FPGA-based high-speed serial transceiver implementation of optical interconnection transmission method and system

Abstract

The application relates to the technical field of optical fiber communication, in particular to a high-speed serial transceiver implementation optical fiber interconnection transmission method and system based on FPGA, which comprises the following steps: the bit error rate BER and the eye pattern characteristic parameter of an optical fiber link are monitored in real time, a link characteristic baseline is constructed, a multi-dimensional weighted quality index is combined, an improved particle swarm optimization algorithm is driven to dynamically adjust the pre-emphasis coefficient and the receiving equalization parameter of the transceiver, the signal jitter characteristic is extracted, the clock data recovery loop bandwidth is dynamically adjusted based on the jitter characteristic and the bit error rate, the forward error correction mode is dynamically switched through the hysteresis threshold value logic, the time delay and the error correction capability are balanced, the baseline parameter is continuously updated through a closed-loop feedback mechanism, and real-time optimization of the optical fiber communication link quality is realized.

Description

Technical Field

[0001] This invention relates to the field of optical fiber communication technology, specifically to a method and system for realizing optical fiber interconnection transmission using a high-speed serial transceiver based on FPGA. Background Technology

[0002] With the rapid development of optical fiber communication, optical fiber interconnect technology has become a core transmission solution due to its high bandwidth, low latency, and anti-interference capabilities. However, in high-speed optical fiber transmission, signals are affected by factors such as fiber dispersion, temperature fluctuations, and mechanical vibrations, leading to problems such as increased bit error rate (BER), eye diagram closure, and increased jitter. Furthermore, traditional solutions often employ fixed parameter equalization, which can easily limit the quality of optical fiber communication links.

[0003] To address this, a method and system for implementing fiber optic interconnect transmission using a high-speed serial transceiver based on FPGA is proposed. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for realizing fiber optic interconnect transmission using a high-speed serial transceiver based on FPGA. This invention constructs a link characteristic baseline by real-time monitoring of the bit error rate (BER) and eye diagram characteristic parameters of the fiber optic link. Combined with multi-dimensional weighted quality indicators, it drives an improved particle swarm optimization algorithm to dynamically adjust the pre-emphasis coefficients and receive equalization parameters of the transceiver. It extracts signal jitter characteristics, dynamically adjusts the clock data recovery loop bandwidth based on jitter characteristics and BER, and dynamically switches the forward error correction mode through hysteresis threshold logic to balance latency and error correction capability. Through a closed-loop feedback mechanism, it continuously updates the baseline parameters to achieve real-time optimization of the fiber optic communication link quality.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for implementing fiber optic interconnect transmission using a high-speed serial transceiver based on FPGA includes:

[0007] S1. Obtain the bit error rate (BER) and eye diagram characteristic parameters of the received signal during optical fiber transmission. The eye diagram characteristic parameters include eye height (EH) and eye width (EW). Combine the BER and eye diagram characteristic parameters to establish a link characteristic baseline.

[0008] S2. Construct a multi-dimensional weighted quality index (MQM). The MQM is calculated by fusing a normalized scoring function that integrates BER, EH, and EW with weighting coefficients. Based on the MQM, the improved particle swarm optimization algorithm in the FPGA is used to adjust the transmit pre-emphasis coefficient and receive equalization parameters of the high-speed serial transceiver.

[0009] S3. Extract the jitter characteristics of the received signal in the optical fiber link, including the jitter standard deviation and jitter peak value, and dynamically adjust the clock data to restore the CDR loop bandwidth according to the jitter characteristics and the current BER, so as to balance jitter tracking capability and noise suppression performance, and use nonlinear mapping rules to precisely control the recovery of optical signal;

[0010] S4. Filter the BER data and calculate the trend slope. Combine the BER, trend slope and preset hysteresis threshold logic to dynamically switch the FEC mode and optimize the signal transmission delay.

[0011] S5. Update baseline parameters in real time to form a closed loop and continuously optimize the quality of fiber optic links.

[0012] Preferably, the optical fiber transmission process includes photoelectric conversion, modulation and demodulation processes. The FPGA modulates the optical signal through an embedded high-speed serial transceiver and transmits the modulated optical signal to the optical fiber link. At the receiving end, the optical signal is recovered through a demodulation process.

[0013] Preferably, the improved particle swarm optimization algorithm in S2 includes the following process:

[0014] S2.1. Initialize the parameter range of the pre-weighting coefficient and the receiving equalization parameter, and randomly generate an initial particle swarm, where the position of each particle represents a set of parameter combinations;

[0015] S2.2. Calculate the MQM for each particle under the current parameters; estimate the gradient of the MQM with respect to the parameters using the finite difference method; and update the particle's current velocity based on the gradient;

[0016] S2.3. Continue iterating; if the rate of change of MQM is less than the threshold, then convergence is determined and the optimal parameter combination is output.

