High-sensitivity qsfp28 optical module integrating soa+pin architecture and intelligent management and control

Abstract

The application discloses a high-sensitivity QSFP28 optical module fusing a SOA+PIN architecture and intelligent management and control, and particularly relates to the technical field of intelligent management and control of high-speed optical communication modules, which comprises the following steps: executing transmission signal execution emission processing, and sequentially executing optical amplification, photoelectric conversion, transimpedance amplification, clock recovery and error correction recovery on input received optical signals, and outputting transmission optical signals and received recovery signals; synchronously reading laser bias values, module temperature values, received optical power values, SOA control write-in values, clock recovery state values and error correction count values at fixed periods on the optical module side, and writing control instructions and outputting monitoring record tables; and executing maintenance cost generation, hidden instability identification and single-item limited release adjustment on the monitoring records in the continuous management and control period on the module side, so as to identify hidden instability states of internal compensation continuous consumption and output corresponding adjustment bases when the received recovery signals remain effective.

Description

Technical Field

[0001] This invention relates to the field of intelligent management and control technology for high-speed optical communication modules, and more specifically, to a high-sensitivity QSFP28 optical module that integrates SOA+PIN architecture and intelligent management and control. Background Technology

[0002] In long-distance high-speed optical modules for metropolitan area interconnection and data center interconnection, existing technologies mainly focus on achieving stable transmission at the 80km level within the QSFP28 package size. In engineering implementation, EML transmit link, SOA+PIN receive link, TIA, CDR and FEC error correction circuit are generally configured inside the module. Digital diagnostic interface is used to continuously read operating information such as laser bias, temperature and received optical power. Then, based on the current reading and bit error performance, relevant parameters are compensated, corrected or alarmed. Taking the continuous operation scenario of the 100GQSFP28ZR4 module deployed in the metropolitan area link or the data center interconnection link as an example, the module needs to maintain 80km of repeaterless transmission within the range of 0°C to 70°C. However, due to its compact size, limited power consumption, and autonomous management conditions on the module side, it cannot rely on external optical amplification equipment, nor is it suitable to maintain the link status through frequent manual intervention. Under these operating conditions, a common phenomenon occurs in the field: for a period of time before the link enters an anomaly, the FEC continues to correct errors, the CDR remains locked, and the received optical power does not fall below the alarm threshold. On the surface, the module can still transmit and receive normally, but internally it has already shown changes such as encrypted SOA compensation actions, repeated recovery adjustments, and a continuous increase in error correction burden with a slow fallback. This indicates that the current link is not truly stable, but is barely maintained after relying on the continuous consumption of internal compensation capabilities. The existing method of processing based only on instantaneous monitoring values ​​and over-limit results is difficult to identify this transitional state in time. Therefore, once temperature fluctuations, link disturbances, or device drift are superimposed, the module is prone to directly transition from superficial normality to actual instability. The technical problem to be solved by this application is: how to identify, in a high-sensitivity QSFP28 optical module that integrates SOA+PIN architecture and intelligent management, a hidden instability state in which the link surface is normal but the internal compensation capability has been continuously consumed in a timely manner at the module side. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a high-sensitivity QSFP28 optical module that integrates SOA+PIN architecture and intelligent management. By performing maintenance cost generation, hidden instability identification, and single-item restricted release adjustment on the monitoring records during the continuous management cycle at the module side, the hidden instability state of continuous internal compensation consumption can be identified and the corresponding adjustment basis can be output when the received recovery signal remains effective, thereby solving the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management, comprising: The transceiver module is used to perform transmission processing on the input transmitted electrical signal and sequentially perform optical amplification, photoelectric conversion, transimpedance amplification, clock recovery and error correction recovery on the input received optical signal, and output the transmitted optical signal and the received recovered signal. The edge control module is used to synchronously read the laser bias value, module temperature value, received optical power value, SOA control write value, clock recovery status value and error correction count value on the optical module side at a fixed period, and write control commands and output monitoring record table; The cost generation module is used to divide the monitoring record table into current record groups according to three consecutive cycles, count the number of SOA control writes, clock recovery relocks, error correction count increments, and recovery cycles, and output a maintenance cost table. The instability identification module is used to perform a line-by-line comparison between the current maintenance cost table and the previous maintenance cost table and the two previous maintenance cost tables when the received recovery signal is continuously valid. When the number of writes, the number of relocks, and the number of increments increase for two consecutive times and the number of recovery cycles is extended for two consecutive times, the hidden instability result is output. The edge publishing module is used to extract the first increased cost item from the current record group corresponding to the hidden instability result, determine the corresponding target control object, perform single-item restricted adjustment while keeping the other control objects unchanged, regenerate the maintenance cost table for the next current record group after adjustment, write the adjustment result of the corresponding cost item reduction into the historical record table, and output the adjustment basis for the next cycle call.

[0005] In a preferred embodiment, the transceiver module includes: The input transmitted electrical signal is processed for transmission shaping and modulation drive, and the modulated transmitted optical signal is written into the transmission record of the current transmission cycle, and the transmitted optical signal corresponding to the current transmission cycle is output. The input received optical signal is sequentially processed through SOA amplification, PIN photoelectric conversion, transimpedance amplification, clock recovery, and error correction recovery. The SOA control write changes, clock recovery relock changes, error correction count changes, and recovery completion cycle count in the corresponding current reception cycle are read and written into the current reception record. The reception recovery signal of the current reception cycle is then output. Based on the current received record, the recovery and retention check is performed. When the received recovery signal corresponding to the current receiving cycle is continuously valid, the SOA control write change precedes the clock recovery relock change and the error correction count change, and the number of recovery completed cycles is greater than the number of recovery completed cycles corresponding to the previous receiving cycle, the current received record is retained as the input for subsequent hidden instability identification.

[0006] In a preferred embodiment, the edge control module includes: Within the current control cycle, the laser bias value, module temperature value, received optical power value, SOA control write value, clock recovery status value, and error correction count value are read synchronously, and each read value is written into the current cycle record according to the same read time, and the original record group of the current control cycle is output. Based on the original record group of the current control cycle and the original record group of the previous control cycle, the same position difference is calculated and the change direction is determined to obtain the laser bias change value, module temperature change value, received optical power change value, SOA control change value, clock recovery change value and error correction count change value, and output the change record group of the current control cycle.

[0007] In a preferred embodiment, the edge control module further includes: The laser bias change value, module temperature change value, received optical power change value, SOA control change value, clock recovery change value, and error correction count change value in the change record group are read item by item according to the preset field order. The first change item with a non-zero change value in the field order is taken as the first change item. The target control object is determined according to the fixed mapping relationship between the first change item and the corresponding control object, and the target control record of the current control cycle is output. Based on the target control record, a single restricted control instruction is written to the target control object while keeping the written values ​​of the other control objects unchanged. Then, the original record group and the change record group of the next control cycle after writing are compared before and after execution. Under the condition that the change value corresponding to the first variable decreases and the other change values ​​are not added, the current control instruction is retained as the subsequent call record and the monitoring record table is output.

[0008] In a preferred embodiment, the cost generation module includes: The monitoring record table is divided into a current record group for every three consecutive control cycles according to time sequence. The SOA control write value, clock recovery status value, error correction count value and recovery completion cycle number in the current record group are expanded in the same position to form the control sequence, recovery sequence, error correction sequence and cycle sequence corresponding to the three control cycles, and the sequence set in the group is output. Based on the intra-group sequence set, construct an intra-group association graph consisting of write change edges, relock change edges, error correction increment edges, and recovery extension edges. For each edge in the intra-group association graph, calculate the edge occurrence order, edge duration span, and edge fallback span. Retain the association edges whose occurrence order is earlier and whose fallback span does not cover the occurrence time of the next edge as valid edges, and output the valid association set of the current record group.

