Intelligent multi-dimensional active and passive combined tuning method for improving transmission capacity of multiband optical fiber

Abstract

The invention provides an intelligent multi-dimensional active and passive combined tuning method for improving the transmission capacity of a multi-band optical fiber. The active tuning comprises the steps of performing power optimization on a link side, and accelerating the optimization speed of an optimization algorithm through optimal configuration of power initialization so as to realize power pre-emphasis on a plurality of channels under multiple wavebands; aiming at unique damage characteristics of different channels at a receiving end, designing an equalization scheme responsible for characterization signal transmission characteristics of respective channel transmission characteristics, and integrating factors such as an equalizer tap and an equalization step length to realize optimal configuration; the passive tuning comprises the step of carrying out self-adaptive optimization on the signal of the transmitting end through the system SNR and NGMI parameters after the tuning of the link end and the receiving end. According to the invention, active and passive multi-dimensional joint tuning of the system is completed in multiple bands, so that the purpose of improving the single-fiber capacity of the multi-band system is achieved.

Description

Technical Field

[0001] The present invention relates to optical fiber communication technology, in particular to a wavelength division multiplexing multi-band optical fiber transmission technology. Background Art

[0002] Driven by continuous breakthroughs in innovative applications such as cloud computing, distributed storage architectures, the interconnected intelligent Internet of Things (IoT), 4K / 8K ultra-high-definition video, and fifth-generation mobile communications, global digital information traffic is experiencing exponential growth. This transformative demand is driving the evolution of optical fiber transmission networks toward a new generation of architectures with ultra-high throughput and low-latency transmission performance to meet the information society's demands for high-speed and massive data transmission. With the commercialization of C+L-band optical fiber communication systems, further increasing single-fiber transmission capacity is limited by factors such as the bandwidth of optoelectronic devices and fiber nonlinearities, making it challenging to further increase single-fiber transmission capacity. Mode-multiplexing-based few-mode or multi-core fiber transmission solutions are affected by mode coupling, intermodal dispersion, and the availability of related components. Currently, increasing the wavelength window (i.e., the transmission band) of optical fibers is an effective solution, such as introducing multi-band S+C+L fiber transmission in the S-band, or even full-band fiber transmission. However, cross-band power transfer caused by stimulated Raman scattering (SRS) and the gain and noise figure constraints of optical amplifiers in these new wavelength windows prevent the increase in S+C+L transmission capacity from scaling linearly with bandwidth. To increase the transmission capacity of S+C+L systems, hybrid amplification—a combination of Raman amplification and lumped amplification—can be used to compensate for the S-band power drop caused by SRS. However, the introduction of Raman amplifiers increases system complexity and power consumption; it also makes ongoing maintenance of the Raman amplifiers difficult. Furthermore, S+C+L systems have a large number of channels, and factors such as the number of Raman pump sources and amplification methods all affect performance and power consumption.

[0003] In existing C+L band transmission systems, erbium-doped fiber amplifiers (EDFAs) can amplify the entire C+L wavelength window, resulting in minimal overall performance variation. Therefore, overall system capacity is typically reduced through transceiver-side signal processing and flattened optical amplification to minimize performance variations between wavelengths, ensuring that overall transmission capacity approaches the channel capacity. However, the S band is more severely affected by SRS, resulting in insufficient amplifier gain and a high noise figure, making increasing transmission capacity challenging. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an optical fiber transmission optimization solution for optical power configuration, signal processing and entropy loading in multi-band scenarios in order to rationally utilize channel resources, give full play to the bandwidth advantages of the newly introduced wavelength window, and comprehensively improve the single-fiber transmission capacity.

[0005] The technical solution adopted by the present invention to solve the above technical problems is an intelligent multi-dimensional active and passive joint tuning method for improving the transmission capacity of multi-band optical fiber, which is characterized by comprising the steps of:

[0006] Active tuning steps on the upstream side of the receiver:

[0007] The system is divided into several bands, and the average transmit power and power gradient of each band are set as channel power setting parameters. The average power and power gradient of all bands are jointly optimized. During the optimization process, each update of the channel power setting parameters completes power pre-emphasis. The effect of the optimized power pre-emphasis is tested. When the average BER of all channels in the system converges to a stable value and the difference between the signal-to-noise ratio (SNR) of each channel meets the preset SNR flatness requirement, the link-side active tuning process is terminated, and the current channel power setting parameters are determined to optimize the power pre-emphasis configuration. The difference between the SNRs of each channel is recorded for subsequent passive tuning steps.

[0008] Active tuning steps on the receiving end:

[0009] The tap length, tap coefficient, and step size parameters of the equalizer for all bands are jointly optimized. When the average BER of all channels in the system converges to a stable value, the receiver's active tuning is terminated, and the tap coefficient and step size range set for each band are determined as the optimal configuration for the equalizer.