[0017] Preferably, the nonlinear mapping rule in S3 is implemented through a lookup table, wherein the input of the lookup table is the joint quantization range of the jitter indicator factor and the bit error rate, and the output is the preset CDR bandwidth level.

[0018] Preferably, the joint quantization interval is used to perform joint analysis on the jitter interval and the BER interval; the jitter interval includes a low jitter interval, a medium jitter interval, and a high jitter interval; the BER interval includes a low bit error interval, a medium bit error interval, and a high bit error interval.

[0019] Preferably, the hysteresis threshold logic in S4 is defined as follows:

[0020] When BER exceeds the threshold T up And if the trend slope is greater than 0, switch from low error correction mode to high error correction mode;

[0021] When BER is less than the threshold T upFurthermore, if the trend slope is less than 0, the system switches from high error correction mode to low error correction mode.

[0022] A fiber optic interconnect transmission system is implemented using a high-speed serial transceiver based on FPGA. The system includes:

[0023] The link characteristic baseline construction module obtains the bit error rate (BER) and eye diagram characteristic parameters of the received signal during optical fiber transmission. The eye diagram characteristic parameters include eye height (EH) and eye width (EW). The link characteristic baseline is established by combining the BER and eye diagram characteristic parameters.

[0024] The equalization optimization module constructs a multi-dimensional weighted quality index (MQM). The MQM is calculated by fusing a normalized scoring function that integrates BER, EH, and EW with weighting coefficients. Based on the MQM, the improved particle swarm optimization algorithm within the FPGA is used to adjust the transmit pre-emphasis coefficient and receive equalization parameters of the high-speed serial transceiver.

[0025] The optical signal control module extracts the jitter characteristics of the received signal in the optical fiber link, including the jitter standard deviation and jitter peak value, and dynamically adjusts the clock data to restore the CDR loop bandwidth based on the jitter characteristics and the current BER, so as to balance jitter tracking capability and noise suppression performance. It uses nonlinear mapping rules to precisely control the recovery of optical signal.

[0026] The signal transmission delay optimization module filters the BER data and calculates the trend slope. Combining the BER, trend slope, and preset hysteresis threshold logic, it dynamically switches the FEC mode to optimize the signal transmission delay.

[0027] The feedback module updates baseline parameters in real time, forming a closed loop and continuously optimizing the quality of the fiber optic link.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] 1. This invention constructs a multi-dimensional weighted quality index (MQM) by real-time monitoring of BER, eye diagram feature parameters (EH / EW), and jitter characteristics. Combined with an improved particle swarm optimization algorithm, it dynamically adjusts the pre-emphasis coefficient and receive equalization parameters, which can effectively improve the quality of optical fiber communication links.

[0030] 2. This invention calculates the partial derivatives of the parameters to be optimized (transmit pre-emphasis coefficient and receive equalization parameter) based on multi-dimensional weighted quality indicators to obtain the gradient, and updates the speed of the particle swarm optimization algorithm based on the gradient, and dynamically adjusts the transmit pre-emphasis coefficient and receive equalization parameter, thereby realizing the convergence capability of the particle swarm optimization algorithm and effectively improving the quality of optical fiber communication links.

[0031] 3. This invention determines the joint quantization interval based on jitter characteristics and BER data, dynamically adjusts the CDR bandwidth level through a lookup table (LUT), and dynamically switches the FEC mode through hysteresis threshold logic, thereby improving the signal jitter suppression capability and effectively improving the quality of optical fiber communication links.

[0032] 4. This invention achieves closed-loop control of fiber optic link quality by dynamically adjusting baseline parameters, thereby effectively improving the quality of fiber optic communication links. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating a method for implementing fiber optic interconnect transmission using a high-speed serial transceiver based on FPGA, provided as an embodiment of the present invention.

[0034] Figure 2 This is a schematic diagram of a fiber optic interconnect transmission system implemented using a high-speed serial transceiver based on an FPGA, as provided in an embodiment of the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Example 1

[0037] To improve the quality of the A-fiber communication link, a high-speed serial transceiver based on FPGA was used to implement a fiber optic interconnect transmission method, such as... Figure 1 A flowchart illustrating a method for implementing fiber optic interconnect transmission using a high-speed serial transceiver based on FPGA, provided in this embodiment of the invention, includes:

[0038] A method for implementing fiber optic interconnect transmission using a high-speed serial transceiver based on FPGA includes:

[0039] S1. Obtain the bit error rate (BER) and eye diagram characteristic parameters of the received signal during optical fiber transmission. The eye diagram characteristic parameters include eye height (EH) and eye width (EW). Combine the BER and eye diagram characteristic parameters to establish a link characteristic baseline.

