CN118075207B - FPGA-based ultra-high definition JPEG-XS IP data packet processing method, device and system - Google Patents
FPGA-based ultra-high definition JPEG-XS IP data packet processing method, device and system Download PDFInfo
- Publication number
- CN118075207B CN118075207B CN202410465671.9A CN202410465671A CN118075207B CN 118075207 B CN118075207 B CN 118075207B CN 202410465671 A CN202410465671 A CN 202410465671A CN 118075207 B CN118075207 B CN 118075207B
- Authority
- CN
- China
- Prior art keywords
- flow control
- data packet
- data
- model
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 13
- 238000012545 processing Methods 0.000 claims abstract description 107
- 238000011217 control strategy Methods 0.000 claims abstract description 98
- 238000000034 method Methods 0.000 claims abstract description 77
- 238000012549 training Methods 0.000 claims abstract description 56
- 230000005540 biological transmission Effects 0.000 claims abstract description 49
- 238000003066 decision tree Methods 0.000 claims description 55
- 238000013145 classification model Methods 0.000 claims description 49
- 230000002159 abnormal effect Effects 0.000 claims description 47
- 230000006870 function Effects 0.000 claims description 27
- 238000004364 calculation method Methods 0.000 claims description 20
- 230000008569 process Effects 0.000 claims description 18
- 238000012360 testing method Methods 0.000 claims description 18
- 238000011156 evaluation Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 12
- 238000003860 storage Methods 0.000 claims description 12
- 238000009826 distribution Methods 0.000 claims description 10
- 238000012544 monitoring process Methods 0.000 claims description 10
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000012216 screening Methods 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 6
- 238000005192 partition Methods 0.000 claims description 6
- 125000004122 cyclic group Chemical group 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 230000000007 visual effect Effects 0.000 abstract description 7
- 238000010586 diagram Methods 0.000 description 8
- 238000012986 modification Methods 0.000 description 8
- 230000004048 modification Effects 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 5
- 238000000605 extraction Methods 0.000 description 5
- 230000005856 abnormality Effects 0.000 description 4
- 230000004075 alteration Effects 0.000 description 4
- 238000001914 filtration Methods 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000012163 sequencing technique Methods 0.000 description 3
- 238000012790 confirmation Methods 0.000 description 2
- 230000003111 delayed effect Effects 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- HRULVFRXEOZUMJ-UHFFFAOYSA-K potassium;disodium;2-(4-chloro-2-methylphenoxy)propanoate;methyl-dioxido-oxo-$l^{5}-arsane Chemical compound [Na+].[Na+].[K+].C[As]([O-])([O-])=O.[O-]C(=O)C(C)OC1=CC=C(Cl)C=C1C HRULVFRXEOZUMJ-UHFFFAOYSA-K 0.000 description 2
- 238000011425 standardization method Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The embodiment of the application provides an IP data packet processing method, device and system of ultra-high definition JPEG-XS based on FPGA. The method comprises the following steps: decapsulating each preprocessed first IP data packet to obtain each second IP data packet; calculating based on each second IP data packet to obtain a flow control reference coefficient; performing model training on the initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model; performing policy matching by adopting a target flow control policy matching model based on each second IP data packet to obtain a target flow control policy; and distributing and processing each second IP data packet by adopting a target flow control strategy to obtain a processing result. The application can optimize the utilization of the ultra-high definition signal network transmission resources with 8K and 4K visual lossless quality, avoid network congestion and reduce transmission delay.
Description
Technical Field
The application relates to the technical field of IP signal processing of 8K and 4K ultra-high definition JPEG-XS, in particular to an FPGA-based method, an FPGA-based device, an FPGA-based storage medium, an FPGA-based electronic device and an FPGA-based system for processing an IP data packet of the ultra-high definition JPEG-XS.
Background
At present, with the development of ultra-high definition video technology, the demands for decapsulation and data output processing of 8K and 4K ultra-high definition visual lossless quality JPEG-XS IP data packets are also increasing. In the prior art, when processing JPEG-XS IP data packets, the waste of flow bandwidth is large, in order to avoid the situations of JPEG-XS data packet transmission errors, data sequence disorder or data loss caused by large flow when processing a large number of data packets, or network data mutation, more than one time of network bandwidth needs to be reserved for data redundancy and mirroring, so that the data transmission cost is increased, the accuracy and the flexibility of flow control are lower, the processing efficiency and the receiving capability of 8K and 4K ultra-high definition JPEG-XS IP data packets are influenced, and the ultra-high definition JPEG-XS IP data transmission delay is increased.
The above information disclosed in the background section is only for enhancement of understanding of the background of the application and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art.
Disclosure of Invention
The embodiment of the application provides an ultrahigh-definition JPEG-XS IP data packet processing method, device, storage medium, electronic equipment and system based on an FPGA (field programmable gate array), which are used for solving the technical problems of high cost and low efficiency in the processing process of the ultrahigh-definition JPEG-XS IP data packet in the prior art 8K and 4K.
The embodiment of the application provides an IP data packet processing method of ultra-high definition JPEG-XS based on FPGA, which comprises the following steps:
Decapsulating each first IP data packet of the 8K4K ultra-high definition JPEG-XS obtained through pretreatment to obtain second IP data packets of the 8K and 4K ultra-high definition JPEG-XS corresponding to each first IP data packet;
Calculating based on each second IP data packet in the second IP data packet stream to obtain a flow control reference coefficient corresponding to the second IP data packet stream;
performing model training on the initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model;
Performing policy matching by using the target flow control policy matching model based on each second IP data packet to obtain a target flow control policy for each second IP data packet;
And distributing the second IP data packets by adopting the target flow control strategy to obtain an IP data packet processing result of 8K4K ultra-high definition JPEG-XS.
The embodiment of the application also provides a device for processing the ultra-high definition JPEG-XS IP data packet based on the FPGA, which comprises:
And (5) a decapsulation module: the method comprises the steps of performing decapsulation processing on each first IP data packet of 8K4K ultra-high definition JPEG-XS obtained through pretreatment to obtain a second IP data packet of 8K4K ultra-high definition JPEG-XS corresponding to each first IP data packet;
the calculation module: the flow control reference coefficient is used for calculating based on each second IP data packet in the second IP data packet flow to obtain a flow control reference coefficient corresponding to the second IP data packet flow;
model training module: the method comprises the steps of performing model training on an initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model;
And a strategy matching module: the target flow control strategy matching module is used for carrying out strategy matching by utilizing the target flow control strategy matching model based on each second IP data packet to obtain a target flow control strategy for each second IP data packet;
And the distribution processing module is used for: and the method is used for carrying out distribution processing on each second IP data packet by adopting the target flow control strategy to obtain an IP data packet processing result of 8K4K ultra-high definition JPEG-XS.