[0010] Passive tuning steps on the sender:

[0011] Channels are grouped according to their SNR differences, with channels with small SNR differences grouped together. At the transmitter, a passive tuning parameter set is established for the parameters related to transmission capacity in each group and jointly optimized. During the optimization process, when the system normalized generalized mutual information (NGMI) converges to stability, the optimal configuration of the current passive tuning parameter set is determined.

[0012] Optionally, the optimal configuration of the power pre-emphasis in the active tuning step upstream of the receiving end is implemented by the transmitting end and / or the link side.

[0013] Specifically, the parameters related to transmission capacity include shaping factor, error correction code overhead, peak-to-average power clipping ratio, and pulse shaping roll-off factor.

[0014] The beneficial effects of the present invention are as follows: the proposed active multi-dimensional parameter joint optimization scheme through link-level and transceiver joint equalization actively improves the channel response and reduces the performance difference between different wavelength windows due to the SRS effect; the passive tuning is performed by group entropy loading to achieve the maximum channel transmission capacity. Through the collaborative active and passive joint optimization mechanism of the transceiver and the optical fiber link, the overall capacity is improved in the multi-band optical fiber communication system, and the nonlinear distortion caused by the cross-band SRS power transfer is effectively suppressed; through the high-dimensional joint optimization of entropy-bit error rate BER-shaping coefficient by the multi-objective genetic algorithm, the parameter combination limitations of the traditional table lookup method are broken through, and under the constraint of the bit error rate threshold, the total system capacity can be further improved, and the spectrum utilization and nonlinear tolerance are collaboratively optimized. The present invention provides core support for ultra-high system capacity and low-damage transmission of multi-band systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Active and passive joint tuning process. DETAILED DESCRIPTION

[0016] Specifically, the present invention adopts an active multi-dimensional parameter joint optimization scheme through link-level and transceiver-end joint equalization:

[0017] 1. Active tuning of the upstream receiving end:

[0018] At the upstream receiver, a multi-dimensional power control scheme based on target criteria optimizes the signal's transmit power according to the SRS transfer characteristics, mitigating the impact of SRS. Due to the large number of wavelength channels in the S+C+L scheme, heuristic algorithms, such as the particle swarm algorithm, are employed to jointly optimize the average transmit power and power gradient across each band to reduce the complexity of multi-dimensional tuning.

[0019] The present invention sets the average transmission power P of each band x and power gradient T x , the i-th center frequency of a band is f i The power of the channel is P i,x =P x +T x (f i -f x ). Among them, f x The center frequency of the center channel of the band. Since the S+C+L band contains 3 bands and 3 power slopes, the present invention converts the power tuning of all channels into 6 channel power setting parameters. and The superscripts S, C, and L represent the corresponding bands respectively.

[0020] According to the above method, it is still used for other multi-band situations, that is, first divide the bands, and then determine the average transmission power P of each band x and power gradient T x , the channel power P of each channel in a band can be determined i,x .

[0021] The upstream of the receiving end may be the optical fiber link side and / or the transmitting end.

[0022] 2. Active tuning of the receiving end:

[0023] At the receiving end, characteristic equalization processing of multi-band signals is achieved by optimizing parameters such as the equalizer tap coefficients. Since the S-band is limited by the influence of SRS and noise, and the C-band and L-band are limited by the influence of fiber nonlinear effects, the tap length, step size and other parameters of the equalizer are jointly optimized through multi-dimensional joint parameter optimization to achieve targeted and effective compensation for damage in different bands.

[0024] 3. Passive tuning on the sending end:

[0025] Under the above-mentioned active optimization to improve channel response, at the transmitting end, by grouping all wavelength channels of S+C+L according to the channel response or theoretically predicted signal-to-noise ratio (SNR) model, a heuristic multi-dimensional parameter optimization method is adopted for different wavelength channel groups. The shaping factor, error correction code FEC overhead, peak-to-average power clipping ratio, pulse shaping roll-off factor, etc. of each wavelength group are jointly tuned to maximize the channel transmission capacity while meeting the error requirements, ultimately completing the capacity improvement of the entire single-fiber transmission system.

[0026] Example

[0027] The joint tuning method includes three steps: active tuning at the upstream of the receiving end, active tuning at the receiving end, and passive tuning at the sending end. Figure 1 As shown, taking the link side as the upstream of the receiving end as an example, the following steps are included:

[0028] 1. Active link-side tuning:

[0029] On the link side, signal transmission power optimization generates complex nonlinear catastrophic effects due to stimulated Raman scattering (SRS), leading to power transfer between channels. Specifically, power differences between signals at different wavelengths trigger the accumulation of nonlinear impairments (NLI) and the asymmetric distribution of amplified spontaneous emission (ASE) noise, exacerbating cross-band signal-to-noise ratio (SNR) degradation. Because these link effects are strongly correlated with SRS perturbations, increasing transmit power faces non-convex optimization constraints.