[0040] The link characteristic baseline is [BER0, EH0, EW0]; where BER0 represents the initial bit error rate; EH0 represents the initial eye height; and EW0 represents the initial eye width.

[0041] Furthermore, the eye diagram feature parameters are generated by capturing 1000 consecutive UI (Unit Interval) signal waveforms in the FPGA and superimposing them; and the eye height EH and eye width EW are obtained.

[0042] S2. Construct a multi-dimensional weighted quality indicator (MQM). The MQM is calculated by integrating the normalized scoring function of BER, EH, and EW with the weight coefficients.

[0043] The multi-dimensional weighted quality index is expressed as follows:

[0044] MQM=κ1*S BER +κ2*S EH +κ3*S EW ;

[0045] Wherein, MQM represents a multi-dimensional weighted quality indicator; S BER S represents the bit error rate normalization scoring function; EH S represents the normalized scoring function for eye level. EW κ1, κ2, and κ3 represent the eye width normalized scoring function; κ1, κ2, and κ3 represent the weighting coefficients.

[0046] The bit error rate normalization scoring function is:

[0047]

[0048] Where BER represents the bit error rate; BER re Bit Error Rate (BER) is a reference value. re Take 1e-3;

[0049] The normalized scoring function for eye height is:

[0050]

[0051] Among them, EH re Reference values ​​indicating eye height;

[0052] The normalized eye width scoring function is as follows:

[0053]

[0054] Among them, EW re Reference values ​​indicating eye width;

[0055] Furthermore, the optical fiber transmission process includes photoelectric conversion, modulation, and demodulation. The FPGA modulates the optical signal through an embedded high-speed serial transceiver and transmits the modulated optical signal to the optical fiber link. At the receiving end, the optical signal is recovered through a demodulation process.

[0056] Adjusting the transmit preemphasis coefficient and receive equalization parameters of a high-speed serial transceiver based on an improved particle swarm optimization algorithm driven within the MQM-driven FPGA.

[0057] Furthermore, the improved particle swarm optimization algorithm in S2 includes the following optimization process:

[0058] S2.1. Initialize the parameter range of the pre-weighting coefficient and the receiving equalization parameter, and randomly generate an initial particle swarm, where the position of each particle represents a set of parameter combinations;

[0059] The parameter combination consists of the pre-emphasis coefficient and the coefficients of each equalizer tap;

[0060] S2.2. Calculate the MQM for each particle under the current parameters; estimate the gradient of the MQM with respect to the parameters using the finite difference method; and update the particle's current velocity based on the gradient;

[0061] The gradient is:

[0062]

[0063] Among them, ILI j This indicates that MQM is trying to optimize parameter p. j Partial derivatives; Δ represents the perturbation step size; MQM(p j +Δ) represents the MQM value after a positive perturbation; MQM(p j -Δ) represents the MQM value after a negative perturbation;

[0064] The update speed is:

[0065]

[0066] Among them, v i t+1 v represents the update velocity of particle i at time t+1; i t ι represents the velocity of particle i at time t; j Indicates the parameter p to be optimized j The weights; M represents the number of parameters to be optimized;

[0067] The parameter dynamic adjustment process is as follows:

[0068] p j new =p j old +η*ILI j ;

[0069] Where, p j new Indicates the parameter p to be optimized j The adjustment value; p jold Indicates the parameter p to be optimized j The initial value; η represents the learning rate;

[0070] S2.3. Continue iterating; if the rate of change of MQM is less than the threshold, then convergence is determined and the optimal parameter combination is output;

[0071] This embodiment constructs a multi-dimensional weighted quality index (MQM) by real-time monitoring of BER, eye diagram feature parameters (EH / EW), and jitter characteristics. Combined with an improved particle swarm optimization algorithm, it dynamically adjusts the pre-emphasis coefficient and receive equalization parameters, which can effectively improve the quality of optical fiber communication links.

[0072] This invention calculates the partial derivatives of the parameters to be optimized (transmit pre-emphasis coefficient and receive equalization parameter) based on multi-dimensional weighted quality indicators to obtain the gradient, and updates the speed of the particle swarm optimization algorithm based on the gradient, and dynamically adjusts the transmit pre-emphasis coefficient and receive equalization parameter, thereby realizing the convergence capability of the particle swarm optimization algorithm and effectively improving the quality of optical fiber communication links.

