyond using IM/DD, current developments are focused on increasing the bandwidth of electrical drivers, optical modulators, lasers and receivers. However, even with such improved components, the use of standard C-band DWDM systems is limited as the transmission reach is significantly reduced due to the dispersion penalty. Alternatively, laser wavelengths within the O-band, i.e. within the wavelength range of 1260 nm to 1360 nm, are considered to avoid the chromatic dispersion, but the propagation loss is larger in the single-mode fiber and WDM as well as amplification is less mature. The above limitation is the essential obstacle to save the precious fiber resources for mobile fronthaul applications.
There is also academic research for providing optical transmission systems showing record-high data rate using advanced IM modulation formats, such as n-level pulse-amplitude modulation (PAM-n) and discrete multitone (DMT) modulation, using narrow-bandwidth components. However, these developments are still far from commercial readiness and require complicated digital signal processing on the receiver side (see N. Eiselt et al., “Performance Comparison of 112 Gb/s DMT, Nyquist PAM4 and Partial-Response PAM4 for Future 5G Ethernet-based Fronthaul Architecture,” Journal of Lightwave Technology, vol. 36, no. 10, pp. 1807-1814, May 2018; Z. Li et al., “Investigation on the equalization techniques for 10G-class optics enabled 25G-EPON”, Optical Express, vol. 25, no. 14, pp. 16228-16234, Jul. 2017).
An early tutorial paper describes an approach to use a communication channel from the far-end transceiver of an optical transmission link to optimize an equalizer (more correctly designated as a pre-emphasizer) that is provided in the transmitter at the near end. However, the approach uses the far end bit error rate (BER) to estimate the link penalty. This, however, is a rather inaccurate method as it comprises an indirect calculation and takes longer time to converge. It is shown in this paper that the transmitter pre-compensation clearly outperforms the equalization at the receiver side (see D. McGhan, “Electronic Dispersion Compensation”, OFC 2006, paper OWK1).
Thus, it is an object of the present invention to provide a method of operating a bidirectional optical transmission link which comprises an initialization method or process for automatically determining transmission parameters which contribute to reduce the inter-symbol interference and thus the BER. It is a further object of the invention to provide a corresponding transmission link.
The invention achieves these objects with the combinations of features as described herein. Further embodiments are apparent from the dependent claims.
The invention starts from the finding that both the bandwidth limitations of an optical transmission link comprising an optical transceiver at each end of an optical path, e.g. a standard optical fiber, that are introduced by the system components and the chromatic dispersion (CD) of the optical path can be pre-compensated or pre-emphasized at the respective transmitting side or transceiver, respectively. The method according to the invention allows to determine pre-compensation parameters, especially filter coefficients of a digital filter provided in the transmitting path and a chirp parameter that defines a positive or negative chirp created by an electro-optical converter, using the same information that is obtained at the receiving side or transceiver, respectively. According to the invention, the pre-compensation parameters are automatically determined during an initialization process for the optical transmission link. As a result, the invention contributes to reducing the transceiver costs while improving the performance of the optical transmission link, especially the reach of the transmission link. It shall be noted that within this description, the technical terms “pre-compensation” at the transmitting side of the optical transmission link and “pre-emphasis” are used equivalently.
According to the invention, the optical transmission link comprises a first and a second optical transceiver at a dedicated end of the optical transmission link and an optical transmission path connecting the first and second optical transceiver. The optical transceivers apply the methods of converting an electrical digital (usually binary) transmit signal into an electrical PAM-n transmit signal, pre-emphasizing the electrical PAM-n transmit signal by digital filtering and using the pre-emphasized electrical PAM-n signal as modulating signal for optically modulating an optical carrier signal, wherein the optical modulation method deployed is configured to create an optical PAM-n transmit signal with a positive or negative chirp. For initializing the optical transmission link, an initialization process is performed in which at least one loop comprising the following steps is run through:
The operating values for the filter parameters and the operating value for the chirp parameter are deployed during normal operation of the optical transmission link in the first optical transceiver.