[0009] In a preferred embodiment, the cost generation module further includes: Perform consistency checks on the effective association set, expand the write change edge, relock change edge, error correction increment edge, and recovery extension edge into candidate cost chains according to the same control object, count the number of control writes, clock recovery relocks, error correction increments, and recovery cycles for each candidate cost chain, and delete the candidate cost chains with reversed edge order, overlapping edge spans, and broken edge fallback relationships, and output the set of retained cost chains for the current record group; Based on the retained cost chain set, perform cost convergence calculation on the current record group, write the number of control writes, the number of clock recovery relocks, the number of error correction count increments, and the number of recovery cycles into the corresponding cost items in a fixed field order, merge the retained cost chains with identical cost items between two adjacent control cycles, stop expanding the current record group if there are no new valid edges, and output the maintenance cost table corresponding to the current record group.

[0010] In a preferred embodiment, the instability identification module includes: Under the condition that the recovery signal is continuously valid, perform the same-position expansion on the current maintenance cost table, the previous maintenance cost table and the previous two maintenance cost tables in the same field order, extract the SOA control write count, clock recovery relock count, error correction count increment count and recovery cycle count from each table, and output the three-cycle cost control group. Based on the three-cycle cost control group, cross-cycle difference calculations are performed on the number of SOA control writes, the number of clock recovery relocks, and the number of error correction count increments to obtain the first difference group of the current maintenance cost table relative to the previous maintenance cost table and the second difference group of the previous maintenance cost table relative to the previous two maintenance cost tables. Cross-cycle difference calculations with the same caliber are performed on the number of recovery cycles, and the number difference group and cycle difference group are output.

[0011] In a preferred embodiment, the instability identification module further includes: For the consistency check of execution direction of the number difference group and the cycle difference group, when the first difference and the second difference corresponding to the number of SOA control writes, the number of clock recovery relocks, and the number of error correction count increments are all greater than zero, and the first difference and the second difference corresponding to the number of recovery cycles are all greater than zero, a hidden instability candidate record corresponding to the current maintenance cost table is formed, and the hidden instability candidate set is output. Perform a corresponding verification on the monitoring record table corresponding to the hidden instability candidate set and the current reception cycle. If the reception recovery signal corresponding to the hidden instability candidate record remains continuously effective and the SOA control write value, clock recovery status value and error correction count value are consistent with the field correspondence of the hidden instability candidate record, retain the hidden instability candidate record as the hidden instability result output.

[0012] In a preferred embodiment, the edge publishing module includes: For the current record group, current maintenance cost table and historical record table corresponding to the hidden instability results, perform corresponding expansion according to the same field order, extract the current cost item, previous cost item and historical adjustment item corresponding to SOA control write count, clock recovery relock count, error correction count increment count and recovery cycle count, construct the release relationship chain from the current cost item to the previous cost item and from the previous cost item to the historical adjustment item, and delete the release relationship chain with inconsistent field order, inconsistent control object or inconsistent adjustment direction, and output the retained release chain set; Based on the retained release chain set, the restricted cost solution is performed on each release relationship chain. The difference between the current cost item and the previous cost item, the cost reduction amount corresponding to the historical adjustment item, and the number of maintenance periods after adjustment are written into the same chain cost group. For the chain cost group corresponding to the same target control object, the adjustment inheritance value and adjustment conflict value are calculated in a fixed field order. The release relationship chain with non-zero adjustment inheritance value and zero adjustment conflict value is retained as the candidate release chain, and the candidate release chain set is output.

[0013] In a preferred embodiment, the edge publishing module further includes: For each candidate release chain set, a forward trial release is performed. For each candidate release chain, only one adjustment value is written to its corresponding target control object while keeping the values ​​written to the other control objects unchanged. Then, the next current record group after the writing is input into the cost generation chain to obtain the trial maintenance cost table. The trial maintenance cost table is compared with the current maintenance cost table. The candidate release chains in which the cost items corresponding to the target control object are reduced, the cost items corresponding to the non-target control objects remain unchanged, and the received recovery signal is continuously effective are retained as the passing release chains. The passing release chain set is output. Based on the release convergence process performed through the release chain set, release chains with identical test results for the same target control object are merged into a formal release chain. The target control object, adjustment amount, cost term fallback result, and hold period number corresponding to the formal release chain are written into the historical record table. Release expansion stops when there are no new hidden instability results in the current record group or when the release chain set is empty, and the adjustment basis for the next cycle call is output.

[0014] The technical effects and advantages of this invention are as follows: 1. This scheme generates a maintenance cost table and identifies hidden instability results through three consecutive control cycles, so that the internal compensation consumption when the received recovery signal is still continuously effective can be characterized in advance, thereby relatively improving the problem that it is difficult to detect transient instability in time by relying solely on instantaneous readings; 2. The SOA control write, clock recovery relock, error correction count increment and recovery cycle number are sequentially associated and expanded according to the same field, so that the link maintenance cost has a unified calculation caliber, thereby relatively reducing the state judgment deviation caused by the dispersed existence of multiple source monitoring quantities; 3. Based on the first variable, the target control object is determined and only a single-item restricted control command is executed, so that the adjustment action is limited to the range of a single object, thereby relatively suppressing the cross disturbance and misadjustment propagation caused by simultaneous writing of multiple control objects; 4. By constructing a release relationship chain and combining it with historical adjustment items to solve for adjustment inheritance values ​​and adjustment conflict values, the current hidden instability results are made to correspond with the previous effective adjustment paths, thereby relatively improving the targeting and continuity of module-side adjustment releases; 5. Perform forward probing releases on candidate release chains and use the probing maintenance cost table to verify the adjustment results. This ensures that the release action is locally verified before being retained, thereby relatively reducing the possibility of link state deterioration caused by direct writing of invalid compensation. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the module structure of the present invention. Detailed Implementation

[0016] 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.