[0073] S3. Extract the jitter characteristics of the received signal in the optical fiber link, including the jitter standard deviation and jitter peak value, and dynamically adjust the clock data to restore the CDR loop bandwidth according to the jitter characteristics and the current BER, so as to balance jitter tracking capability and noise suppression performance, and use nonlinear mapping rules to precisely control the recovery of optical signal;

[0074] The jitter feature is extracted from the received signal using LSTM;

[0075] Furthermore, the nonlinear mapping rule in S3 is implemented through a lookup table, the input of which is the joint quantization range of the jitter indicator factor and the bit error rate, and the output is the preset CDR bandwidth level.

[0076] The jitter indicator factor is obtained based on the jitter standard deviation and jitter peak value; the jitter indicator factor is:

[0077] TJ = β1*DJ + β2*IJ;

[0078] Where TJ represents the jitter indicator factor; DJ represents the jitter standard deviation; IJ represents the jitter peak value; β1 and β2 represent the jitter indicator factor influence coefficients;

[0079] Furthermore, the joint quantization interval is used to perform joint analysis on the jitter interval and the BER interval; the jitter interval includes a low jitter interval, a medium jitter interval, and a high jitter interval; the BER interval includes a low bit error interval, a medium bit error interval, and a high bit error interval.

[0080] When TJ≤TJ1, it is the low jitter interval; when TJ1<TJ<TJ2, it is the medium jitter interval; when TJ2<TJ, it is the high jitter interval; TJ1 and TJ2 are the threshold values ​​of the first jitter indicator factor and the second jitter indicator factor, respectively.

[0081] When BER≤BER1, it is the low error rate range; when BER1<BER<BER2, it is the medium error rate range; when BER2<BER, it is the high error rate range; BER1 and BER2 are the first and second bit error rate thresholds, respectively; the lookup table is shown in Table 1.

[0082] Table 1 Lookup Table

[0083] jitter range Error range CDR bandwidth level Low jitter range Low error range low width Low jitter range Medium error range medium width Low jitter range High error range Height and Width Medium jitter range Low error range medium width Medium jitter range Medium error range medium width Medium jitter range High error range Height and Width High jitter range Low error range Height and Width High jitter range Medium error range Height and Width High jitter range High error range Height and Width

[0084] S4. Filter the BER data and calculate the trend slope. Combine the BER, trend slope and preset hysteresis threshold logic to dynamically switch the FEC mode and optimize the signal transmission delay.

[0085] The filtering process includes a weighted average of real-time BER data and historical BER data, with weights of 0.1 and 0.9, respectively.

[0086] Filtered BER data within a past time window are selected for linear regression to fit the slope of the straight line, thus obtaining the trend slope; the trend slope is:

[0087]

[0088] Where k represents the trend slope; n represents the number of sampling points in the time window; t i Represents a timestamp; BER i Indicates t i The filtered BER value at time 1;

[0089] Furthermore, the hysteresis threshold logic in S4 is defined as follows:

[0090] When BER exceeds the threshold T up And if the trend slope is greater than 0, switch from low error correction mode to high error correction mode;

[0091] When BER is less than the threshold T up Furthermore, if the trend slope is less than 0, the system switches from high error correction mode to low error correction mode.

[0092] This embodiment determines the joint quantization interval based on jitter characteristics and BER data, dynamically adjusts the CDR bandwidth level through a lookup table (LUT), and dynamically switches the FEC mode through hysteresis threshold logic, thereby improving the signal jitter suppression capability and effectively improving the quality of optical fiber communication links.

[0093] S5. Update baseline parameters in real time to form a closed loop and continuously optimize fiber optic link quality; update baseline parameters as follows: the link characteristic baseline is [BER0*0.1+BER new *0.9,EH0*0.1+EH new *0.9,EW0*0.1+EW new *0.90]; where BER new Indicates the current bit error rate; EH new Indicates current eye level; EW new This indicates the current eye width.

[0094] This embodiment achieves closed-loop control of fiber optic link quality by dynamically adjusting baseline parameters, thereby effectively improving the quality of fiber optic communication links.

[0095] The updated baseline parameters are then transmitted to the sender until the MQM change rate is less than the threshold.

[0096] To verify the effectiveness of the FPGA-based high-speed serial transceiver method for fiber optic interconnection provided in this embodiment, the quality of different methods applied to an A-type fiber optic communication link was compared. The quality comparison was conducted by comparing the transmission delay, jitter standard deviation, and BER of different methods, as shown in Table 2. Specifically, Method 1 is the FPGA-based high-speed serial transceiver method for fiber optic interconnection provided in this embodiment; Method 2 is Method 1 without considering the adjustment of pre-emphasis coefficients and receive equalization parameters; Method 3 is Method 1 without considering particle swarm optimization; Method 4 is Method 1 without considering dynamic adjustment of CDR bandwidth; and Method 5 is Method 1 without considering dynamic switching of FEC mode.