The embodiment of the application also provides a storage medium which stores a computer program, and the computer program realizes the steps of the method for processing the IP data packet of the ultra-high-definition JPEG-XS based on the FPGA when being executed by a processor.
The embodiment of the application also provides an electronic device, wherein the FPGA chip resource at least comprises an on-chip high-speed interface, an on-chip processor, a partition cache and an on-chip memory, the on-chip high-speed interface is used for receiving data, reading and writing the data and transmitting the data to each data processing module in parallel, the partition cache is used for dividing and forming a data array to generate data reference information, the on-chip memory stores a VHDL instruction program, and the on-chip processor directly runs the VHDL instruction program to execute calculation and processing so as to realize the steps of the method for processing the IP data packet of the ultra-high definition JPEG-XS based on the FPGA.
The embodiment of the application also provides an IP data packet processing system of the ultra-high definition JPEG-XS based on the FPGA, which comprises the following steps:
One or more processors;
A storage means for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the FPGA-based ultra-high definition JPEG-XS IP packet processing method as described above.
By adopting the technical scheme, the embodiment of the application has the following technical effects: by applying the method provided by the embodiment, each first IP data packet of 8K and 4K ultra-high-definition JPEG-XS obtained through pretreatment is subjected to decapsulation treatment to obtain a second IP data packet of 8K and 4K ultra-high-definition JPEG-XS corresponding to each first IP data packet; the data quality is improved, and the data volume of subsequent processing is reduced. Calculating based on each second IP data packet in the second IP data packet stream to obtain a flow control reference coefficient corresponding to the second IP data packet stream; performing model training on the initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model; performing policy matching by using the target flow control policy matching model based on each second IP data packet to obtain a target flow control policy for each second IP data packet; and distributing the second IP data packets by adopting the target flow control strategy to obtain IP data packet processing results of 8K and 4K ultra-high definition JPEG-XS. Classifying the data packet flows can realize the control of the service quality. Through classification and marking, the priority of each second IP data packet can be identified and distinguished, and then the sending sequence or discarding strategy of the second IP data packet is determined, so that the utilization of network resources can be optimized, network congestion is avoided, and smooth operation of key services can be ensured.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute a limitation on the application. In the drawings:
FIG. 1 is a flow chart of an FPGA-based ultra-high definition JPEG-XS IP data packet processing method according to an embodiment of the present application;
Fig. 2 is a block diagram of an IP packet processing apparatus of ultra-high definition JPEG-XS based on an FPGA according to an embodiment of the present application.
Detailed Description
In order to make the technical solutions and advantages of the embodiments of the present application more apparent, the following detailed description of exemplary embodiments of the present application is provided in conjunction with the accompanying drawings, and it is apparent that the described embodiments are only some embodiments of the present application and not exhaustive of all embodiments. It should be noted that, without conflict, the embodiments of the present application and features of the embodiments may be combined with each other.
Example 1
Fig. 1 is a flow chart of an FPGA-based ultra-high definition JPEG-XS IP packet processing method according to an embodiment of the present application, which includes:
Step S101: decapsulating each first IP data packet of the 8K and 4K ultrahigh-definition JPEG-XS obtained through pretreatment to obtain a second IP data packet of the 8K and 4K ultrahigh-definition JPEG-XS corresponding to each first IP data packet;
In practice, each initial IP data packet is received; extracting basic data from each initial IP data packet to obtain each intermediate IP data packet; preprocessing the IP data in each intermediate IP data packet to obtain each first IP data packet; wherein the preprocessing comprises: processing methods such as duplicate removal processing, verification processing, redundant information removal processing and the like.
In the specific implementation, capturing an initial IP data packet through a high-bandwidth optical fiber network interface, and performing caching and preliminary processing on the data packet; the preliminary processing comprises the steps of extracting basic data from each initial IP data packet to obtain each intermediate IP data packet; the basic data includes: general information such as IPv6, IPv4, UDP and the like of the initial IP data packet. The decapsulation processing tool may perform SMPTE ST 2110 protocol transmission and IP to SDI/SDI to IP protocol conversion through the device to decapsulate each first IP data packet into a second IP data packet corresponding to each first IP data packet. The first IP data packets are 8K and 4K ultra-high definition JPEG-XS data packets containing video data, audio data and a plurality of auxiliary information, and the second IP data packets containing 8K and 4K ultra-high definition JPEG-XS of the video data in the first IP data packets corresponding to the first IP data packets are obtained after the decapsulation processing of the application.
Step S102: calculating based on each second IP data packet in the second IP data packet stream to obtain a flow control reference coefficient corresponding to the second IP data packet stream;
In the implementation, feature extraction is carried out on each second IP data packet to obtain a plurality of data transmission feature values of the second IP data packet flow; calculating based on each data transmission characteristic value and a corresponding preset characteristic weight value to obtain a comprehensive characteristic value corresponding to the second IP data packet flow; based on each second IP data packet, calculating by adopting a preset variance function to obtain a time delay variation degree coefficient corresponding to the second IP data packet flow; calculating based on the byte number of each second IP data packet, the comprehensive characteristic value and the time delay variation degree coefficient to obtain a flow control reference coefficient corresponding to the second IP data packet flow; wherein the data transmission feature comprises: bandwidth utilization, transmission rate, number of retransmissions, traffic burstiness, and inter-packet correlation.
In a specific implementation, the data transmission features include bandwidth utilization, transmission rate, retransmission times, traffic burstiness, inter-packet correlation, and the like. The bandwidth utilization rate refers to the current network bandwidth use condition, usually expressed by percentage, and is used for evaluating the network load condition so as to perform flow control and resource allocation; in particular, bandwidth utilization may be calculated by monitoring the input and output rates of links or lanes. The transmission rate may be calculated based on the number of IP packets arriving per unit time or the total number of bytes. The retransmission times, specifically, the retransmission times of the IP data packet, are used for evaluating the network congestion condition; whether retransmission conditions exist or not can be judged according to the serial number or the confirmation number of the protocol, and the times of retransmission are counted. Traffic burstiness, which is the condition that a large number of data packets arrive in a short time; the method can detect whether the flow is higher than a preset threshold value or not by counting the number of data packets or bits which arrive in unit time, and analyze the flow to obtain the burstiness of the flow. The inter-packet correlation can judge whether the packet loss or disorder phenomenon exists according to the sequence number and the confirmation number fields of the data packets, and the correlation between the data packets, namely the inter-packet correlation, is obtained through analysis.