[0017] Refer to the instruction manual appendix Figure 1 The high-sensitivity QSFP28 optical module of the present invention, which integrates SOA+PIN architecture and intelligent management, includes: The transceiver module is used to perform transmission processing on the input transmitted electrical signal and sequentially perform optical amplification, photoelectric conversion, transimpedance amplification, clock recovery and error correction recovery on the input received optical signal, and output the transmitted optical signal and the received recovered signal. In this embodiment, the transceiver module performs same-cycle correspondence processing on the transmitting and receiving links, and generates basic records that can be directly invoked for subsequent edge control, cost generation, and hidden instability identification while completing photoelectric transmission and reception. Specifically, the transmitting and receiving cycles are used as the basic time stamps within the module. The transmitting cycle identifies the processing interval where a transmitted electrical signal enters the transmitting link and forms a transmitted optical signal, and the receiving cycle identifies the processing interval where a received optical signal enters the receiving link and forms a received recovery signal. When the transmitting and receiving sides are driven by the same master clock, adjacent transmitting and receiving cycles under the same master clock are numbered separately. When the transmitting and receiving sides do not share a clock, the unified time stamp output by the edge control module is written into the transmitting record and the current receiving record as a correspondence key to ensure that subsequent monitoring record tables, maintenance cost tables, and hidden instability results can all be performed with the same time stamp. This implementation process includes the following steps: First, to complete the transmission link processing and form a traceable transmission record, the input transmission electrical signal is fed into the transmission shaping circuit as the input quantity. Amplitude shaping, edge shaping, and modulation drive processing corresponding to the code pattern are performed on the input transmission electrical signal to obtain the modulation drive signal corresponding to the current transmission cycle. Then, the transmitting device outputs the transmission optical signal based on the modulation drive signal. Simultaneously with the output of the transmission optical signal, the current transmission cycle identifier, the input transmission electrical signal identifier, the modulation drive state, and the transmission optical signal state are extracted and written into the transmission record in a fixed field order. This fixed field order includes at least the transmission cycle identifier field, the input signal identifier field, the drive state field, and the transmission result field. If the input transmission electrical signal is missing in the current transmission cycle, the transmitting device does not output a transmission optical signal, or the modulation drive state is incomplete, the current transmission cycle identifier and the error reason code are written into the transmission record, and the transmission result field is set to invalid to prevent this transmission record from entering the subsequent normal transmission corresponding chain. Secondly, to complete the receiving link processing and form a calculable current receiving record, the input received optical signal is sequentially fed into the SOA amplification link, PIN optoelectronic conversion link, transimpedance amplification link, clock recovery link, and error correction recovery link. SOA amplification, PIN optoelectronic conversion, transimpedance amplification, clock recovery, and error correction recovery are sequentially performed on the input received optical signal, outputting the receiving recovery signal for the current receiving cycle. Specifically, the SOA control write change is determined by whether the difference between the SOA control write value corresponding to the current receiving cycle and the SOA control write value corresponding to the previous receiving cycle is zero. A difference of non-zero indicates the presence of SOA control write change, while a difference of zero indicates the absence of SOA control write change; clock recovery relock change is the number of times the clock recovery state changes from unlocked to relocked completed within the current reception cycle; error correction count change is the difference between the error correction count value at the end of the current reception cycle and the error correction count value at the beginning of the current reception cycle; recovery completion cycle count is the number of reception cycles spanned from the first occurrence of any of the SOA control write change, clock recovery relock change, or error correction count change in the current reception cycle until the reception recovery signal is restored to a continuously valid state; Then, the current reception cycle identifier, unified time stamp, SOA control write change, clock recovery relock change, error correction count change, recovery completion cycle count, and reception recovery signal status are written into the current reception record in a fixed field order. The fixed field order includes at least the reception cycle identifier field, unified time stamp field, SOA change field, relock change field, error correction change field, recovery cycle number field, and recovery result field. When the input received optical signal is missing, any link processing step fails to output a valid result, or the unified time stamp is missing, the current reception cycle identifier and the abnormal reason code are written into the current reception record, and the SOA control write change, clock recovery relock change, error correction count change, and recovery completion cycle number field are set to missing markers. This current reception record is not included in the normal hidden instability identification input. Furthermore, to filter out valid received records that can be used as inputs for subsequent hidden instability identification, the current received record and the corresponding current received record from the previous received cycle are read as input quantities, and a recovery and maintenance check is performed on the current received record. The criteria for determining the continuous validity of the received recovery signal are: the received recovery signal status field is valid within the current received cycle, and no unlocking flag, recovery interruption flag, or error correction recovery failure flag appears before the end of the current received cycle. The criteria for determining whether SOA control write changes precede clock recovery relock changes and error correction count changes are: the occurrence sequence code of each change item is recorded in the current received record; if the occurrence sequence code corresponding to the SOA control write change is less than the occurrence sequence code corresponding to the clock recovery relock change and less than the occurrence sequence code corresponding to the error correction count change, then the SOA control write change is determined to have occurred first. The criteria for determining whether the number of recovery completion cycles is greater than the number of recovery completion cycles corresponding to the previous reception cycle are as follows: the value of the recovery completion cycle number segment in the current reception record is greater than the value of the recovery completion cycle number segment in the previous reception record. When all three conditions are met, the current reception record is retained as the subsequent hidden instability identification input and written into the valid reception record area of ​​the monitoring record table. When any condition is not met, the current reception record is written into the non-retained reception record area of ​​the monitoring record table and a reason code for not retaining it is attached. The reason code for not retaining it includes at least three categories: invalid recovery signal, incorrect occurrence order, and recovery cycle not extended, so that the edge control module can directly distinguish between valid and invalid inputs when reading them later. Through the above processing, the transceiver module not only completes the basic optoelectronic transceiver of sending optical signals and receiving recovery signals, but also completes the structured writing of the sending record and the current receiving record. It also outputs the current receiving record that meets the recovery and maintenance verification conditions as the input for subsequent hidden instability identification, thereby providing a unified and calculable input basis for subsequent monitoring record table generation, maintenance cost table calculation and hidden instability result determination. In practical applications: Taking the continuous operation scenario of a 100GQSFP28 optical module in an 80km metropolitan area link as an example, the SOA control write value changes relative to the previous receive cycle within a certain receive cycle. Subsequently, a clock recovery relock change and an increase in the error correction count are recorded within the same receive cycle. The receive recovery signal recovers to a continuously valid state after crossing two receive cycles. At this time, the transceiver module writes the SOA control write change, clock recovery relock change, error correction count change, and number of recovery completed cycles corresponding to the receive cycle into the current receive record. If the current receive record simultaneously satisfies the following conditions: the receive recovery signal is continuously valid, the SOA control write change occurs before the clock recovery relock change and the error correction count change, and the number of recovery completed cycles is greater than the number of recovery completed cycles corresponding to the previous receive cycle, then the current receive record is written into the valid receive record area of ​​the monitoring record table for direct use by the subsequent instability identification module.

[0018] The edge control module is used to synchronously read the laser bias value, module temperature value, received optical power value, SOA control write value, clock recovery status value and error correction count value on the optical module side at a fixed period, and write control commands and output monitoring record table; In this embodiment, the edge control module is responsible for synchronously collecting, extracting changes, determining control objects, and writing individual records to the optical module side under a unified control timescale. This ensures that the monitoring record tables read by the subsequent cost generation module have a unified source, unified field order, and unified correspondence. To avoid period misalignment, field mismatch, and non-unique control objects in subsequent calculations, this embodiment first determines the control period as a unique control period unit within the edge control module. The current control period, the previous control period, and the next control period after writing are all determined by consecutive period numbers. The unified sampling time before the end of the current control period is taken at the same reading time. The unified sampling time is triggered by a timer within the module or determined by a periodic trigger signal issued by the host device and written to each record object as the corresponding key. This implementation process includes the following: First, to form the basic input required for subsequent change calculations, the laser bias register, temperature acquisition register, received optical power register, SOA control register, clock recovery status register, and error correction count register are read within the current control period to obtain the laser bias value, module temperature value, received optical power value, SOA control write value, clock recovery status value, and error correction count value. After reading, the control period identifier, unified sampling time, six read values, and read validity identifier are written into the current period record in a fixed field order. Then, all current period records within the same current control period are grouped into an original record group according to time sequence. The fixed field order adopts the sequence code given by the preset field order table and includes at least the period identifier field, sampling time field, laser bias field, module temperature field, received optical power field, SOA control field, clock recovery status field, error correction count field, and validity identifier field. If any register returns a null value, the check bit is abnormal, or the value exceeds the register code range, a missing flag is written in the corresponding field and an abnormal reason code is attached. This field is not used as a valid read value for this control period in subsequent change calculations. Subsequently, to convert discrete read values ​​into readable changes, the original record group of the current control cycle and the original record group of the previous control cycle are read, and corresponding positions are performed according to the rule of adjacent cycle numbers, the same field sequence code, and the same sampling time. The difference between the current value and the previous value is calculated for the laser bias value, module temperature value, received optical power value, SOA control write value, and error correction count value, and the difference is recorded as positive change if the difference is greater than zero, negative change if the difference is less than zero, and zero change if the difference is equal to zero. The clock recovery status value is differentially analyzed according to a fixed status code table. If the current status code is different from the previous status code, it is recorded as a clock recovery change value. If the status codes are the same, it is recorded as a zero change. After the difference calculation is completed, the laser bias change value, module temperature change value, received optical power change value, SOA control change value, clock recovery change value, error correction count change value, and their respective direction codes are written into the change record group, and the field sequence codes are kept the same as those in the original record group. If there is a missing marker in the corresponding field in the original record group of the current control cycle or the original record group of the previous control cycle, the change value corresponding to that field is written as an undecidable field, and the undecidable reason code is written into the change record group to prevent missing values ​​from being misjudged as valid changes. Furthermore, to ensure that only one control object is generated within the same control cycle, the edge control module performs a step-by-step reading of the laser bias change value, module temperature change value, received optical power change value, SOA control change value, clock recovery change value, and error correction count change value in the change record group according to a preset field sequence table. The first change item in the field sequence table with a non-zero change value and not belonging to an undecidable field is taken as the first change item. The preset field sequence table is written using the system factory configuration or deployment configuration, and the field sequence code is fixed as the laser bias field, module temperature field, received optical power field, SOA control field, clock recovery status field, and error correction count field. After the first variable is determined, the fixed mapping table is read, and the target control object is determined according to the one-to-one correspondence between the first variable and the control object. Among them, the laser bias change value corresponds to the laser bias control object, the module temperature change value corresponds to the temperature control control object, the received optical power change value corresponds to the receive link compensation control object, the SOA control change value corresponds to the SOA control object, the clock recovery change value corresponds to the clock recovery control object, and the error correction count change value corresponds to the error correction processing control object. After the determination is completed, the current control cycle identifier, the first variable field sequence code, the first variable change value, the first variable direction code, the target control object identifier, and the mapping code are written into the target control record. If all six change values ​​are zero or all are undecidable fields, an empty first variable mark and an empty target control object mark are written, and the current control cycle is recorded as a no-control cycle, and no control writing is performed. After the target control object is determined, to verify whether the control write should be retained as a subsequent call record, only a single-item restricted control instruction is written to the target control object, while keeping the current valid write values ​​of the other control objects unchanged. The single-item restricted control instruction means that within the current control cycle, the target control object is only allowed to change by one preset level relative to the previous value. The preset level is given by the register level table, which is written using a factory configuration table or a field configuration table. The code value difference between adjacent levels is fixed. After writing, the original record group for the next control cycle is read, and this original record group is compared with the change record group for the current control cycle according to the field sequence code. The absolute value of the change after writing is compared with the absolute value of the change before writing for the first change item. When the absolute value of the change after writing is less than the absolute value of the change before writing, it is determined that the change value corresponding to the first change item has decreased. Meanwhile, each of the remaining change values ​​is checked. If there are no new change values ​​that change from zero to positive or negative, it is determined that no new change values ​​have been added. Only when the change value corresponding to the first change value decreases and no new change values ​​have been added, the current single-item restricted control instruction is retained as a subsequent call record. The original record group, change record group, target control record, control write result, and subsequent call record are summarized and written into the monitoring record table in cyclical order. If the comparison between the previous and subsequent entries fails, the current control write is marked as an invalid call record, and an invalid reason code is written. The valid write value of the target control object before the write is restored in the next control cycle. Through the above processing, the edge control module transforms the dispersed operation volume on the optical module side into a monitoring record table with fixed field order, clear correspondence, and direct write-back and callability. This provides a unified input for the subsequent cost generation module and avoids the uncertainty caused by concurrent writing of multiple control objects, period misalignment, and field jumps. In practical applications: When the received optical power value decreases relative to the previous control cycle within a certain current control cycle, the clock recovery state value changes state while other read values ​​remain unchanged, the received optical power change value is calculated as a reverse change and the clock recovery change value is a state change after the same position difference is calculated. The edge control module first takes the received optical power change value as the first variable item according to the preset field sequence table, and then determines the receiving link compensation control object as the target control object according to the fixed mapping table, and writes the adjacent first-level control code only to the target control object. After the next control cycle after writing, if the absolute value of the received optical power change decreases and the laser bias change value, module temperature change value, SOA control change value, clock recovery change value, and error correction count change value are not newly added, then the control write is retained as a subsequent call record and written to the monitoring record table for direct call by the subsequent cost generation module.