[0097] Table 2. Quality Comparison of Different Methods Applied to A Fiber Optic Communication Links

[0098] method Transmission delay BER jitter standard deviation Method 1 40ns 5e-6 0.1 UI Method 2 60ns 1e-4 0.3 UI Method 3 50ns 8e-5 0.2 UI Method 4 55ns 3e-5 0.2 UI Method 5 50ns 8e-5 0.3 UI

[0099] As shown in Table 2, the method for implementing fiber optic interconnection transmission using a high-speed serial transceiver based on FPGA provided in this embodiment has a certain degree of effectiveness.

[0100] Example 2

[0101] To improve the quality of the B-fiber communication link, a high-speed serial transceiver based on FPGA was used to implement the fiber optic interconnect transmission system, such as... Figure 2 A schematic diagram of a fiber optic interconnect transmission system based on a high-speed serial transceiver using an FPGA, provided as an embodiment of the present invention, includes:

[0102] A fiber optic interconnect transmission system is implemented using a high-speed serial transceiver based on FPGA. The system includes:

[0103] The link characteristic baseline construction module obtains the bit error rate (BER) and eye diagram characteristic parameters of the received signal during optical fiber transmission. The eye diagram characteristic parameters include eye height (EH) and eye width (EW). The link characteristic baseline is established by combining the BER and eye diagram characteristic parameters.

[0104] The link characteristic baseline is [BER0, EH0, EW0]; where BER0 represents the initial bit error rate; EH0 represents the initial eye height; and EW0 represents the initial eye width.

[0105] Furthermore, the eye diagram feature parameters are generated by capturing 1000 consecutive UI (Unit Interval) signal waveforms in the FPGA and superimposing them; and the eye height EH and eye width EW are obtained.

[0106] The balanced optimization module constructs a multi-dimensional weighted quality indicator MQM, which is calculated by integrating the normalized scoring function of BER, EH and EW with weight coefficients.

[0107] The multi-dimensional weighted quality index is expressed as follows:

[0108] MQM=κ1*S BER +κ2*S EH +κ3*S EW ;

[0109] Wherein, MQM represents a multi-dimensional weighted quality indicator; S BER S represents the bit error rate normalization scoring function; EH S represents the normalized scoring function for eye level. EW κ1, κ2, and κ3 represent the eye width normalized scoring function; κ1, κ2, and κ3 represent the weighting coefficients.

[0110] The bit error rate normalization scoring function is:

[0111]

[0112] Where BER represents the bit error rate; BER re Bit Error Rate (BER) is a reference value. re Take 1e-3;

[0113] The normalized scoring function for eye height is:

[0114]

[0115] Among them, EH re Reference values ​​for eye height;

[0116] The normalized eye width scoring function is as follows:

[0117]

[0118] Among them, EW re Reference values ​​indicating eye width;

[0119] Adjusting the transmit preemphasis coefficient and receive equalization parameters of a high-speed serial transceiver based on an improved particle swarm optimization algorithm driven within the MQM-driven FPGA.

[0120] Furthermore, the improved particle swarm optimization algorithm in S2 includes the following optimization process:

[0121] S2.1. Initialize the parameter range of the pre-weighting coefficient and the receiving equalization parameter, and randomly generate an initial particle swarm, where the position of each particle represents a set of parameter combinations;

[0122] The parameter combination consists of the pre-emphasis coefficient and the coefficients of each equalizer tap;

[0123] S2.2. Calculate the MQM for each particle under the current parameters; estimate the gradient of the MQM with respect to the parameters using the finite difference method; and update the particle's current velocity based on the gradient;

[0124] The gradient is:

[0125]

[0126] Among them, ILI j This indicates that MQM is trying to optimize parameter p. j Partial derivatives; Δ represents the perturbation step size; MQM(p j +Δ) represents the MQM value after a positive perturbation; MQM(p j -Δ) represents the MQM value after a negative perturbation;

[0127] The update speed is:

[0128]

[0129] Among them, v i t+1 v represents the update velocity of particle i at time t+1; i t ι represents the velocity of particle i at time t; j Indicates the parameter p to be optimized j The weights; M represents the number of parameters to be optimized;

[0130] The parameter dynamic adjustment process is as follows:

[0131] p j new =p j old +η*ILI j ;

[0132] Where, p j new Indicates the parameter p to be optimized j The adjustment value; p j old Indicates the parameter p to be optimized j The initial value; η represents the learning rate;

[0133] S2.3. Continue iterating; if the rate of change of MQM is less than the threshold, then convergence is determined and the optimal parameter combination is output;

[0134] The optical signal control module extracts the jitter characteristics of the received signal in the optical fiber link, including the jitter standard deviation and jitter peak value. Based on the jitter characteristics and the current BER, it dynamically adjusts the clock data to restore the CDR loop bandwidth in order to balance jitter tracking capability and noise suppression performance. It uses a lookup table (LUT) to precisely control the recovery of the optical signal.