Of course, this method can be applied for determining the pre-compensation parameters, i.e. the operating values for the filter equalization and the operating value for the chirp parameter, of both optical transceivers by performing this method twice, namely, for each transmission direction of the bidirectional optical transmission link.
According to an embodiment of the invention, the operating values for the filter parameters and the operating value for the chirp parameter are determined by the second optical transceiver and transmitted, via a communication channel, preferably an out-of-band communication channel, to the first optical transceiver, wherein the second optical transceiver knows the filter structure of the digital filter in the first optical transceiver and the ideal electrical PAM-n transmit signal or respective ideal sampled values. This requires to transmit only a few values, namely, the (final or preliminary) operating values for the filter parameters and the chirp parameter from the second to the first optical transceiver.
According to another embodiment, the operating values for the filter parameters and the operating value for the chirp parameter are determined by the first optical transceiver, wherein the sampled values of the electrical PAM-n training receive signal are transmitted, via a communication channel, preferably an out-of-band communication channel, to the first optical transceiver. In this case, a specific knowledge of the optical transceiver at the receiving end concerning the digital filter of the optical transceiver at the transmitting end is not required (of course, transceiver at the transmitting end knows the ideal electrical PAM-n transmit signal or respective ideal sampled values). However, a plurality of values, namely, the sampled values of the electrical PAM-n training receive signal, needs to be transmitted from the second to the first optical transceiver. If, for example a PAM-n training signal comprising 4096 symbols is used during the initialization process, a corresponding number of sampled values is to be transmitted to the respective other end.
According to an embodiment of the invention, the sampled values obtained and the ideal sampled values are used as input information for an algorithm that is configured to output operating values for the filter parameters, wherein, preferably, the sampled values obtained and the ideal sampled values are normalized (e.g. using the respective highest value of each set of values).
The algorithm may be, in one alternative, a zero-forcing algorithm, wherein the above-mentioned loop is run through preferably only once. As a zero-forcing algorithm is in principle configured to determine the inverse of the frequency response of the optical transmission link and thus completely eliminates inter-symbol interference (ISI), this alternative is disadvantageous if the channel frequency response has small magnitudes or even zeroes in the interesting range. The reason therefore is that the inverse of the frequency response in such ranges becomes high or even infinite so that noise components in these ranges are greatly amplified.
According to another alternative, the algorithm may be a least mean square error (LMSE) algorithm, wherein the above-mentioned loop is run through multiple times until the mean square error (MSE) between the sampled values obtained and the corresponding sampled values of the ideal electrical PAM-n transmit signal (that is determined in each pass of the loop) is interpreted as being an LMSE, wherein this interpretation can be made if the MSE reaches a minimum or is lower than a predetermined threshold value or if the MSE obtained in the current loop deviates from the MSE obtained in the previous loop by less than a predetermined threshold value. An LMSE algorithm is a more balanced algorithm and does not usually eliminate ISI completely but instead minimizes the total power of the noise and ISI components in the output.
According to a further embodiment of the invention, a change of the chirp parameter value is determined from a predetermined dependency of the chirp parameter from a remaining MSE between the sampled values obtained and the corresponding sampled values of the ideal electrical PAM-n transmit signal that is determined when using the operating values for the filter parameters and keeping the initial operating value for the chirp parameter.
This relationship may be stored in a control device of the optical transceiver that carries out the respective calculations.
Thus, according to the present invention, both the operating values for the filter parameters and the operating value for the chirp parameter are determined on the basis of the same information, namely, the two sets of sampled values of the electrical PAM-n receive signal and the ideal electrical PAM-n transmit signal.