[0019] The cost generation module is used to divide the monitoring record table into current record groups according to three consecutive cycles, count the number of SOA control writes, clock recovery relocks, error correction count increments, and recovery cycles, and output a maintenance cost table. In this embodiment, the cost generation module converts the monitoring record table output by the edge control module into a structured result that characterizes the maintenance cost of the receiving link. This allows the subsequent instability identification module to rely on the correlation between control writing, clock recovery, error correction changes, and recovery time within a continuous control cycle, rather than directly on a single monitoring value. To ensure that the maintenance cost table has a unified source and consistent definition, this embodiment uses the control cycle as the sole expansion cycle. Three control cycles arranged sequentially in the monitoring record table are defined as a current record group, and each field within the current record group is assigned to the same position according to the fixed field order table in the monitoring record table. The number of recovery completion cycles is directly taken from the corresponding field in the monitoring record table. In this embodiment, the number of recovery cycles uses the same field and definition as the number of recovery completion cycles, without setting a separate field. This implementation process includes the following: First, to form the intra-group sequence set required for subsequent edge construction calculations, records from every three consecutive control cycles are extracted from the monitoring record table in chronological order to form a current record group. The SOA control write values, clock recovery status values, error correction count values, and recovery completion cycle counts corresponding to the three control cycles within the current record group are used as inputs for in-place expansion. During in-place expansion, the control cycle identifier and field sequence code are used as dual mapping keys. SOA control write values ​​with the same field sequence code within the three control cycles are sequentially written into the control sequence, clock recovery status values ​​with the same field sequence code are sequentially written into the recovery sequence, and error correction count values ​​with the same field sequence code are sequentially written into the recovery sequence. Error counts are written sequentially into the error correction sequence, and recovery completion cycles with the same field sequence code are written sequentially into the cycle sequence, thus forming a sequence set within the current record group. If any control cycle in the monitoring record table is missing an SOA control write value, clock recovery status value, error correction count, or recovery completion cycle, a missing marker is written at the corresponding sequence position, and a sequence missing reason code is written into the current record group. The missing position is also prohibited from participating in the edge construction calculation of the current record group. If there are fewer than two control cycles of valid fields in three consecutive control cycles, the current record group is marked as an unexpandable record group and will not be included in the subsequent internal association graph construction. Subsequently, to transform the intra-group sequence set into an intra-group correlation graph that can be used to statistically analyze maintenance costs, the cost generation module reads the control sequence, recovery sequence, error correction sequence, and cycle sequence as inputs, and performs item-by-item comparison of the corresponding sequence values ​​between two adjacent control cycles. When the SOA control write value of the later control cycle is different from that of the earlier control cycle, a write change edge is established between the corresponding two control cycle nodes. When the clock recovery state value of the later control cycle changes state relative to the earlier control cycle, a relock change edge is established between the corresponding two control cycle nodes. When the error correction count value of the later control cycle is greater than that of the earlier control cycle, an error correction increment edge is established between the corresponding two control cycle nodes. When the number of recovery completion cycles of the later control cycle is greater than that of the earlier control cycle, a recovery extension edge is established between the corresponding two control cycle nodes. After edge construction, the edge occurrence order, edge duration span, and edge fallback span are calculated for each edge. The edge occurrence order is the control cycle sequence number corresponding to the edge's starting node. The edge duration span is the number of control cycles between the edge's starting and ending nodes. The edge fallback span is the number of control cycles between the control cycle in which the edge first returns to its previous value after its ending node and the end node itself. If the previous value is not returned in subsequent control cycles, the edge fallback span is recorded as a fixed extension bit after the last cycle of the current record group. The fixed extension bit is determined by rule constraints. 1. Then, perform filtering according to the order of edge occurrence from smallest to largest, retaining the associated edges whose occurrence order is earlier and whose fallback span does not cover the occurrence time of the next edge as valid edges. The occurrence time of the next edge whose fallback span does not cover it means that the period position obtained by adding the fallback span to the current edge's termination node is less than the period position corresponding to the starting node of the next edge. If two edges have the same starting node, they are retained according to the fixed edge sequence code, which is in the order of writing change edge, relocking change edge, error correction increment edge, and recovery extension edge. After filtering, output the valid association set of the current record group. Subsequently, to eliminate cross-interference between different controlled objects and obtain a cost chain that can be directly statistically analyzed, the cost generation module performs a consistency check on the effective association set. Specifically, it reads the control object identifier, edge type identifier, start cycle identifier, and end cycle identifier corresponding to each edge in the effective association set, and expands the write change edges, relock change edges, error correction increment edges, and recovery extension edges with the same control object identifier into the same candidate cost chain. After the candidate cost chain is formed, it counts the number of write change edges as the number of control writes, the number of relock change edges as the number of clock recovery relocks, the number of error correction increment edges as the number of error correction increments, and the sum of the cycle extension values ​​corresponding to the recovery extension edges as the number of recovery cycles. Then, each candidate cost chain is screened out, deleting those with reversed edge order, overlapping edge span, or broken edge fallback relationships. Reversed edge order means the starting period of the next edge is less than the starting period of the previous edge; overlapping edge span means the starting period of the next edge is not greater than the ending period of the previous edge; broken edge fallback relationships mean that after the previous edge terminates, there is no fallback record corresponding to the previous edge type before the end of the current record group, and the next edge has already started. For cases where the same controlled object forms multiple candidate cost chains, priority is given to retaining the candidate cost chain with more edges. If the number of edges is the same, priority is given to retaining the candidate cost chain with an earlier starting period position. If the starting period positions are the same, a unique candidate cost chain is retained according to the dictionary order of the fixed edge sequence code. After screening, the retained cost chain set for the current record group is output. If the effective association set is empty, an empty retained cost chain set is directly output, and a reason code for no valid chain is written to the current record group. Finally, to converge the set of retained cost chains into a maintenance cost table that can be directly called later, the cost generation module performs cost convergence calculation on the current record group based on the set of retained cost chains. Specifically, the control write count, clock recovery relock count, error correction count increment count, and recovery cycle count in each retained cost chain are written to the corresponding cost item in a fixed field order. The fixed field order adopts the field sequence code in the monitoring record table, which is the control write count field, relock count field, error correction count increment field, and recovery cycle number field in sequence. After the cost items are written, the retained cost chains with identical cost items between two adjacent control cycles are merged. The identical cost items mean that the values ​​of the control write count, clock recovery relock count, error correction count increment count, and recovery cycle count are all the same. When merging, the start cycle identifier of the previous retained cost chain and the end cycle identifier of the next retained cost chain are retained, and the merged cost chain is renumbered. Next, check if there are any new valid edges in the current record group. If there are no new valid edges relative to the retained cost chain set, stop the expansion and do not continue to connect to subsequent cycles. If there are new valid edges, expand once only under the same control object and adjacent cycle conditions, and repeat the consistency check and cost convergence calculation. When the stop expansion condition is met, write the current record group identifier, control object identifier, control write count, clock recovery relock count, error correction count increment count, and recovery cycle count into the maintenance cost table. At the same time, write the cost chain source identifier and merge identifier so that the instability identification module can perform a peer comparison in the same field order. If the retained cost chain set is empty, write the empty cost value and empty chain identifier into the maintenance cost table, and keep the current record group identifier unchanged to prevent misalignment in subsequent cycles. Through the above processing, the cost generation module transforms the discrete monitoring quantities in the monitoring record table into a maintenance cost table that corresponds one-to-one with the controlled object. This allows the correlation between SOA control writing, clock recovery relocking, error correction count increment, and recovery completion cycle number to be uniformly written into the same cost structure. This not only eliminates field misalignment and control object mixing, but also provides an input basis for subsequent hidden instability identification that can be directly compared in the same position for three consecutive cycles. In practical applications: When the same controlled object corresponding to three consecutive control cycles in the monitoring record table experiences one change in SOA control write value, one switch in clock recovery status value, two increases in error correction count value, and an increase in the number of recovery completion cycles from one to two, the cost generation module first forms the current record group according to the three control cycles, and then forms the control sequence, recovery sequence, error correction sequence, and cycle sequence through co-position expansion; then, write change edges, relock change edges, error correction increment edges, and recovery extension edges are established between adjacent cycle nodes, and effective association sets are screened according to the edge occurrence order, edge duration span, and edge fallback span; after consistency verification, the effective edges corresponding to the same controlled object are expanded into candidate cost chains, and the control write count is one, the clock recovery relock count is one, the error correction count increment count is two, and the number of recovery cycles is one; if there is no edge order reversal, edge span overlap, or edge fallback relationship break, the candidate cost chain is retained and written into the maintenance cost table in a fixed field order for subsequent instability identification module to read.