[0135] The jitter feature is extracted from the received signal using LSTM;

[0136] Furthermore, the optical fiber transmission process includes photoelectric conversion, modulation, and demodulation. The FPGA modulates the optical signal through an embedded high-speed serial transceiver and transmits the modulated optical signal to the optical fiber link. At the receiving end, the optical signal is recovered through a demodulation process.

[0137] Furthermore, the improved particle swarm optimization algorithm includes the following optimization process:

[0138] Initialize the parameter range of the pre-emphasis coefficient and the receiver equalization parameter, randomly generate the initial particle swarm, and the position of each particle represents a set of parameter combinations;

[0139] Calculate the MQM for each particle with the current parameters; estimate the gradient of the MQM with respect to the parameters using the finite difference method; and update the particle's current velocity based on the gradient.

[0140] Continue iterating; if the rate of change of MQM is less than the threshold, then convergence is determined and the optimal parameter combination is output.

[0141] Furthermore, the nonlinear mapping rule is implemented through a lookup table, the input of which is the joint quantization range of the jitter indicator factor and the bit error rate, and the output is the preset CDR bandwidth level.

[0142] The jitter indicator factor is obtained based on the jitter standard deviation and jitter peak value; the jitter indicator factor is:

[0143] TJ = β1*DJ + β2*IJ;

[0144] Where TJ represents the jitter indicator factor; DJ represents the jitter standard deviation; IJ represents the jitter peak value; β1 and β2 represent the jitter indicator factor influence coefficients;

[0145] Furthermore, the joint quantization interval is used to perform joint analysis on the jitter interval and the BER interval; the jitter interval includes a low jitter interval, a medium jitter interval, and a high jitter interval; the BER interval includes a low bit error interval, a medium bit error interval, and a high bit error interval.

[0146] When TJ≤TJ1, it is the low jitter interval; when TJ1<TJ<TJ2, it is the medium jitter interval; when TJ2<TJ, it is the high jitter interval; TJ1 and TJ2 are the threshold values ​​of the first jitter indicator factor and the second jitter indicator factor, respectively.

[0147] When BER≤BER1, it is the low error rate range; when BER1<BER<BER2, it is the medium error rate range; when BER2<BER, it is the high error rate range; BER1 and BER2 are the first and second bit error rate thresholds, respectively; the lookup table is shown in Table 3.

[0148] Table 3 Lookup Table

[0149] jitter range Error range CDR bandwidth level Low jitter range Low error range Low width Low jitter range Medium error range medium width Low jitter range High error range Height and Width Medium jitter range Low error range medium width Medium jitter range Medium error range medium width Medium jitter range High error range Height and Width High jitter range Low error range Height and Width High jitter range Medium error range Height and Width High jitter range High error range Height and Width

[0150] Furthermore, the joint quantization interval is used to perform joint analysis on the jitter interval and the BER interval; the jitter interval includes a low jitter interval, a medium jitter interval, and a high jitter interval; the BER interval includes a low bit error interval, a medium bit error interval, and a high bit error interval.

[0151] The signal transmission delay optimization module filters the BER data and calculates the trend slope. Combining the BER, trend slope, and preset hysteresis threshold logic, it dynamically switches the FEC mode to optimize the signal transmission delay.

[0152] The filtering process includes a weighted average of real-time BER data and historical BER data, with weights of 0.1 and 0.9, respectively.

[0153] Filtered BER data within a past time window are selected for linear regression to fit the slope of the straight line, thus obtaining the trend slope; the trend slope is:

[0154]

[0155] Where k represents the trend slope; n represents the number of sampling points in the time window; t i Represents a timestamp; BER i Indicates t i The filtered BER value at time 1;

[0156] Furthermore, the hysteresis threshold logic is defined as follows:

[0157] When BER exceeds the threshold T up And if the trend slope is greater than 0, switch from low error correction mode to high error correction mode;

[0158] When BER is less than the threshold T up Furthermore, if the trend slope is less than 0, the system switches from high error correction mode to low error correction mode.

[0159] The feedback module updates baseline parameters in real time, forming a closed loop and continuously optimizing the quality of the fiber optic link.