In a specific implementation, based on each data transmission characteristic value and each preset characteristic weight value, calculation processing is performed to obtain a comprehensive characteristic value corresponding to the second IP data packet flowThe mathematical expression is as shown in the following formula (1):
(1)
Wherein, In order for the bandwidth to be utilized,In order to achieve a transmission rate of the data,The number of retransmissions is determined based on the number of retransmissions,In order for the traffic to be bursty,In order to be a correlation between the packets,、、、、For each predetermined feature weight value. Based on each second IP data packet, adopting a preset variance function to perform calculation processing to obtain a time delay variation degree coefficient corresponding to the second IP data packet flowThe mathematical expression is as shown in the following formula (2):
(2) ; wherein, For the number of each of said second IP packets,Indicating that the i-th second IP packet,The average value of the samples is shown. Byte count based on each of the second IP packetsThe integrated characteristic valueThe time delay variation degree coefficientCalculating to obtain flow control reference coefficient corresponding to the second IP data packet flowThe mathematical expression is as shown in the following formula (3):
(3) ; wherein, 、、Each representing a predetermined weight coefficient of each item in the flow control reference coefficient formula.
Step S103: performing model training on the initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model;
In implementation, policy feature screening is performed based on each second IP packet and each historical packet, so as to obtain each sample data set for model training, wherein each second IP packet carries the flow control reference coefficient at the current moment, and each historical packet carries a historical flow control reference coefficient corresponding to each historical moment; extracting features of the strategy labels in each sample data set to obtain a feature data set; dividing the characteristic data set into data sets to obtain a plurality of training sets and a plurality of test sets; model training is carried out on the initial flow control strategy matching model by adopting each training set and each testing set, and a target flow control strategy matching model meeting preset conditions is generated; wherein the policy tag feature comprises: one or more of a flow control reference coefficient, a slope, a priority flag, and a bandwidth utilization; the feature data in the feature data set includes: one or more of the type, length and priority of the data packet.
In the implementation, a flow control reference coefficient change trend graph is generated and displayed based on the flow control reference coefficient at the current moment and the historical flow control reference coefficients corresponding to the historical moments.
In particular implementations, the slope in the policy tag featureThe mathematical expression of (2) is as shown in the following formula (4): (4) ; wherein, Indicating the flow control reference coefficient at the current time,A flow control reference coefficient representing a previous time,Representing the time interval between these two moments. If the slope is positive, indicating that the flow control reference coefficient is increasing, indicating that more BRAM is needed to be allocated and utilized to control the data flow; if the slope is negative, the flow control reference coefficient is reduced, and the surplus BRAM resource can be released; if the slope is close to zero, it means that the flow control reference remains substantially stable. Performing policy feature screening based on each second IP data packet and each historical data packet to obtain each sample data set for model training; performing policy feature screening on each second IP data packet to obtain a first sample data set corresponding to each second IP data packet; performing policy feature screening on each historical data packet to obtain a second sample data set corresponding to each historical data packet; wherein each sample data set comprises each first sample data set and each second sample data set, the characteristics of each first sample data set comprising: flow control reference coefficient, slope, priority mark, bandwidth utilization rate and other data at the current moment; the policy features of each second sample dataset include: data such as flow control reference coefficients, slopes, priority flags, bandwidth utilization, and the like at each historical time over a period of time. The priority mark is used for indicating the priority of the data packet and is used for determining the sending sequence or discarding strategy of the data packet. Carrying out feature extraction on the strategy marking features in each sample data set to obtain a plurality of feature data sets; performing feature scaling processing on the strategy marking features in each sample data set by using a target feature scaling method to obtain each feature scaling data set, wherein the target feature scaling method can be as follows: the application does not limit the target feature scaling method by using the Min-Max standardization method or the Z-score standardization method, and adopts the target feature selection method to screen the feature scaling data in each feature scaling data set to obtain a feature filtering data packet corresponding to each feature scaling data set; and carrying out feature transformation processing on the feature filtering data in each feature filtering data packet to obtain a feature data set corresponding to each feature filtering data packet. Dividing the feature data sets into data sets to obtain a plurality of training sets and a plurality of test sets; each of the feature data sets may be divided into a training set and a test set in a ratio of 7:3, 8:2, or 9:1, wherein the training set is used to train the model and the test set is used to evaluate the performance of the model. The application does not limit the dividing proportion of the training set and the testing set, and the dividing proportion of the training set and the testing set can be set according to actual needs.
In the implementation, training an initial decision tree classification model in the initial flow control strategy matching model and an initial regression model in the initial flow control strategy matching model by adopting the training set to generate an intermediate decision tree classification model and an intermediate regression model; evaluating the intermediate decision tree classification model and the intermediate regression model based on each test set to obtain an evaluation result; when the evaluation result is that the classification accuracy of the intermediate decision tree classification model is lower than a first preset threshold value and/or the regression error value of the intermediate regression model exceeds a second preset threshold value, carrying out model parameter adjustment on model parameters of the intermediate decision tree classification model and/or the intermediate regression model to obtain an updated intermediate decision tree classification model and/or an updated intermediate regression model, and carrying out parameter adjustment on the updated intermediate decision tree classification model and the updated intermediate regression model in a cyclic iteration mode until the evaluation result is that the classification accuracy of the intermediate decision tree classification model is higher than the first preset threshold value and the regression error value of the intermediate regression model is lower than the second preset threshold value, determining the updated intermediate decision tree classification model as a target decision tree classification model and determining the updated intermediate regression model as a target regression model to obtain the target flow control strategy matching model; determining the intermediate decision tree classification model as a target decision tree classification model and determining the intermediate regression model as a target regression model under the condition that the classification accuracy of the intermediate decision tree classification model is larger than a first preset threshold and the regression error value of the intermediate regression model is smaller than a second preset threshold as the evaluation result, so as to obtain the target flow control strategy matching model; wherein the target flow control strategy matching model comprises: a target decision tree classification model and a target regression model.