The predetermined dependency of the chirp parameter from the remaining MSE can be obtained by a factory calibration process using an optical calibration transmission link comprising the first and second optical transceiver and two or more optical paths of known dispersion load or a single optical path comprising an adjustable dispersion component, or by simulating such a factory calibration process. Of course, a simulation is only an adequate substitute for a calibration process that comprises a measurement of the remaining MSE if the properties of the components are known with sufficient accuracy.
Such a factory calibration process may comprise two stages. In one stage, the following steps are carried out:
This first calibration dependency can be used, during an initialization of the optical transmission link, to determine the actual CD load from the MSE that remains after having optimized the filter parameters as explained above.
In an additional stage of the calibration process, the following steps are carried out:
This second calibration dependency can be used, during an initialization of the optical transmission link, to determine an optimum value for the chirp parameter from the CD load value that has been determined from the first calibration dependency.
Thus, the relationship between the CD load and the MSE can be used during the initialization process for determining the CD load once the operating values for the filter parameters have been determined using a selected algorithm, e.g. an LMSE algorithm. The relationship may be stored in the respective optical transceiver, especially in the control device thereof, e.g. as a two-dimensional table or an analytical dependency.
Likewise, the relationship between the CD load and the chirp parameter may be stored in the respective optical transceiver, especially in the control device thereof, e.g. as a two-dimensional table or an analytical dependency.
It is of course also possible to combine these two relationships to a single relationship between the chirp parameter and the MSE.
In the following, the invention will be described in more detail with reference to the drawings. In the drawings,
The first optical transceiver 102 is configured to receive an electrical digital transmit signal Tel,1 at a local input port and converts this signal into an optical PAM-4 transmit signal TPopt,1, which is fed to the respective optical fiber 108 at a remote transmit port of the optical transceiver 1. The optical PAM-4 transmit signal TPopt,1 travels via the optical fiber 108 and encounters a respective attenuation and chromatic dispersion (other transmission characteristics of the optical fiber, e.g. nonlinear effects, are neglected within this description) before it is received as optical PAM-4 receive signal RPopt,1 at a remote receive port of the optical transceiver 104. The optical transceiver 104 converts the optical PAM-4 receive signal RPopt,1 into an electrical digital receive signal Rel,1. The electrical digital transmit and receive signals Tel,1, Rel,1 may be binary digital signals, e.g. in the form of NRZ coded signals.
Likewise, the second optical transceiver 104 receives an electrical digital transmit signal Stx,2 at a local input port and converts this signal into an optical PAM-4 transmit signal TPopt,2. This signal is supplied to the further optical fiber 108 at a remote transmit port of the optical transceiver 104. The further optical fiber 108 guides the optical PRM-4 transmit signal in the direction to the first optical transceiver 1, wherein the signal is received as optical PAM-4 receive signal RPopt,2 at a remote receive port of the optical transceiver 102. The optical transceiver 102 converts this signal into an electrical digital receive signal Rel,2.
It is noted that, even if the embodiments described with reference to the drawings use a PAM-4 coding for the transmission of the optical signals over the optical path 106, any arbitrary PAM-n coding is appropriate to realize the present invention.
As already mentioned above, the signal transmission quality between the local input and output ports of the optical transceivers 102, 104 is decisively determined by both the bandwidth of the system components comprised by the optical transceivers 102, 104 and the attenuation and chromatic dispersion of the optical path, i.e. the optical fibers 108.
Each of the optical transceivers 102, 104 comprises the identical components, namely, a transmit signal processing unit 110 and an electro-optical converter 112 provided in a transmit path, a opto-electrical converter 114, a signal detector 116 and a receive signal converter 118 provided in a receive path and a control device 120 that is connected to a control port of the transmit signal processing unit 110, a control port of the electro-optical converter 112, an output port of the opto-electrical converter 114 and a control port of the receive signal converter 118.