[0020] The instability identification module is used to perform a line-by-line comparison between the current maintenance cost table and the previous maintenance cost table and the two previous maintenance cost tables when the received recovery signal is continuously valid. When the number of writes, the number of relocks, and the number of increments increase for two consecutive times and the number of recovery cycles is extended for two consecutive times, the hidden instability result is output. In this embodiment, the instability identification module identifies, under the condition that the received recovery signal remains continuously effective, a hidden instability state where the surface recovery is effective but the internal maintenance cost continues to rise from the maintenance cost changes corresponding to three consecutive control cycles. The identification result is then checked against the monitoring record table of the current receiving cycle to avoid misjudging short-term fluctuations, invalid records, or field mismatches as hidden instability results. To ensure consistent identification criteria, this embodiment uses the three consecutive control cycles corresponding to the previous two maintenance cost tables, the previous maintenance cost table, and the current maintenance cost table as the sole identification window. All three maintenance cost tables originate from the cost generation module output and have the same field sequence code. Simultaneously, the determination result of the continuous effectiveness of the received recovery signal is directly read from the recovery result field in the monitoring record table and is not regenerated. This implementation process includes the following: First, to form a directly comparable continuous three-cycle cost input, the instability identification module, under the condition that the recovery signal is continuously valid, reads the current maintenance cost table, the previous maintenance cost table, and the previous two maintenance cost tables as input quantities, and performs in-situ expansion based on the same field sequence code, the same controlled object identifier, and the continuous control cycle identifier. During expansion, the number of SOA control writes, clock recovery relocks, error correction count increments, and recovery cycles with the same field sequence code and controlled object identifier in the three maintenance cost tables are extracted in sequence to form the current cycle item, the previous cycle item, and the previous two cycle items. Then, the three types of cycle items are written into the three-cycle cost control group in the same order. Among them, the recovery cycle number uses the recovery cycle number segment already written in the maintenance cost table and is not recalculated. If any of the three maintenance cost tables is missing the controlled object identifier, field sequence code, or any of the four cost items, the corresponding cost item of the controlled object is written with an incomparable mark, and the missing reason code is written in the three-cycle cost control group. The controlled object is not included in the subsequent difference calculation. Subsequently, to transform the three-cycle cost changes into a determinable incremental result, the instability identification module performs cross-cycle difference calculations on the number of SOA control writes, clock recovery relocks, and error correction count increments based on the three-cycle cost control group. Specifically, the first difference group is obtained by subtracting the corresponding value from the previous maintenance cost table from the corresponding value in the current maintenance cost table; the second difference group is obtained by subtracting the corresponding value from the previous two maintenance cost tables from the corresponding value in the previous maintenance cost table; and the recovery cycle number is calculated using the same caliber to obtain the first difference of the recovery cycle number. The first difference value and the second difference value are calculated. After the calculation is completed, the first difference value of the SOA control write count, the first difference value of the clock recovery relock count, and the first difference value of the error correction count increment count are written into the first difference value field of the count difference value group. The corresponding second difference value is written into the second difference value field of the count difference value group. The first difference value and the second difference value of the recovery cycle count are written into the cycle difference value group. If any subtracted item or subtracted item in the difference value calculation has an incomparable mark, the corresponding difference value is written into the incomparable mark. The control object identifier and field sequence code are retained in the difference value group so that the source can be traced during subsequent corresponding verification. Subsequently, to distinguish between general cost fluctuations and hidden instability candidates, the instability identification module performs a direction consistency check on the number difference group and the cycle difference group. Specifically, it reads the first and second differences corresponding to the number of SOA control writes, the first and second differences corresponding to the number of clock recovery relocks, the first and second differences corresponding to the number of error correction count increments, and the first and second differences corresponding to the number of recovery cycles. For each item, it determines whether the first difference is greater than zero and whether the second difference is greater than zero. When the first and second differences corresponding to the number of SOA control writes, the number of clock recovery relocks, and the number of error correction count increments are all greater than zero, and the first and second differences corresponding to the number of recovery cycles are all greater than zero, it is determined that the controlled object simultaneously satisfies the continuous increase in cost and the continuous extension of recovery time within three consecutive control cycles, thereby forming a hidden instability candidate record corresponding to the current maintenance cost table. The hidden instability candidate record should at least include the current control cycle identifier, the controlled object identifier, the field sequence code, the four cost values, the four first differences, the four second differences, and the candidate establishment identifier, and output the hidden instability candidate set; if the first or second difference of any of the three counts is not greater than zero, or the first or second difference of the recovery cycle count is not greater than zero, then no hidden instability candidate record will be formed, but a non-establishment reason code will be written into the hidden instability candidate set, where the non-establishment reason code includes at least three categories: the count is not continuously increasing, the recovery cycle is not continuously extending, and the difference cannot be determined; Finally, to converge hidden instability candidate records into publishable hidden instability results, the instability identification module performs a corresponding verification between the hidden instability candidate set and the monitoring record table corresponding to the current reception cycle. Specifically, it reads the current control cycle identifier, control object identifier, and field sequence code from the hidden instability candidate records, and retrieves the corresponding current reception record in the monitoring record table according to the unified time scale, control object identifier, and field sequence code. After retrieving the corresponding record, it first reads the reception recovery signal status field. If the reception recovery signal status field is continuously valid, it passes the recovery validity verification. Then, it reads the SOA control write value, clock recovery status value, and error correction count value, and compares these three with the control object identifier and cost source in the hidden instability candidate records. Fields and field sequence codes undergo consistency checks. The consistency check criteria are as follows: SOA control write values ​​correspond to SOA control objects, clock recovery status values ​​correspond to clock recovery control objects, and error correction count values ​​correspond to error correction processing control objects. Furthermore, the field sequence codes of these three values ​​in the monitoring record table must match the field sequence codes in the hidden instability candidate records. Only when the received recovery signal remains continuously valid and the field correspondence is consistent, the hidden instability candidate record is retained as a hidden instability result output, and the hidden instability result is written to the result table. If the received recovery signal is invalid, the monitoring record table does not contain a corresponding record, or the field correspondence is inconsistent, the hidden instability candidate record is deleted, and the deletion reason code is written to the invalid candidate area of ​​the result table. Through the above processing, the instability identification module takes the maintenance cost table corresponding to three consecutive control cycles as input. Under the premise that the received recovery signal remains continuously effective, it first completes the three-cycle co-position expansion, and then completes the cross-cycle difference calculation, direction consistency verification and monitoring record table corresponding verification. In this way, the state where the reception is still effective but the internal maintenance cost continues to rise is separated from the general cost fluctuation, forming a hidden instability result that can be directly called by the edge publishing module. In practical applications: when the number of SOA control writes to the same controlled object in the first two maintenance cost tables, the previous maintenance cost table, and the current maintenance cost table are one, two, and three respectively; the number of clock recovery relocks is zero, one, and two respectively; the number of error correction count increments is one, two, and three respectively; and the number of recovery cycles is one, two, and three respectively, the instability identification module performs a co-position expansion on the three maintenance cost tables. The first difference of the number of SOA control writes, the number of clock recovery relocks, the number of error correction count increments, and the number of recovery cycles is one, and the second difference is one. After the direction consistency check is passed, a hidden instability candidate record is formed. If the receive recovery signal status field corresponding to the current receiving cycle in the monitoring record table is still continuously valid, and the SOA control write value, clock recovery status value, and error correction count value are consistent with the field sequence code and controlled object identifier of the hidden instability candidate record, then the hidden instability candidate record is retained as a hidden instability result output.