In specific implementation, training the initial decision tree model and the initial regression model respectively by using a training set, so that the initial decision tree model can classify the data packet stream, and the initial regression model can generate strategies for the classified data packet stream. The method comprises the steps of classifying a data packet stream by using a decision tree model, specifically classifying the data packet of the stream, wherein the classification of the data packet can take various classification modes into consideration, and common classification modes are classification modes such as message types, message lengths, priority and the like. For example, the classification may be performed by header information of the IP packet, or may be performed according to the priority of the packet. For some specific network protocol messages, such as VLAN messages and IP messages, 802.1p, DSCP, etc. may be used as priority labels to classify the data packet flows. When the regression model is trained, the historical data packet and the characteristic data related to strategy generation are used for adjusting model parameters so that the model parameters can be predicted to be suitable for the flow control strategy of the corresponding category. In the flow control policy generation process, the characteristic data includes various policy-related parameters. Such as type, length, priority, etc. of IP packets. In addition, historical data packet and real-time monitoring data packet information can also be used as characteristic data for training a regression model. These characteristic data are used to input into the regression model, which by adjusting the model parameters enables it to predict flow control strategies appropriate for a particular class.
In specific implementation, the mathematical calculation formula of the classification accuracy is shown as the following formula (5):
(5) ; the mathematical calculation formula of the regression error value is shown as the following formula (6):
(6) ; wherein, Indicating the number of samples that are correctly classified,The number of samples to be taken in total is indicated,Representing the actual policy tag value that is to be used,Policy tag values representing model predictions. And evaluating the classification model based on the calculated classification accuracy and evaluating the regression model based on the calculated regression error value, and optimally adjusting the model, including adjusting model parameters or increasing training sample size, to obtain a target classification model and a target regression model which meet preset conditions so as to train and obtain the target flow control strategy matching model.
Step S104: performing policy matching by using the target flow control policy matching model based on each second IP data packet to obtain a target flow control policy for each second IP data packet;
In the implementation, classifying treatment is carried out by adopting a target decision tree classification model of the target flow control strategy matching model based on each second IP data packet to obtain classification identification information corresponding to each second IP data packet; based on the classification identification information, predicting the flow control strategy of each second IP data packet by adopting a target regression model of the target flow control strategy matching model to obtain an initial flow control strategy for each second IP data packet; and performing strategy adjustment on the initial flow control strategy based on the flow control reference coefficient and each historical flow control reference coefficient to obtain a target flow control strategy for each second IP data packet.
In the implementation, inputting each second IP data packet into a target decision tree classification model for classification processing to obtain classification identification information corresponding to each second IP data packet so as to obtain a classified data packet stream after classification processing; and then inputting the classified data packet flow after the classification processing into a target regression model for flow control strategy prediction to obtain an initial flow control strategy for each second IP data packet, wherein the flow control strategy comprises the following steps: adjusting the sending rate, discarding the data packet or reallocating the bandwidth, etc.; and performing strategy adjustment on the initial flow control strategy based on the flow control reference coefficient and each historical flow control reference coefficient to obtain a target flow control strategy for each second IP data packet.
Step S105: and distributing the second IP data packets by adopting the target flow control strategy to obtain IP data packet processing results of 8K and 4K ultra-high definition JPEG-XS.
In a specific implementation, the second IP packet stream is duplicated and distributed to a plurality of processing units or threads for data monitoring and flow control processing. Specifically, the buffer area and sequencing mechanism technology can be adopted to process disordered and delayed data packets, a multi-copy redundancy processing strategy is used to copy the data packet flow for multiple copies, the data packets are distributed to multiple processing units or threads in a random distribution or load balancing algorithm-based distribution mode, and the full-frame 8K or 4K signal visual lossless processing function of the VFN28-XS series device is utilized for processing. The parallel processing module fully utilizes the parallelism of the multi-core processor, improves the data processing speed and efficiency, adopts a buffer zone, a sequencing mechanism technology and a multi-copy redundancy processing strategy, enhances the fault tolerance and the anti-interference capability of the system, enables the system to better cope with faults and interference, and maintains the reliability and the stability of data processing; the buffer area, the ordering mechanism technology, the multi-copy redundancy processing strategy and the load balancing algorithm belong to the prior art means, so the embodiment does not make a specific description; the full-frame 8K or 4K signal visual lossless processing function of the VFN28-XS series device can process high-resolution video signals and perform real-time transmission or conversion operation.
Further, the embodiment of the application further comprises:
In the implementation, the BRAM utilization rate index value, the cache utilization rate index value and the resource utilization rate index value on the FPGA chip, which are obtained based on real-time monitoring, are calculated by adopting preset mapping functions to obtain the BRAM utilization rate abnormal score, the cache utilization rate abnormal score and the resource utilization rate abnormal score on the FPGA chip; based on the real-time reading, the number of the sent second IP data packets, the number of the received second IP data packets, the transmission time parameter, the time interval average value, the time interval number and the preset adjustment coefficients, calculating by adopting a preset data transmission quality function to obtain a real-time data transmission quality value; calculating and processing by adopting a preset abnormal index function based on the BRAM utilization rate abnormal score on the FPGA chip, the cache utilization rate abnormal score on the FPGA chip, the resource utilization rate abnormal score, the real-time data transmission quality value and each preset abnormal index proportionality coefficient to obtain an abnormal index value; and under the condition that the abnormal index value is greater than or equal to a preset abnormal index threshold value, generating a reference IP data packet and reference information, and writing the reference information into a data matrix of the FPGA to process the IP data packet again.
In particular implementations, a real-time data transmission quality valueThe mathematical calculation formula of (2) is shown as the following formula (7):
(7) ; wherein, Indicating the number of each of said second IP packets transmitted,Indicating the number of received second IP packets,A transmission time parameter is indicated and a time of flight parameter is indicated,A time interval parameter indicating receipt of the u-th packet,Representing the average value of the time interval over which the data packet is received,Indicating the number of time intervals in which the data packet was received,、、Representing the values of the predetermined adjustment coefficient parameters in the data transmission quality formula. Abnormal index valueThe mathematical calculation formula of (2) is shown as the following formula (8):
(8) ; wherein, For the BRAM utilization anomaly score,For the abnormal score of the memory utilization,For the resource utilization anomaly score,For the real-time data transmission quality value,、、、Scaling coefficients for each predetermined abnormality index. And under the condition that the abnormal index value is smaller than a preset abnormal index threshold value, the current data packet processing process is normal, and under the condition that the abnormal index value is larger than or equal to the preset abnormal index threshold value, a reference IP data packet and reference information are generated and written into a data matrix of the FPGA to be cached so as to process the IP data packet again.
The present application takes a specific scenario as an example, and a first embodiment of the present application will be described in detail.
Capturing each initial IP data packet through a high-bandwidth optical fiber network interface, caching each captured initial IP data packet, extracting basic data information such as IPv6, IPv4, UDP and the like of the initial IP data packet to obtain each intermediate IP data packet, and carrying out preprocessing operations such as de-duplication processing, verification processing, redundant information removal and the like on each intermediate IP data packet to obtain each IP data packet.