The transmit signal processing units 110 convert the respective electrical digital transmit signal Tel,1, Tel,2 into a corresponding electrical PAM-4 transmit signal TPel,1 TPel,2, which is converted into the respective optical PRM-4 transmit signal TPopt,1, TPopt,1 by the corresponding electro-optical converter 112. As explained below, the electro-optical converters 112 are configured to introduce a predetermined positive or negative optical chirp into the respective optical PAM-4 transmit signal TPopt,1, TPopt,1.
The optical converters 112 may be realized as directly modulated lasers, which may provide an additional chirp control port configured to receive a respective chirp control signal created by the control device 120. In a further preferred alternative, the electro-optical converter is realized by a laser that creates an optical CW carrier signal and an optical modulator that is configured to modulate the optical CW carrier signal. In this alternative, the optical modulator is configured to introduce a controllable positive or negative chirp into the respective optical PAM-4 transmit signal TPopt,1, TPopt,1.
The control device 120 further controls at least a part of the signal processing properties of the transmit signal processing unit 110. Especially, as will be described below, the transmit signal processing unit 110 comprises a digital filter which is configured to pre-emphasize an electrical PAM-4 transmit signal TPel,1, TPel,2 created by the transmit signal processing unit 110. In the following, the pre-emphasized and the non-pre-emphasized electrical PAM-4 transmit signals will both be designated with the reference signs TPel,1, TPel,2, respectively.
In the receiving arm of each optical transceiver 102, 104, the respective opto-electrical converter receives the respective optical PAM-4 receive signal RPopt,1 RPopt,2 and converts it into an electrical PAM-4 receive signal RPel,1 RPel,2. The signal detector 116 samples the electrical PAM-4 receive signal RPel,1 RPel,2 at predetermined sampling points in time. The sampled values are fed to the receive signal converter 118 which is configured to convert these values into an electrical digital receive signal Rel,1, Rel,2.
As mentioned above, the transmit signal processing unit 110 is configured to pre-emphasize the electrical PAM-4 transmit signal TPel,1, TPel,2 and the electro-optical converter 112 is configured to add a positive or negative chirp to the optical PAM-4 transmit signal TPopt,1, TPopt,2. In this way, it is possible to at least partially compensate bandwidth limitations that are introduced by the system components and the linear dispersion, especially the chromatic dispersion, introduced by the optical path 106.
In the following, a more detailed embodiment of an optical transceiver 102, 104 will be described with reference to
According to the more detailed block diagram of
Preferably, the digital filter 124 is realized as a FIR filter as this filter type is simpler to design and reveals a high stability than other types of digital filters, especially infinite impulse response (IIR) filters. Preferably, the digital filter 124 may be realized as a 3-tap FIR filter, the filter properties of which are determined by three tap coefficients.
Further, the more detailed structure of the optical transceiver 1 shown in
It is also well known that an MZM can be controlled in such a way that a desired arm imbalance is achieved, i.e. the optical average power is controlled to be different in each of the arms. Such an imbalance can be reached in different ways: In one alternative, the imbalance may be created by the electrical driver 130. In this alternative, the electrical driver creates driving signals S+ and S− for each arm that reveal a different maximum amplitude. In another alternative, the arm imbalance may be achieved by providing an optical controllable attenuator in one of the arms. It is also possible to use a single-drive MZM and a controllable optical attenuator in one of the arms in order to achieve a controllable chirp functionality of the MZM.
In practice, it is advantageous to bias the MZM at the quadrature point and to use a swing for driving signals S+ and S− that represent the digital filter electrical PAM-4 transmit signal TPel,1 which matches the voltage V-pi of the MZM, i.e. the voltage at which a phase shift of pi is achieved by the respective phase shifter.
As apparent from
The control device 120 is further configured to supply values for the filter parameters, especially values for the tap coefficients of the preferably used FIR filter, to the digital filter 124. In
The control device 120 is also capable of modulating the bias voltage of the MZM according to a low-frequency signal (low-frequency here means at least two, preferably three orders lower than the symbol rate of the PAM-4 signal to be transmitted). In this way, and out-of-band communication channel (OOBC) can be realized as a low-frequency modulation component of the optical PAM-4 transmit signal TPopt,1. This OOBC makes it possible to transmit control information to the respective other optical transceiver 104.