[0021] The edge publishing module is used to extract the first increased cost item from the current record group corresponding to the hidden instability result, determine the corresponding target control object, perform single-item restricted adjustment while keeping the other control objects unchanged, regenerate the maintenance cost table for the next current record group after adjustment, write the adjustment result of the corresponding cost item reduction into the history table, and output the adjustment basis for the next cycle call. In this implementation, the edge release module transforms hidden instability results into restricted, verifiable, and inheritable adjustment release results. This prevents adjustment actions from being directly issued based solely on the current hidden instability result. Instead, it establishes a continuous relationship between the current record group, the current maintenance cost table, and the historical record table. It first filters out inheritable and conflict-free release relationship chains, then performs single-item trial writes and verification convergence, ultimately forming an adjustment basis that can be directly invoked in the next cycle. To avoid field misalignment, mixed control objects, incorrect inheritance of historical adjustments, and concurrent writes to multiple objects during the release process, this implementation uses the current record group identifier, control object identifier, and field sequence code as the corresponding keys throughout the release process. The current cost item, the previous cost item, and historical adjustment items are all expanded according to the same field order. Forward trial releases only allow one target control object to change within one adjustment level, while the remaining control objects retain their current valid write values. This implementation process includes the following: First, to form a comparable link required for subsequent deployment solutions, the edge deployment module reads the current record group, current maintenance cost table, and historical record table corresponding to the hidden instability results as input, and performs corresponding expansion according to the same field sequence code, the same control object identifier, and the relationship between adjacent cycles. During expansion, the current cost item corresponding to the number of SOA control writes, clock recovery relocks, error correction count increments, and recovery cycles is extracted from the current maintenance cost table; the previous cost item corresponding to the same field is extracted from the maintenance cost table of the previous cycle; and the historical adjustment item corresponding to the current control object and the same field sequence code is extracted from the historical record table. After extraction, a release relationship chain is constructed, pointing from the current cost item to the previous cost item, and from the previous cost item to the historical adjustment item. The current cost item must contain at least the current record group identifier, control object identifier, field sequence code, and four cost values. The previous cost item must contain at least the previous record group identifier, control object identifier, field sequence code, and four cost values. The historical adjustment item must contain at least the historical adjustment cycle identifier, control object identifier, field sequence code, historical adjustment amount, cost reduction result, and maintenance cycle number. Subsequently, the release relationship chain is filtered out, deleting release relationship chains with inconsistent field sequence codes, inconsistent control object identifiers, or inconsistent adjustment directions. The adjustment direction is determined by the correspondence between the adjustment amount direction code in the historical adjustment item and the difference direction code of the current cost item relative to the previous cost item. After filtering, a set of retained release chains is output. If any object in the current maintenance cost table, the previous maintenance cost table, or the historical record table is missing a corresponding field, the corresponding release relationship chain is marked as unconstructable and a missing reason code is appended; it is not included in the set of retained release chains. Subsequently, to establish executable release candidates between historical inheritance and current hidden instability, the edge release module performs constrained cost solving on each release relationship chain based on the reserved release chain set. Specifically, for each reserved release chain, the difference between the current cost item and the previous cost item is calculated. The difference includes the difference in the number of control writes, the difference in the number of relocks, the difference in the number of error correction increments, and the difference in the number of recovery cycles. Then, the cost fallback amount and the number of retention cycles after adjustment are extracted from the historical adjustment items. The cost fallback amount is the difference between the pre-release value and the post-release value of the corresponding cost item after historical adjustment, and the number of retention cycles after adjustment is the number of cycles in which the cost item after historical adjustment remains unchanged. Then, the four differences, the cost fallback amount, and the number of retention cycles after adjustment are written into the same chain cost group, and constrained solving is performed in a fixed field order, where the fixed field order is the control write count field, the relock count field, the error correction increment count field, and the recovery cycle number field. Based on this, the adjustment inheritance value and adjustment conflict value are calculated for the chain cost group corresponding to the same target control object. The adjustment inheritance value is the number of cost items whose adjustment direction code of the historical adjustment item matches the difference direction code of the current cost item. The adjustment conflict value is the number of cost items whose adjustment direction code of the historical adjustment item is opposite to the difference direction code of the current cost item. When the adjustment inheritance value is not zero and the adjustment conflict value is zero, it is determined that the release relationship chain can be inherited by the historical adjustment path, and thus it is retained as a candidate release chain and the candidate release chain set is output. If the adjustment inheritance value is zero or the adjustment conflict value is not zero, the corresponding release relationship chain is deleted and the deletion reason code is written. If the same target control object forms multiple candidate release chains, the candidate release chain with the larger adjustment inheritance value is retained first. If the adjustment inheritance values ​​are the same, the candidate release chain with the longer maintenance period after adjustment is retained first. If they are still the same, the one with the closer historical adjustment period identifier is retained uniquely. Subsequently, to verify whether the adjustment actions corresponding to the candidate release chain can actually take effect in the current hidden instability state, the edge release module performs a forward trial release for the candidate release chain set. Specifically, for each candidate release chain, only one adjustment amount is written to its corresponding target control object, while keeping the written values ​​of the other control objects unchanged. One adjustment amount is given by the register level table corresponding to the target control object. The register level table is written using the factory configuration table or the field configuration table, and the code value difference between adjacent levels is fixed. After the writing is completed, the next current record group after the writing is read, and the next current record group is input into the cost generation module to generate the trial maintenance cost table corresponding to the next current record group. Then, the trial maintenance cost table and the current maintenance cost table are compared bitwise according to the field sequence code and the control object identifier. The difference before and after the trial is calculated item by item for the cost item corresponding to the target control object. If at least one of the control writing count, clock recovery relock count, error correction count increment count, and recovery cycle count corresponding to the target control object decreases and the other items do not increase, it is determined that the cost item corresponding to the target control object has decreased. Simultaneously, each cost item corresponding to the non-target control object is checked. If all four cost items of the non-target control object are the same as the corresponding values ​​in the current maintenance cost table, it is determined that the cost item corresponding to the non-target control object remains unchanged. Then, the receiving recovery signal status field corresponding to the next current record group after being written in the monitoring record table is read. If the status field is continuously valid, it is determined that the receiving recovery signal remains valid. Only when the cost item corresponding to the target control object decreases, the cost item corresponding to the non-target control object remains unchanged, and the receiving recovery signal remains valid, is the candidate release chain retained as the passing release chain and the passing release chain set output. If any condition is not met, the corresponding candidate release chain is deleted and an invalid reason code is written in the trial result table, while the valid write value before the target control object was written is restored. Finally, to converge the results of a single trial into a formal release result that can be used for a long time, the edge release module performs release convergence processing based on the release chain set. Specifically, the release chains corresponding to the same target control object are compared in chronological order. If the control object identifier, field sequence code, adjustment amount, cost term fallback result, and maintenance period number field are all the same in two consecutive trial results, then the two consecutive release chains are merged into a formal release chain. During the merging, the starting trial period identifier of the previous release chain and the ending trial period identifier of the next release chain are retained, and a new formal release chain identifier is generated. After the formal release chain is generated, its corresponding target control object, adjustment amount, cost term fallback result, and hold period number are written to the historical record table. At the same time, the formal release identifier and call priority identifier are written for direct call in the next period. Then, it checks whether there are any new hidden instability results in the current record group and whether the release chain set is empty. If there are no new hidden instability results in the current record group or the release chain set is empty, the release expansion stops, and no further trial writing is performed in subsequent periods. The current formal release chain or empty release result is written to the adjustment basis table, and the adjustment basis for the call in the next period is output. If there are still new hidden instability results and the release chain set is not empty, then only the control object corresponding to the new hidden instability result is subjected to one round of forward trial release to avoid repeated expansion of the control object that has already formed a formal release chain. Through the above processing, the edge release module first establishes a release relationship chain between the hidden instability result, the current maintenance cost table, and the historical record table. Then, it filters out inheritable and conflict-free candidate release chains by solving the chain cost group. It verifies the actual release effect by writing a single-item restricted trial and verification with the next current record group. Finally, it converges the release chain that passes through the release chain twice in a row into a formal release chain and writes it into the historical record table. This makes the adjustment basis both from the current hidden instability state and retains the inheritance relationship of the historical adjustment results, while avoiding concurrent adjustment of multiple control objects and repeated unfolding of invalid releases. In practical applications: When the hidden instability result corresponding to a certain current record group indicates a continuous increase in the number of control writes, relocks, and error correction increments of the receiving link compensation control object, and there are historical adjustment items in the historical record table with the same control object, the same field sequence code, and which have previously produced cost fallback results, the edge publishing module first constructs a publishing relationship chain composed of the current cost item, the previous cost item, and the historical adjustment item. If the publishing relationship chain satisfies the requirements of consistent field order, consistent control object, and consistent adjustment direction, it enters the chain cost group solution. When the adjustment inheritance value is not zero and the adjustment conflict value is zero, the publishing relationship chain is retained as a candidate publishing chain. Subsequently, the edge publishing module only writes one adjustment amount to the receiving link compensation control object. After the next current record group generates a trial maintenance cost table, if the cost item corresponding to the receiving link compensation control object decreases, the cost items corresponding to other control objects remain unchanged, and the receiving recovery signal remains valid, then the candidate publishing chain becomes a passed publishing chain. If two consecutive trials yield a passed publishing chain with identical fields, it is merged into a formal publishing chain, and the corresponding adjustment amount, cost item fallback result, and maintenance cycle number are written into the historical record table as the adjustment basis for direct invocation in the next cycle.