And adopting the functions of full-frame 8K or 4K signal visual lossless processing, ST 2110 IP data conversion and real-time transmission provided by a preset decapsulation processing device, and adopting a built-in 8K or 4K full-frame JPEG-XS visual lossless format encoding and decoding module to encode and decode each IP data packet to obtain a second IP data packet corresponding to each IP data packet.
Extracting bandwidth utilization in a second IP packet streamTransmission rateNumber of retransmissionsBurstiness of flowInter-package correlationThe data transmission characteristics are equal, and the comprehensive characteristic value is calculated by adopting a comprehensive characteristic value calculation formula according to the data transmission characteristics; Calculating a delay variation degree coefficient of a data packet by using a variance function; Based on packet sizeIntegrated characteristic valueAnd a coefficient of degree of variation of the time delayCalculating a flow control reference coefficient。
Based on the flow control reference coefficient at the current moment and the historical flow control reference coefficients corresponding to each historical moment, a flow control reference coefficient change trend graph is generated and displayed so as to analyze the change trend of the flow control reference coefficient, and a foundation is laid for the subsequent strategy adjustment of the initial flow control strategy based on the flow control reference coefficient and each historical flow control reference coefficient.
A data set for training and testing is collected, including historical data packets, real-time monitored second IP data packets, and policy features associated with the flow control policy, including flow control reference coefficients, slopes, priority flags, and bandwidth utilization, among others. And performing preprocessing, feature scaling, feature selection or feature transformation and other operations on the strategy marking features in the data set, and dividing the data set into a training set and a testing set. Training the decision tree model and the regression model using the training set to enable the decision tree model to classify the packet stream and enable the regression model to strategically generate the classified packet stream. For example: the classification may be performed by header information of the data packet, and the IP header includes fields of version, protocol, source IP address, destination IP address, etc., and the data packet may be classified according to characteristics of the IP header. The classification may also be based on the priority of the messages. Such as VLAN messages and IP messages, 802.1p, DSCP, etc. may be used as priority labels. In the process of flow control policy generation, the characteristic data includes various policy-related data, such as the type, length, priority, etc., of the data packet. When the regression model is trained, the model parameters are adjusted by using the historical data and the characteristic data related to strategy generation, so that the control strategy suitable for adjusting the sending rate, discarding the data packet or reallocating the bandwidth flow and the like of the corresponding category can be predicted. And evaluating the decision tree model and the regression model by using a test set, calculating a classification accuracy and a regression error index, performing optimization adjustment on the model under the condition that the classification accuracy is lower than a first preset threshold value and/or the regression error value of the intermediate regression model exceeds a second preset threshold value, including adjusting model parameters or increasing training sample size, and training the model in a cyclic iteration mode until a final target classification model and a target regression model are obtained under the condition that the classification accuracy is higher than the first preset threshold value and the regression error value is lower than the second preset threshold value so as to obtain a target flow control strategy matching model.
Classifying the data packet flows by using a target decision tree model meeting the evaluation requirements, and generating a preliminary flow control strategy for the classified data packet flows generated after classification processing by using a target regression model meeting the evaluation requirements. In the specific operation, the second IP data packet after being unpacked is input into a target decision tree model, each data packet is classified according to a decision path, classified data packet flows are output through the target decision tree model, the classified data packet flows after being classified are input into a target regression model, and the target regression model predicts the preliminary flow control strategy of the corresponding category. And performing policy adjustment on the initial flow control policy based on the flow control reference coefficient and each historical flow control reference coefficient to obtain a policy of adjusting a sending rate and the like for the target flow control policy of each second IP data packet.
The method can process disordered and delayed data packets by adopting a buffer area and sequencing mechanism technology, copy multiple data packet streams by using a multiple copy redundancy processing strategy, distribute the data packets to multiple processing units or threads by a mode of random distribution or distribution based on a load balancing algorithm, process the data packets by utilizing a full-frame 8K or 4K signal visual lossless processing function of VFN28-XS series equipment, and transmit the data packets in real time by adopting a ST2110 data transmission protocol.
The data monitoring module is used for monitoring the performance utilization rate and the data transmission quality in real time to obtain BRAM utilization rate index values, memory utilization rate index values and resource utilization rate index values, and a predetermined mapping function is used for calculating and processing to obtain BRAM utilization rate abnormal scores, memory utilization rate abnormal scores and resource utilization rate abnormal scores. Mapping each index value into a range of [0,1] by using each preset mapping function which is linear or nonlinear to obtain each abnormal score; based on the real-time reading, the number of the sent second IP data packets, the number of the received second IP data packets, the transmission time parameter, the time interval average value, the time interval number and the preset adjustment coefficients, calculating by adopting a preset data transmission quality function to obtain a real-time data transmission quality value; further, based on each abnormal score and each corresponding preset abnormal index proportional coefficient, carrying out calculation processing by adopting a preset abnormal index function to obtain an abnormal index value Q; judging and comparing the system abnormality index Q with a preset system abnormality index threshold value Qt, if Q is more than or equal to Qt, indicating that the system has an abnormality condition, and automatically resetting; otherwise, the system is indicated to be normal; the size of Qt can be set according to practical needs.
And finally, outputting the data packet stream processed by the data monitoring and flow control strategies. And sending the data packet stream to the target device or the target application program for playing, storing or transmitting, wherein the specific output mode is determined according to the actual requirement, for example: and sending the data packet stream processed by the data monitoring and flow control strategies to an audio/video playing module or equipment, and storing the data packet stream in a database or transmitting the data packet stream to other network nodes.
By applying the method provided by the embodiment, each IP data packet obtained through pretreatment is subjected to decapsulation treatment to obtain a second IP data packet corresponding to each IP data packet; the data quality is improved, and the data volume of subsequent processing is reduced. Calculating based on each second IP data packet in the second IP data packet stream to obtain a flow control reference coefficient corresponding to the second IP data packet stream; performing model training on the initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model; performing policy matching by adopting the target flow control policy matching model based on each second IP data packet to obtain a target flow control policy for each second IP data packet; and distributing the second IP data packets by adopting the target flow control strategy to obtain IP data packet processing results. Classifying the data packet flows can realize the control of the service quality. Through classification and marking, the priority of each second IP data packet can be identified and distinguished, and then the sending sequence or discarding strategy of the second IP data packet is determined, so that the utilization of network resources can be optimized, network congestion is avoided, and smooth operation of key services can be ensured.