In order to initialize the optical transmission link 100 (
As apparent from
Also the signal detector 116 is connected the control device 120. Via this connection, the signal detector 116 is able to transmit sampled values of an electrical PAM-4 receive signal TPel,2 to the control device 120 during an initialization procedure.
In the following, an initialization procedure for the optical transmission link 100 (
As explained above, an initialization procedure will be started by one of the optical transceivers 102, 104 if necessary, e.g. if the optical transmission link has been newly installed or is again put into operation after interruption.
During the initialization procedure and optical PAM-4 training transmit signal TPopt,1 (or general, a PAM-n training transmit signal) is created by the first optical transceiver 102 (for simplicity, in the following, the optical and electrical training signals are designated by the same reference signs as the optical training signals during a normal operation of the transmission link). For this purpose, the control device 120 controls the transmit signal converter 122 to create an electrical PAM-4 training transmit signal TPel,1 instead of converting the electrical digital transmit signal Tel,1 into a corresponding electrical PAM-4 signal TPel,1. The control device 120 may be configured to feed to the transmit signal converter 122 a respective training bit sequence that is to be used to create the electrical PAM-4 signal TPel,1 at the same symbol rate that is used for the electrical PAM-4 signal TPel,1 during normal operation.
At the beginning of the initialization procedure, the control device 120 sets the values for the filter parameters, e.g. the tap coefficients for a 3-tapFIR filter, and the value for the chirp parameter, e.g. the arm driving imbalance of the MZM, to arbitrary but reasonable initial values.
According to an embodiment of the invention, the initial value for the chirp parameter may be set to 0, i.e. the respective optical PAM-4 training transmit signal TPopt,1 is created without chirp. It is of course also possible, to start the initialization procedure with initial values for the digital filter that not introduce any pre-emphasis into the respective electrical PAM-4 training transmit signal TPel,1.
This optical PAM-4 training transmit signal TPopt,1 is received by the second optical transceiver 104 as optical PAM-4 training receive signal RPopt,1, wherein this signal is converted into a corresponding electrical PAM-4 training receive signal RPel,1. The signal detector 116 of the second optical transceiver 104 samples signal and feeds the sampled values of the electrical PAM-4 training receive signal RPel,1 to the control unit 120.
The control unit 120 of the second optical transceiver 104 knows the (sampled) values of an ideal electrical PAM-4 training transmit signal, which corresponds to the pre-set amplitude levels of the PAM-4 coding used.
in order to compare the sampled values of the electrical PAM-4 training receive signal RPel,1 and corresponding values of the ideal PAM-4 training transmit signal, both sets of values are normalized to the corresponding highest value.
In a next step, the control device 120 of the second optical transceiver 104 uses the two sets of (sampled) values based on the electrical PAM-4 training receive signal RPel,1 and on the ideal electrical PAM-4 training transmit signal as input values of a feed forward equalization (FFE) algorithm. This algorithm may be based on zero forcing LMSE methods. Such equalization methods are widely known so that it is not necessary to go into further detail in this respect.
If a zero forcing algorithm is applied, operating values for the filter parameters, e.g. the tap coefficients of FIR filter, are the direct result of the respective calculation that is carried out by the control device 120 of the second optical transceiver 104.
If a LMSE algorithm is applied, the control device 120 calculates a respective set of values for the filter parameters and the MSE for the respective electrical PAM-4 training receive signal RPel,1 that has been received in the respective loop (the LMSE algorithm is an iterating algorithm). In a next step, the filter parameter values are transmitted to the first optical transceiver 102 via the OOBC. The control device of the first optical transceiver sets the digital filter 124 to the values for the filter parameters received in the current loop.