[0022] Working Principle: This scheme first uses a transceiver module to process transmitted electrical signals and amplify, convert, recover, and correct received optical signals. It records SOA control changes, clock recovery changes, error correction changes, and recovery time for each reception cycle. The edge control module then synchronously reads laser bias, module temperature, received optical power, SOA control write values, clock recovery status values, and error correction count values ​​according to a fixed control cycle, forming a monitoring record table. Based on changes between adjacent cycles, it determines the control object to be monitored. The cost generation module then compiles the monitoring results from three consecutive control cycles into a maintenance cost table, representing the control cost incurred by the link to maintain normal reception. The instability identification module compares the maintenance cost changes over three consecutive cycles, identifying hidden instability states where the surface appears normal but the internal cost is continuously increasing, provided the received recovery signal remains effective. Finally, the edge publishing module, combining historical adjustment records, tentatively performs single-item restricted adjustments on the target control object. Based on whether the adjusted cost decreases, whether other objects remain stable, and whether the received recovery signal remains effective, it decides whether to retain the adjustment result, thus forming the adjustment basis that can be directly invoked in the next cycle. For example, in an 80km metropolitan area link, the optical module may appear to be receiving data normally. However, as the link weakens, temperature changes, or device drift occurs, SOA compensation actions increase, clock recovery relocking times rise, error correction counts increase, and recovery time becomes longer. This indicates that although the link has not yet dropped, it is internally struggling to stay connected. In this situation, our solution does not only look at whether a single monitoring value exceeds the limit, but first analyzes the changes and costs over several consecutive cycles. Once it confirms that this is a hidden instability, it only performs a limited trial adjustment on the corresponding target control object. If the cost of the object decreases after adjustment, and other objects are not affected, and the received recovery signal remains effective, the adjustment is retained for future use. If the adjustment does not improve the situation or causes fluctuations in other objects, the adjustment is canceled to avoid misadjustment. In this way, the module can detect problems before they become truly unstable and maintain the link in a stable state with small local adjustments.

[0023] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management, characterized in that: include: The transceiver module is used to perform transmission processing on the input transmitted electrical signal and sequentially perform optical amplification, photoelectric conversion, transimpedance amplification, clock recovery and error correction recovery on the input received optical signal, and output the transmitted optical signal and the received recovered signal. The edge control module is used to synchronously read the laser bias value, module temperature value, received optical power value, SOA control write value, clock recovery status value and error correction count value on the optical module side at a fixed period, and write control commands and output monitoring record table; The cost generation module is used to divide the monitoring record table into current record groups according to three consecutive cycles, count the number of SOA control writes, clock recovery relocks, error correction count increments, and recovery cycles, and output a maintenance cost table. The instability identification module is used to perform a line-by-line comparison between the current maintenance cost table and the previous maintenance cost table and the two previous maintenance cost tables when the received recovery signal is continuously valid. When the number of writes, the number of relocks, and the number of increments increase for two consecutive times and the number of recovery cycles is extended for two consecutive times, the hidden instability result is output. The edge publishing module is used to extract the first increased cost item from the current record group corresponding to the hidden instability result, determine the corresponding target control object, perform single-item restricted adjustment while keeping the other control objects unchanged, regenerate the maintenance cost table for the next current record group after adjustment, write the adjustment result of the corresponding cost item reduction into the historical record table, and output the adjustment basis for the next cycle call.