Based on the same application conception, the embodiment of the application also provides an IP data packet processing device of the ultra-high-definition JPEG-XS based on the FPGA, and because the principle of solving the problems of the devices is similar to that of the IP data packet processing of the ultra-high-definition JPEG-XS based on the FPGA, the implementation of the devices can be referred to the implementation of the method, and the repetition is omitted.
Fig. 2 is a schematic structural diagram of an IP packet processing device of the FPGA-based ultra-high definition JPEG-XS according to an embodiment of the present application. As shown in fig. 2, an apparatus for processing an IP packet of ultra-high definition JPEG-XS based on an FPGA according to an embodiment of the present application includes:
Decapsulation module 201: the method comprises the steps of performing decapsulation processing on first IP data packets of 8K and 4K ultrahigh-definition JPEG-XS obtained through pretreatment to obtain second IP data packets of 8K and 4K ultrahigh-definition JPEG-XS corresponding to the first IP data packets;
The calculation module 202: the flow control reference coefficient is used for calculating based on each second IP data packet in the second IP data packet flow to obtain a flow control reference coefficient corresponding to the second IP data packet flow;
Model training module 203: the method comprises the steps of performing model training on an initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model;
Policy matching module 204: the target flow control strategy matching module is used for carrying out strategy matching by utilizing the target flow control strategy matching model based on each second IP data packet to obtain a target flow control strategy for each second IP data packet;
distribution processing module 205: and the method is used for carrying out distribution processing on each second IP data packet by adopting the target flow control strategy to obtain IP data packet processing results of 8K and 4K ultra-high definition JPEG-XS.
In implementation, the device for processing the IP data packet of the ultra-high definition JPEG-XS based on the FPGA further comprises:
a receiving unit: for receiving each initial IP packet;
Basic data extraction unit: the method comprises the steps of extracting basic data from each initial IP data packet to obtain each intermediate IP data packet;
Pretreatment unit: the method comprises the steps of preprocessing IP data in each intermediate IP data packet to obtain each first IP data packet; wherein the preprocessing comprises: one or more of the processes of de-duplication, verification and redundancy information removal.
In implementation, the computing module 202 specifically includes:
A first feature extraction unit: the method comprises the steps of extracting characteristics of each second IP data packet to obtain each data transmission characteristic value of the second IP data packet flow;
A first calculation unit: the data transmission characteristic values are used for carrying out calculation processing based on the data transmission characteristic values and the preset characteristic weight values to obtain comprehensive characteristic values corresponding to the second IP data packet flow;
a second calculation unit: the method comprises the steps of calculating and processing by adopting a preset variance function based on each second IP data packet to obtain a time delay variation degree coefficient corresponding to the second IP data packet flow;
A third calculation unit: the flow control reference coefficient corresponding to the second IP data packet flow is obtained by calculating based on the byte number of each second IP data packet, the comprehensive characteristic value and the time delay variation degree coefficient; wherein each of the data transmission features comprises: bandwidth utilization, transmission rate, number of retransmissions, traffic burstiness, and inter-packet correlation.
In implementation, the model training module 203 specifically includes:
Feature screening unit: the method comprises the steps of carrying out strategy marking feature screening based on each second IP data packet and each historical data packet to obtain each sample data set used for model training, wherein each second IP data packet carries the flow control reference coefficient at the current moment, and each historical data packet carries the historical flow control reference coefficient corresponding to each historical moment;
A second feature extraction unit: the method comprises the steps of performing feature extraction on strategy marking features in each sample data set to obtain a plurality of feature data sets;
A data set dividing unit: the method comprises the steps of dividing each characteristic data set into data sets to obtain a plurality of training sets and a plurality of test sets;
model training unit: the method comprises the steps of training the initial flow control strategy matching model by using training sets and testing sets to generate a target flow control strategy matching model meeting preset conditions; wherein the policy tag feature comprises: one or more of a flow control reference coefficient, a slope, a priority flag, and a bandwidth utilization; the feature data in each of the feature data sets includes: one or more of the type, length and priority of the data packet.
In implementation, the model training unit of the model training module 203 specifically includes:
Training unit: the training set is used for training an initial decision tree classification model in the initial flow control strategy matching model and an initial regression model in the initial flow control strategy matching model respectively to generate an intermediate decision tree classification model and an intermediate regression model;
An evaluation unit: the intermediate decision tree classification model and the intermediate regression model are evaluated based on the test sets, and an evaluation result is obtained;
A first determination unit: the method comprises the steps of carrying out model parameter adjustment on model parameters of an intermediate decision tree classification model and/or an intermediate regression model under the condition that the evaluation result is that the classification accuracy of the intermediate decision tree classification model is smaller than or equal to a first preset threshold value and/or the regression error value of the intermediate regression model is larger than or equal to a second preset threshold value, obtaining an updated intermediate decision tree classification model and/or an updated intermediate regression model, carrying out parameter adjustment on the updated intermediate decision tree classification model and the updated intermediate regression model in a cyclic iteration mode until the evaluation result is that the classification accuracy of the intermediate decision tree classification model is larger than the first preset threshold value and the regression error value of the intermediate regression model is smaller than the second preset threshold value, determining the updated intermediate decision tree classification model as a target decision tree classification model and determining the updated intermediate regression model as a target regression model, and obtaining the target flow control strategy matching model;
A second determination unit: the method comprises the steps of determining an intermediate decision tree classification model as a target decision tree classification model and determining the intermediate regression model as a target regression model under the condition that the classification accuracy of the intermediate decision tree classification model is larger than a first preset threshold value and the regression error value of the intermediate regression model is smaller than a second preset threshold value as the evaluation result, so as to obtain the target flow control strategy matching model; wherein the target flow control strategy matching model comprises: a target decision tree classification model and a target regression model.
In implementation, the policy matching module 204 specifically includes:
A classification processing unit: the target decision tree classification model is used for carrying out classification processing by adopting the target flow control strategy matching model based on each second IP data packet to obtain classification identification information corresponding to each second IP data packet;
Prediction unit: the target regression model is used for predicting the flow control strategy of each second IP data packet by adopting the target flow control strategy matching model based on each piece of classification identification information to obtain an initial flow control strategy for each second IP data packet;
Policy adjustment unit: and the method is used for carrying out strategy adjustment on the initial flow control strategy based on the flow control reference coefficient and each historical flow control reference coefficient to obtain a target flow control strategy for each second IP data packet.
In implementation, the device for processing the IP data packet of the ultra-high definition JPEG-XS based on the FPGA further comprises:
Curve generation unit: and the flow control reference coefficient change trend graph is generated based on the flow control reference coefficient at the current moment and the historical flow control reference coefficients corresponding to the historical moments so as to show the change trend of the flow control reference coefficients.