[0030] Therefore, based on the SRS transfer characteristics, a heuristic algorithm and a multi-dimensional power control scheme based on target criteria are used to optimize the signal transmission power on the optical fiber link to suppress the SRS impact. To address the optimization dimensionality explosion problem caused by the S+C+L multi-band system, dimensionality reduction control is achieved by defining characteristic parameters of the band power distribution.

[0031] Specifically, the system is divided into three bands: S, C, and L, and the average transmission power of each band is set to and And set three power inclinations accordingly and Based on heuristic algorithms, such as particle swarm optimization, the average power and power gradient are jointly optimized, thereby converting the power optimization of all wavelengths into an intelligent optimization process for the six channel power setting parameters. and The parameters complete a power pre-emphasis. Because different bands experience different power attenuation after passing through the channel, the attenuation at the receiving end is foreseen, so the attenuation amount is first placed on the link side for pre-compensation. That is, the channel power on each channel in the three bands is adjusted through the power setting parameters so that the receiving end has a flat receiving power when receiving the signal. This process is called power pre-emphasis. The power pre-emphasis effect after this optimization is tested. When the system's average signal-to-noise ratio (SNR) converges to stability (for example, the fluctuation error is less than 5% for more than 10 consecutive iterations or the preset number of iterations is reached) and the difference between the SNRs of each channel meets the preset SNR flatness requirement, the active link-side tuning process is terminated, and the current channel power setting parameters are determined to perform the optimal configuration of power pre-emphasis, significantly reducing the computational complexity of multivariable coupling optimization.

[0032] Optionally, the optimal configuration of power pre-emphasis may also be placed at the transmitter, or the transmitter and the link side may collaborate to achieve the optimal configuration.

[0033] The difference in SNR between channels is recorded for subsequent passive tuning steps.

[0034] 2. Active Tuning at the Receiver:

[0035] On the receiver side, the signal is affected by the inconsistent gain and noise figures of optical amplifiers in different bands in multi-band fiber transmission systems, resulting in large fluctuations in the optical signal-to-noise ratio (OSNR) of signals in different bands. After fiber transmission, different bands are affected by different nonlinear effects. The S-band is mainly affected by stimulated Raman scattering, while the C and L bands are mainly affected by the fiber Kerr nonlinear effect. This further exacerbates the power and OSNR fluctuations at different wavelengths. By optimizing parameters such as the equalizer tap coefficients, characterized equalization processing of multi-band signals is achieved. Since the S-band is limited by SRS and noise, and the C-band and L-band are limited by fiber nonlinear effects, the tap length, tap coefficient, step size and other parameters of the equalizer are jointly optimized to achieve targeted and effective compensation for damage in different bands. In order to achieve characteristic damage compensation on each channel, a differential genetic algorithm can be used to jointly optimize the tap length, tap coefficient, and step size parameters of the equalizers in the three bands. When the average signal-to-noise ratio (SNR) of all channels in the system converges to stability (for example, the fluctuation error is less than 5% for more than 10 consecutive iterations), the active tuning of the receiving end is terminated, and the tap coefficient and step size range set under the band are determined as the optimal configuration of the equalizer, which improves performance while reducing the signal processing complexity under multi-channel conditions.

[0036] 3. Passive Tuning on the Sending Side:

[0037] Based on the combined optimization of signal power in a multi-band optical fiber transmission system and digital signal processing at the receiver, the transmitter performs packet-based modulation, coding, and shaping to fully utilize channel spectrum resources and maximize transmission capacity. The degree of shaping is measured by entropy, and the source entropy is reflected by the size of the shaping factor. The shaping factor and coding overhead at the transmitter are passively tuned based on information such as the system's signal-to-noise ratio (SNR) and next-generation information (NGMI). By adjusting the distribution probability of the source constellation points and increasing the Euclidean distance between constellation points, the transmitter improves tolerance to ASE noise and channel capacity.

[0038] Because multiple bands have varying channel conditions, their signal-to-noise ratios (SNRs) vary. In multi-band transmission, to maximize channel capacity, all channels in the system are grouped based on previously determined differences in channel SNRs. Channels with similar SNRs, indicating similar impairments, are grouped together. SNR differences less than a preset threshold can be considered similar.