[0160] The updated baseline parameters are:

[0161] [BER0*0.1+BER new *0.9,EH0*0.1+EH new *0.9,EW0*0.1+EW new *0.90]; where BER new Indicates the current bit error rate; EH new Indicates current eye level; EW new This indicates the current eye width.

[0162] The updated baseline parameters are then transmitted to the sender until the MQM change rate is less than the threshold.

[0163] To verify the effectiveness of the FPGA-based high-speed serial transceiver-based fiber optic interconnect transmission system provided in this embodiment, the quality of different systems applied to B-fiber communication links was compared. The quality comparison was conducted by comparing the transmission delay, jitter standard deviation, and BER of different systems, as shown in Table 4. System 1 is the FPGA-based high-speed serial transceiver-based fiber optic interconnect transmission system provided in this embodiment; System 2 is System 1 without considering the adjustment of pre-emphasis coefficients and receive equalization parameters; System 3 is System 1 without considering particle swarm optimization; System 4 is System 1 without considering dynamic adjustment of CDR bandwidth; and System 5 is System 1 without considering dynamic switching of FEC mode.

[0164] Table 4. Quality Comparison of Different Methods Applied to A Fiber Optic Communication Links

[0165] system Transmission delay BER jitter standard deviation System 1 40ns 5e-6 0.1 UI System 2 50ns 3e-5 0.3 UI System 3 55ns 8e-5 0.2 UI System 4 60ns 8e-5 0.3 UI System 5 50ns 1e-4 0.4 UI

[0166] As shown in Table 4, the FPGA-based high-speed serial transceiver provided in this embodiment for implementing an optical fiber interconnection transmission system has a certain degree of effectiveness.

[0167] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for implementing fiber optic interconnect transmission using a high-speed serial transceiver based on FPGA, characterized in that, include: S1. Obtain the bit error rate (BER) and eye diagram characteristic parameters of the received signal during optical fiber transmission. The eye diagram characteristic parameters include eye height (EH) and eye width (EW). Combine the BER and eye diagram characteristic parameters to establish a link characteristic baseline. The link characteristic baseline is... ;in Indicates the initial bit error rate; Indicates initial eye height; Indicates the initial eye width; The optical fiber transmission process includes photoelectric conversion, modulation and demodulation. The FPGA modulates the optical signal through an embedded high-speed serial transceiver and transmits the modulated optical signal to the optical fiber link. At the receiving end, the optical signal is recovered through a demodulation process. S2. Construct a multi-dimensional weighted quality indicator (MQM). The MQM is calculated by fusing a normalized scoring function that integrates BER, EH, and EW with weighting coefficients. The multi-dimensional weighted quality indicator is expressed as follows: ； in, This represents a multi-dimensional weighted quality indicator; This represents the bit error rate normalization scoring function; This represents the normalized scoring function for eye level. This represents the eye width normalized scoring function; , and Indicates the weighting coefficient; An improved particle swarm optimization algorithm based on MQM-driven FPGA is used to adjust the transmit pre-emphasis coefficient and receive equalization parameters of the high-speed serial transceiver. The improved particle swarm optimization algorithm includes the following optimization process: S2.

1. Initialize the parameter range of the pre-weighting coefficient and the receiving equalization parameter, and randomly generate an initial particle swarm, where the position of each particle represents a set of parameter combinations; S2.

2. Calculate the MQM for each particle under the current parameters; estimate the gradient of the MQM with respect to the parameters using the finite difference method; and update the particle's current velocity based on the gradient; S2.

3. Continue iterating; if the rate of change of MQM is less than the threshold, then convergence is determined and the optimal parameter combination is output; S3. Extract jitter characteristics of the received signal in the optical fiber link, including jitter standard deviation and jitter peak value. Based on the jitter characteristics and the current BER, dynamically adjust the clock data to restore the CDR bandwidth to balance jitter tracking capability and noise suppression performance. Utilize a nonlinear mapping rule to precisely control the recovery of the optical signal. The nonlinear mapping rule is implemented through a lookup table. The input of the lookup table is a joint quantization interval of the jitter indicator factor and the bit error rate (BER), and the output is a preset CDR bandwidth level. The joint quantization interval is used to jointly analyze the jitter interval and the BER interval. The jitter interval includes low jitter, medium jitter, and high jitter intervals; the BER interval includes low bit error rate, medium bit error rate, and high bit error rate intervals. The jitter indicator factor is obtained based on the jitter standard deviation and jitter peak value. The jitter indicator factor is: ； in, Indicates the jitter indicator factor; Indicates the standard deviation of jitter; Indicates the peak value of jitter; and This represents the influence coefficient of the jitter indicator factor; S4. Filter the BER data and calculate the trend slope. Combine the BER, trend slope, and preset hysteresis threshold logic to dynamically switch the forward error correction (FEC) mode to optimize signal transmission delay. The trend slope is: ； in, Indicates the slope of the trend; Indicates the number of sampling points within the time window; Represents a timestamp; express The filtered BER value at time 1; Hysteresis threshold logic is defined as follows: When the BER exceeds the threshold and the trend slope is greater than 0, switch from low error correction mode to high error correction mode; When the BER is less than the threshold and the trend slope is less than 0, switch from high error correction mode to low error correction mode. S5. Update baseline parameters in real time to form a closed loop and continuously optimize fiber optic link quality; the updated baseline parameters are... ; in Indicates the current bit error rate; Indicates current eye level; This indicates the current eye width.