In an implementation, the IP packet processing apparatus further includes:
A fourth calculation unit: the method comprises the steps that a BRAM utilization rate index value, a cache utilization rate index value and a resource utilization rate index value on an FPGA chip, which are obtained based on real-time monitoring, are calculated by adopting preset mapping functions, so that a BRAM utilization rate abnormal score, a cache utilization rate abnormal score and a resource utilization rate abnormal score on the FPGA chip are obtained;
A fifth calculation unit: the method comprises the steps of calculating and processing the number of the sent second IP data packets, the number of the received second IP data packets, a transmission time parameter value, a time interval average value parameter value, a time interval number and each preset adjustment coefficient parameter value by adopting a preset data transmission quality function based on real-time reading to obtain a real-time data transmission quality value;
A sixth calculation unit: the method comprises the steps of calculating and processing by adopting a preset abnormal index function based on the BRAM utilization abnormal score on the FPGA chip, the cache utilization abnormal score on the FPGA chip, the resource utilization abnormal score, the real-time data transmission quality value and each preset abnormal index proportionality coefficient to obtain an abnormal index value;
A reset unit: and the method is used for generating a reference IP data packet and reference information under the condition that the abnormal index value is larger than or equal to a preset abnormal index threshold value, and writing the reference IP data packet and the reference information into a data matrix of the FPGA so as to process the IP data packet again.
Based on the same inventive concept, the embodiment of the application also provides a storage medium, wherein the storage medium stores a program of instructions and a computer program of a programmable chip, and the program of instructions and the computer program of the programmable chip realize the following method steps when being executed by a processor:
Step one, decapsulating each first IP data packet of the 8K and 4K ultrahigh-definition JPEG-XS obtained through pretreatment to obtain a second IP data packet of the 8K4K ultrahigh-definition JPEG-XS corresponding to each first IP data packet;
calculating based on each second IP data packet in the second IP data packet stream to obtain a flow control reference coefficient corresponding to the second IP data packet stream;
thirdly, performing model training on an initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model;
Step four, based on the second IP data packets, performing policy matching by using the target flow control policy matching model to obtain target flow control policies for the second IP data packets;
And fifthly, distributing the second IP data packets by adopting the target flow control strategy to obtain an IP data packet processing result of 8K4K ultra-high definition JPEG-XS.
The specific implementation process of the steps of the method can be referred to the embodiment of any of the above-mentioned method for processing the IP data packet of the ultra-high definition JPEG-XS based on the FPGA, and the embodiment is not repeated here.
Another embodiment of the present application provides an electronic device, where FPGA on-chip resources of the electronic device at least include an on-chip high-speed interface, an on-chip processor, a partition cache, and an on-chip memory, where the on-chip high-speed interface is used to receive data, read and write data, and transmit the data in parallel to each data processing module, the partition cache is used to divide and form a data array to generate data reference information, and the on-chip memory stores a VHDL instruction program, and when the on-chip processor directly runs the VHDL instruction program to perform computation and processing, the method steps are implemented as follows:
Step one, decapsulating each first IP data packet of 8K and 4K ultrahigh-definition JPEG-XS obtained through pretreatment to obtain a second IP data packet of 8K and 4K ultrahigh-definition JPEG-XS corresponding to each first IP data packet;
calculating based on each second IP data packet in the second IP data packet stream to obtain a flow control reference coefficient corresponding to the second IP data packet stream;
thirdly, performing model training on an initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model;
Step four, based on the second IP data packets, performing policy matching by using the target flow control policy matching model to obtain target flow control policies for the second IP data packets;
and fifthly, distributing the second IP data packets by adopting the target flow control strategy to obtain IP data packet processing results of 8K and 4K ultrahigh-definition JPEG-XS.
The specific implementation process of the steps of the method can be referred to the embodiment of any of the above-mentioned method for processing the IP data packet of the ultra-high definition JPEG-XS based on the FPGA, and the embodiment is not repeated here.
Based on the same inventive concept, the embodiment of the application also provides an IP data packet processing system of the ultra-high-definition JPEG-XS based on the FPGA, and the principle of solving the problems by the equipment is similar to that of an IP data packet processing method of the ultra-high-definition JPEG-XS based on the FPGA, and an IP data packet processing device of the ultra-high-definition JPEG-XS based on the FPGA, so that the implementation of the equipment can be seen in the implementation of the method, and the repeated parts are omitted.
The FPGA-based ultra-high definition JPEG-XS IP data packet processing system can comprise:
One or more processors;
A storage means for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the FPGA-based ultra-high definition JPEG-XS IP packet processing method as described above.
For convenience of description, the parts of the above apparatus are described as being functionally divided into various modules or units, respectively. Of course, the functions of each module or unit may be implemented in the same piece or pieces of software or hardware when implementing the present application.
It will be apparent to those skilled in the art that embodiments of the present application may be provided as a method, apparatus, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the application may take the form of programmable chip instructions and programs, and computer program products, embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flowchart and/or block of the flowchart illustrations and/or block diagrams, and combinations of flowcharts and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (11)
1. The method for processing the IP data packet of the ultra-high definition JPEG-XS based on the FPGA is characterized by comprising the following steps of:
decapsulating each first IP data packet of the 8K and 4K ultrahigh-definition JPEG-XS obtained through pretreatment to obtain a second IP data packet of the 8K4K ultrahigh-definition JPEG-XS corresponding to each first IP data packet;
Calculating based on each second IP data packet in the second IP data packet stream to obtain a flow control reference coefficient corresponding to the second IP data packet stream;
performing model training on the initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model;
Performing policy matching by using the target flow control policy matching model based on each second IP data packet to obtain a target flow control policy for each second IP data packet;
Distributing each second IP data packet by adopting the target flow control strategy to obtain IP data packet processing results of 8K and 4K ultra-high definition JPEG-XS;
Calculating based on each second IP packet in the second IP packet stream to obtain a flow control reference coefficient corresponding to the second IP packet stream, including:
Extracting the characteristics of each second IP data packet to obtain a plurality of data transmission characteristic values of the second IP data packet flow;
Calculating based on each data transmission characteristic value and a corresponding preset characteristic weight value to obtain a comprehensive characteristic value corresponding to the second IP data packet flow;
Based on each second IP data packet, calculating by adopting a preset variance function to obtain a time delay variation degree coefficient corresponding to the second IP data packet flow;
Calculating based on the byte number of each second IP data packet, the comprehensive characteristic value and the time delay variation degree coefficient to obtain a flow control reference coefficient corresponding to the second IP data packet flow;
wherein the data transmission feature comprises: bandwidth utilization, transmission rate, number of retransmissions, traffic burstiness, and inter-packet correlation.