[0039] Based on preset entropy-SNR mapping rules, a passive tuning parameter set is established for each group's shaping factor, FEC overhead, peak-to-average power clipping ratio, and pulse shaping roll-off factor. This allows for joint multi-dimensional parameter tuning, maximizing transmission capacity while meeting bit error requirements, ultimately increasing the capacity of the entire single-fiber transmission system. At the transmitter, a heuristic algorithm is used to optimize the shaping factor, FEC overhead, peak-to-average power clipping ratio, and pulse shaping roll-off factor for each wavelength group. Once the system's normalized generalized mutual information (NGMI) converges and stabilizes, and the fluctuation error remains below 5% for more than 10 consecutive iterations, the optimal passive tuning parameter set is considered.

[0040] In multi-band systems, active and passive joint tuning achieves global optimization through a collaborative mechanism of "active compensation at the link receiving end + passive adaptation at the transmitting end." The active end dynamically adjusts power pre-emphasis allocation, equalizer coefficient size, and step size based on channel state information, and uses a multi-dimensional parameter optimization algorithm to close-loop optimize cross-band power balancing and nonlinear damage suppression. The passive end adaptively adjusts probability shaping parameters to match the statistical characteristics of the channel capacity based on the system SNR packet data and preset entropy-SNR mapping rules. The significance of this strategy lies in breaking through the limitations of a single tuning dimension: active compensation quickly responds to channel changes and suppresses interference such as stimulated Raman scattering; passive adaptation optimizes source entropy through probability shaping, approaching the Shannon limit while ensuring bit error rate. By performing active and passive joint tuning of the system's multi-dimensional parameters, the total capacity of the multi-band transmission system can be further improved, and spectrum utilization and nonlinear tolerance can be collaboratively optimized.

[0041] While the active-passive joint optimization framework proposed in this example is based on an S+C+L multi-band system, its core mechanism possesses cross-band universality. By studying the dynamic coupling relationship between full-band nonlinear interference and power transfer effects, a global optimization model based on joint power characterization and equalization can be constructed to simultaneously optimize multiple bands, breaking through the limitations of the traditional method of distinguishing bands and then optimizing each band separately. An intelligent optimization algorithm based on multi-dimensional parameter joint optimization can achieve capacity improvements for fiber optic transmission systems across the entire band range, bringing channel capacity closer to the Shannon limit.

Claims

1. An intelligent multi-dimensional active and passive joint tuning method for improving multi-band optical fiber transmission capacity, characterized by: Including steps: Active tuning steps on the upstream side of the receiver: The system is divided into several bands, and the average transmit power and power gradient of each band are set as channel power setting parameters. The average power and power gradient of all bands are jointly optimized. During the optimization process, each update of the channel power setting parameters completes power pre-emphasis. The effect of the optimized power pre-emphasis is tested. When the average signal-to-noise ratio (SNR) of all channels in the system converges to a stable state and the difference between the SNRs of each channel meets the preset SNR flatness requirement, the link-side active tuning process is terminated, and the current channel power setting parameters are determined to optimize the power pre-emphasis configuration. The difference between the SNRs of each channel is recorded for subsequent passive tuning steps. Active tuning steps on the receiving end: The tap length, tap coefficient, and step size parameters of the equalizer for all bands are jointly optimized. When the average BER of all channels in the system converges to a stable value, the receiver's active tuning is terminated, and the tap coefficient and step size range set for each band are determined as the optimal configuration for the equalizer. Passive tuning steps on the sender: Channels are grouped according to their SNR differences, with channels with small SNR differences grouped together. At the transmitter, a passive tuning parameter set is established for the parameters related to transmission capacity in each group and jointly optimized. During the optimization process, when the system normalized generalized mutual information (NGMI) converges to stability, the optimal configuration of the current passive tuning parameter set is determined.

2. The method according to claim 1, wherein: The optimal configuration of power pre-emphasis in the active tuning step upstream of the receiver is implemented at the transmitter.

3. The method according to claim 1, wherein: The optimal configuration of power pre-emphasis in the active tuning step upstream of the receiver is implemented on the link side.

4. The method according to claim 1, wherein: The optimal configuration of power pre-emphasis in the active tuning step upstream of the receiver is achieved collaboratively by the transmitter and the link side.

5. The method according to claim 1, wherein: Convergence to stability is manifested by a fluctuation error of less than 5% for more than 10 consecutive iterations.

6. The method according to claim 1, wherein: Parameters related to transmission capacity include shaping factor, error correction code overhead, peak-to-average power clipping ratio, and pulse shaping roll-off factor.

7. The method according to claim 1, wherein: A heuristic algorithm is used to optimize the average power and power gradient of all bands.

8. The method according to claim 1, wherein: A differential genetic algorithm is used to optimize the tap length, tap coefficient, and step size parameters of the equalizer for all bands.

9. The method according to claim 1, wherein: The method according to claim 1 is characterized in that a heuristic algorithm is used when establishing a passive tuning parameter set and optimizing the parameters related to transmission capacity in each group.

10. The method according to claim 1, wherein: The system is divided into three bands: S, C and L.

Images