2. The high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management as described in claim 1, characterized in that: The transceiver module includes: The input transmitted electrical signal is processed for transmission shaping and modulation drive, and the modulated transmitted optical signal is written into the transmission record of the current transmission cycle, and the transmitted optical signal corresponding to the current transmission cycle is output. The input received optical signal is sequentially processed through SOA amplification, PIN photoelectric conversion, transimpedance amplification, clock recovery, and error correction recovery. The SOA control write changes, clock recovery relock changes, error correction count changes, and recovery completion cycle count in the corresponding current reception cycle are read and written into the current reception record. The reception recovery signal of the current reception cycle is then output. Based on the current received record, the recovery and retention check is performed. When the received recovery signal corresponding to the current receiving cycle is continuously valid, the SOA control write change precedes the clock recovery relock change and the error correction count change, and the number of recovery completed cycles is greater than the number of recovery completed cycles corresponding to the previous receiving cycle, the current received record is retained as the input for subsequent hidden instability identification.

3. The high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management as described in claim 2, characterized in that: The edge control module includes: Within the current control cycle, the laser bias value, module temperature value, received optical power value, SOA control write value, clock recovery status value, and error correction count value are read synchronously, and each read value is written into the current cycle record according to the same read time, and the original record group of the current control cycle is output. Based on the original record group of the current control cycle and the original record group of the previous control cycle, the same position difference is calculated and the change direction is determined to obtain the laser bias change value, module temperature change value, received optical power change value, SOA control change value, clock recovery change value and error correction count change value, and output the change record group of the current control cycle.

4. The high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management as described in claim 3, characterized in that: The edge control module also includes: The laser bias change value, module temperature change value, received optical power change value, SOA control change value, clock recovery change value, and error correction count change value in the change record group are read item by item according to the preset field order. The first change item with a non-zero change value in the field order is taken as the first change item. The target control object is determined according to the fixed mapping relationship between the first change item and the corresponding control object, and the target control record of the current control cycle is output. Based on the target control record, a single restricted control instruction is written to the target control object while keeping the written values ​​of the other control objects unchanged. Then, the original record group and the change record group of the next control cycle after writing are compared before and after execution. Under the condition that the change value corresponding to the first variable decreases and the other change values ​​are not added, the current control instruction is retained as the subsequent call record and the monitoring record table is output.

5. The high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management as described in claim 4, characterized in that: The cost generation module includes: The monitoring record table is divided into a current record group for every three consecutive control cycles according to time sequence. The SOA control write value, clock recovery status value, error correction count value and recovery completion cycle number in the current record group are expanded in the same position to form the control sequence, recovery sequence, error correction sequence and cycle sequence corresponding to the three control cycles, and the sequence set in the group is output. Based on the intra-group sequence set, construct an intra-group association graph consisting of write change edges, relock change edges, error correction increment edges, and recovery extension edges. For each edge in the intra-group association graph, calculate the edge occurrence order, edge duration span, and edge fallback span. Retain the association edges whose occurrence order is earlier and whose fallback span does not cover the occurrence time of the next edge as valid edges, and output the valid association set of the current record group.

6. The high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management as described in claim 5, characterized in that: The cost generation module further includes: Perform consistency checks on the effective association set, expand the write change edge, relock change edge, error correction increment edge, and recovery extension edge into candidate cost chains according to the same control object, count the number of control writes, clock recovery relocks, error correction increments, and recovery cycles for each candidate cost chain, and delete the candidate cost chains with reversed edge order, overlapping edge spans, and broken edge fallback relationships, and output the set of retained cost chains for the current record group; Based on the retained cost chain set, perform cost convergence calculation on the current record group, write the number of control writes, the number of clock recovery relocks, the number of error correction count increments, and the number of recovery cycles into the corresponding cost items in a fixed field order, merge the retained cost chains with identical cost items between two adjacent control cycles, stop expanding the current record group if there are no new valid edges, and output the maintenance cost table corresponding to the current record group.

7. The high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management as described in claim 6, characterized in that: The instability identification module includes: Under the condition that the recovery signal is continuously valid, perform the same-position expansion on the current maintenance cost table, the previous maintenance cost table and the previous two maintenance cost tables in the same field order, extract the SOA control write count, clock recovery relock count, error correction count increment count and recovery cycle count from each table, and output the three-cycle cost control group. Based on the three-cycle cost control group, cross-cycle difference calculations are performed on the number of SOA control writes, the number of clock recovery relocks, and the number of error correction count increments to obtain the first difference group of the current maintenance cost table relative to the previous maintenance cost table and the second difference group of the previous maintenance cost table relative to the previous two maintenance cost tables. Cross-cycle difference calculations with the same caliber are performed on the number of recovery cycles, and the number difference group and cycle difference group are output.

8. The high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management according to claim 7, characterized in that: The instability identification module also includes: For the consistency check of execution direction of the number difference group and the cycle difference group, when the first difference and the second difference corresponding to the number of SOA control writes, the number of clock recovery relocks, and the number of error correction count increments are all greater than zero, and the first difference and the second difference corresponding to the number of recovery cycles are all greater than zero, a hidden instability candidate record corresponding to the current maintenance cost table is formed, and the hidden instability candidate set is output. Perform a corresponding verification on the monitoring record table corresponding to the hidden instability candidate set and the current reception cycle. If the reception recovery signal corresponding to the hidden instability candidate record remains continuously effective and the SOA control write value, clock recovery status value and error correction count value are consistent with the field correspondence of the hidden instability candidate record, retain the hidden instability candidate record as the hidden instability result output.

9. The high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management as described in claim 8, characterized in that: The edge publishing module includes: For the current record group, current maintenance cost table and historical record table corresponding to the hidden instability results, perform corresponding expansion according to the same field order, extract the current cost item, previous cost item and historical adjustment item corresponding to SOA control write count, clock recovery relock count, error correction count increment count and recovery cycle count, construct the release relationship chain from the current cost item to the previous cost item and from the previous cost item to the historical adjustment item, and delete the release relationship chain with inconsistent field order, inconsistent control object or inconsistent adjustment direction, and output the retained release chain set; Based on the retained release chain set, the restricted cost solution is performed on each release relationship chain. The difference between the current cost item and the previous cost item, the cost reduction amount corresponding to the historical adjustment item, and the number of maintenance periods after adjustment are written into the same chain cost group. For the chain cost group corresponding to the same target control object, the adjustment inheritance value and adjustment conflict value are calculated in a fixed field order. The release relationship chain with non-zero adjustment inheritance value and zero adjustment conflict value is retained as the candidate release chain, and the candidate release chain set is output.

10. The high-sensitivity QSFP28 optical module integrating SOA+PIN architecture and intelligent management as described in claim 9, characterized in that: The edge publishing module also includes: For each candidate release chain set, a forward trial release is performed. For each candidate release chain, only one adjustment value is written to its corresponding target control object while keeping the values ​​written to the other control objects unchanged. Then, the next current record group after the writing is input into the cost generation chain to obtain the trial maintenance cost table. The trial maintenance cost table is compared with the current maintenance cost table. The candidate release chains in which the cost items corresponding to the target control object are reduced, the cost items corresponding to the non-target control objects remain unchanged, and the received recovery signal is continuously effective are retained as the passing release chains. The passing release chain set is output. Based on the release convergence process performed through the release chain set, release chains with identical test results for the same target control object are merged into a formal release chain. The target control object, adjustment amount, cost term fallback result, and hold period number corresponding to the formal release chain are written into the historical record table. Release expansion stops when there are no new hidden instability results in the current record group or when the release chain set is empty, and the adjustment basis for the next cycle call is output.

Images