Then, the next loop is initiated, wherein the control device 120 of the first optical transceiver 102 again creates an optical PAM-4 training transmit signal TPopt,1 using the values for the filter parameters that have been obtained in the previous loop. The second optical transceiver 104 receives the respective optical PAM-4 training receive signal RPopt,1, converts it into a corresponding electrical PAM-4 training receive signal RPel,1, samples this signal and calculates new values for the filter parameters and a new value for the MSE. If the value for the MSE meets a termination criterion, the LMS algorithm is finished. The values for the filter parameters that have been calculated in the last loop are used as operating values for the filter parameters during a normal operation of the optical transmission link 100.
In both cases, i.e. the use of a zero forcing algorithm or an LMSE algorithm, or generally spoken, for any FFE algorithm that might be applied, a remaining MSE is calculated by the control device 120 of the second optical transceiver 104. This value of the remaining MSE, i.e. the MSE value that is determined when the operational values for the filter parameters are used, can be used to determine the value for the chirp parameter as described below. It shall be noted in this respect that the remaining MSE is a differential value and depends on the initial value for the chirp parameter. That is the initial chirp parameter value determines the (remaining) MSE that is output by the algorithm and thus the (remaining) CD load that is to be compensated.
Next, in a first step, the relationship between the CD load introduced by the optical path and the chirp parameter and the relationship between the CD load and the (remaining) MSE (between sampled values of an appropriate electrical PAM-4 receive signal RPel,2 and an ideal PAM-4 transmit signal) will be explained with reference to
As apparent from
This dependency of the CD load on the (remaining) MSE can be stored in the control device 120, e.g. in the form of analytical expressions or tables (for one of more average powers of the electrical PAM-4 (training) receive signal RPel,2). This dependency can be used, during an initialization process of the optical transmission link, to determine the actual CD load from the (remaining) MSE by performing the process for determining the optimized filter parameter values.
The dashed curve in
Of course, also the dependency of the chirp parameter on the CD load can also be stored in the control device 120 in the form of an analytical dependency or a table.
As mentioned above, the two dependencies according to
It is also possible to combine the two dependencies to a single dependency of the chirp parameter (e.g. the arm driving imbalance) on the (remaining) MSE (for one or more given average powers of the signal received).
According to an embodiment of the invention, the initialization procedure is started with a chirp value of zero. In this way, the lacking uniqueness of the dependency of the chirp parameter on the CD load can be avoided at least in cases in which the general dispersion properties of the optical path are known. For example, it a standard (single-mode) fiber is used, it is clear that the CD load can only be positive, provided that no over-pre-compensation is effected when the initialization process starts, which is guaranteed if the process starts with no pre-compensation.
It shall finally be noted that the calculations (or generally spoken the actions required to determine the values of the filter parameters and the values of the chirp parameter) can also be performed by the control device of the first optical transceiver 102. In this case, it is required to transmit the sampled values of the electrical PAM-4 training receive signal RPel,1 from the second optical transceiver 104 to the first optical transceiver 102 via the OOBC.
It is also possibly to distribute these actions on the two control devices 120. For example, the filter parameter values may be determined and the calculation of the (remaining) MSE may be carried out by the control device 120 of the second optical transceiver 104 and the chirp parameter value may be determined by the control device 120 of the first optical transceiver 102, wherein this control device receives the MSE value (and of course the filter parameter values) form the control device 120 of the second optical transceiver 104.
Of course, the roles of the first and second optical transceiver described above may also be exchanged.
In this way, the invention makes it possible to increase the data rate (while maintaining the reach and the BER required) by determining, during an initialization procedure, optimum parameters for the pre-emphasis by means of a digital filter and for the chirp by means of controlling the electro-optical converter.
| Number | Date | Country | Kind |
|---|---|---|---|
| 21206554.4 | Nov 2021 | WO | international |