2. A fiber optic interconnect transmission system implemented using a high-speed serial transceiver based on FPGA, characterized in that: The system includes: The link characteristic baseline construction module acquires the bit error rate (BER) and eye diagram characteristic parameters of the received signal during optical fiber transmission. The eye diagram characteristic parameters include eye height (EH) and eye width (EW). Combining the BER and eye diagram characteristic parameters, a link characteristic baseline is established. This link characteristic baseline is... ;in Indicates the initial bit error rate; Indicates initial eye height; The initial eye width is indicated; the optical fiber transmission process includes photoelectric conversion, modulation and demodulation. The FPGA modulates the optical signal through an embedded high-speed serial transceiver and transmits the modulated optical signal to the optical fiber link. At the receiving end, the optical signal is recovered through a demodulation process. The balanced optimization module constructs a multi-dimensional weighted quality index (MQM). MQM is calculated by fusing a normalized scoring function that integrates BER, EH, and EW with weighting coefficients. The multi-dimensional weighted quality index is expressed as follows: ； in, This represents a multi-dimensional weighted quality indicator; This represents the bit error rate normalization scoring function; This represents the normalized scoring function for eye level. This represents the eye width normalized scoring function; , and Indicates the weighting coefficient; An improved particle swarm optimization algorithm based on MQM-driven FPGA is used to adjust the transmit pre-emphasis coefficient and receive equalization parameters of the high-speed serial transceiver. The improved particle swarm optimization algorithm includes the following optimization process: S2.

1. Initialize the parameter range of the pre-weighting coefficient and the receiving equalization parameter, and randomly generate an initial particle swarm, where the position of each particle represents a set of parameter combinations; S2.

2. Calculate the MQM for each particle under the current parameters; estimate the gradient of the MQM with respect to the parameters using the finite difference method; and update the particle's current velocity based on the gradient; S2.

3. Continue iterating; if the rate of change of MQM is less than the threshold, then convergence is determined and the optimal parameter combination is output; The optical signal control module extracts jitter characteristics of the received signal in the optical fiber link, including jitter standard deviation and jitter peak value. Based on the jitter characteristics and the current bit error rate (BER), it dynamically adjusts the clock data to restore the CDR loop bandwidth, balancing jitter tracking capability and noise suppression performance. It utilizes a nonlinear mapping rule to precisely control the optical signal recovery. This nonlinear mapping rule is implemented using a lookup table. The input to the lookup table is a joint quantization interval of the jitter indicator factor and the bit error rate (BER), and the output is a preset CDR bandwidth level. The joint quantization interval is used to jointly analyze the jitter interval and the BER interval. The jitter interval includes low jitter, medium jitter, and high jitter intervals; the BER interval includes low BER, medium BER, and high BER intervals. The jitter indicator factor is obtained based on the jitter standard deviation and jitter peak value. The jitter indicator factor is: ； in, Indicates the jitter indicator factor; Indicates the standard deviation of jitter; Indicates the peak value of jitter; and This represents the influence coefficient of the jitter indicator factor; The signal transmission delay optimization module filters the BER data and calculates the trend slope. Combining the BER, trend slope, and a preset hysteresis threshold, it dynamically switches between FEC modes to optimize signal transmission delay. The trend slope is: ； in, Indicates the slope of the trend; Indicates the number of sampling points within the time window; Represents a timestamp; express The filtered BER value at time 1; Hysteresis threshold logic is defined as follows: When the BER exceeds the threshold and the trend slope is greater than 0, switch from low error correction mode to high error correction mode; When the BER is less than the threshold and the trend slope is less than 0, switch from high error correction mode to low error correction mode. The feedback module updates baseline parameters in real time, forming a closed loop for continuous fiber optic link quality tuning; the updated baseline parameters are... ; in Indicates the current bit error rate; Indicates current eye level; This indicates the current eye width.

Images