2. The method of claim 1, wherein the decapsulating process is performed on each of the first IP packets of the 8K and 4K ultra-high definition JPEG-XS obtained by the preprocessing, and the method further comprises:
receiving each initial IP data packet;
extracting basic data from each initial IP data packet to obtain each intermediate IP data packet;
preprocessing the IP data in each intermediate IP data packet to obtain each first IP data packet;
Wherein the preprocessing comprises: one or more of the processes of de-duplication, verification and redundancy information removal.
3. The method of claim 2, wherein model training an initial flow control policy matching model based on the flow control reference coefficient, each of the second IP packets, and each of the historical packets, generating a target flow control policy matching model comprises:
Performing policy feature screening based on the second IP data packets and the historical data packets to obtain sample data sets for model training, wherein the second IP data packets carry the flow control reference coefficients at the current moment, and the historical data packets carry the historical flow control reference coefficients corresponding to the historical moments;
Extracting features of the strategy labels in each sample data set to obtain a feature data set;
dividing the characteristic data set into data sets to obtain a plurality of training sets and a plurality of test sets;
Model training is carried out on the initial flow control strategy matching model by adopting each training set and each testing set, and a target flow control strategy matching model meeting preset conditions is generated;
Wherein the policy feature comprises: one or more of a flow control reference coefficient, a slope, a priority flag, and a bandwidth utilization; the feature data in the feature data set includes: one or more of the type, length and priority of the data packet.
4. The method of claim 3, wherein model training the initial flow control strategy matching model using each of the training sets and each of the test sets to generate a target flow control strategy matching model that meets a preset condition, comprising:
Training an initial decision tree classification model in the initial flow control strategy matching model and an initial regression model in the initial flow control strategy matching model by adopting the training set to generate an intermediate decision tree classification model and an intermediate regression model;
Evaluating the intermediate decision tree classification model and the intermediate regression model based on each test set to obtain an evaluation result;
When the evaluation result is that the classification accuracy of the intermediate decision tree classification model is lower than a first preset threshold value and/or the regression error value of the intermediate regression model exceeds a second preset threshold value, carrying out model parameter adjustment on model parameters of the intermediate decision tree classification model and/or the intermediate regression model to obtain an updated intermediate decision tree classification model and/or an updated intermediate regression model, and carrying out parameter adjustment on the updated intermediate decision tree classification model and the updated intermediate regression model in a cyclic iteration mode until the evaluation result is that the classification accuracy of the intermediate decision tree classification model is higher than the first preset threshold value and the regression error value of the intermediate regression model is lower than the second preset threshold value, determining the updated intermediate decision tree classification model as a target decision tree classification model and determining the updated intermediate regression model as a target regression model to obtain the target flow control strategy matching model;
Determining the intermediate decision tree classification model as a target decision tree classification model and determining the intermediate regression model as a target regression model under the condition that the classification accuracy of the intermediate decision tree classification model is larger than a first preset threshold and the regression error value of the intermediate regression model is smaller than a second preset threshold as the evaluation result, so as to obtain the target flow control strategy matching model;
Wherein the target flow control strategy matching model comprises: a target decision tree classification model and a target regression model.
5. The method according to claim 1 or 4, wherein performing policy matching by using the target flow control policy matching model based on each of the second IP packets to obtain a target flow control policy for each of the second IP packets, includes:
Classifying the second IP data packets by adopting a target decision tree classification model of the target flow control strategy matching model to obtain classification identification information corresponding to the second IP data packets;
based on the classification identification information, predicting the flow control strategy of each second IP data packet by adopting a target regression model of the target flow control strategy matching model to obtain an initial flow control strategy for each second IP data packet;
And performing strategy adjustment on the initial flow control strategy based on the flow control reference coefficient and each historical flow control reference coefficient to obtain a target flow control strategy for each second IP data packet.
6. The method of claim 5, wherein the method further comprises:
and generating and displaying a flow control reference coefficient change trend graph based on the flow control reference coefficient at the current moment and the historical flow control reference coefficients corresponding to the historical moments.
7. The method of claim 2, wherein after distributing each of the second IP packets using the target flow control policy, the method further comprises:
The method comprises the steps that based on a BRAM utilization index value, a cache utilization index value and a resource utilization index value on an FPGA chip obtained through real-time monitoring, calculation processing is conducted through preset mapping functions, and a BRAM utilization abnormal score, a cache utilization abnormal score and a resource utilization abnormal score on the FPGA chip are obtained;
Based on the real-time reading, the number of the sent second IP data packets, the number of the received second IP data packets, the transmission time parameter, the time interval average value, the time interval number and the preset adjustment coefficients, calculating by adopting a preset data transmission quality function to obtain a real-time data transmission quality value;
Calculating and processing by adopting a preset abnormal index function based on the BRAM utilization rate abnormal score on the FPGA chip, the cache utilization rate abnormal score on the FPGA chip, the resource utilization rate abnormal score, the real-time data transmission quality value and each preset abnormal index proportionality coefficient to obtain an abnormal index value;
And under the condition that the abnormal index value is greater than or equal to a preset abnormal index threshold value, generating a reference IP data packet and reference information, and writing the reference information into a data matrix of the FPGA to process the IP data packet again.
8. An FPGA-based ultra-high definition JPEG-XS IP packet processing apparatus implementing the method of claim 1, comprising:
And (5) a decapsulation module: the method comprises the steps of performing decapsulation processing on first IP data packets of 8K and 4K ultrahigh-definition JPEG-XS obtained through pretreatment to obtain second IP data packets of 8K4K ultrahigh-definition JPEG-XS corresponding to the first IP data packets;
the calculation module: the flow control reference coefficient is used for calculating based on each second IP data packet in the second IP data packet flow to obtain a flow control reference coefficient corresponding to the second IP data packet flow;
model training module: the method comprises the steps of performing model training on an initial flow control strategy matching model based on the flow control reference coefficient, each second IP data packet and each historical data packet to generate a target flow control strategy matching model;
And a strategy matching module: the target flow control strategy matching module is used for carrying out strategy matching by utilizing the target flow control strategy matching model based on each second IP data packet to obtain a target flow control strategy for each second IP data packet;
And the distribution processing module is used for: and the method is used for carrying out distribution processing on each second IP data packet by adopting the target flow control strategy to obtain an IP data packet processing result of 8K4K ultra-high definition JPEG-XS.
9. A storage medium storing a computer program which when executed by a processor performs the steps of the FPGA-based ultra-high definition JPEG-XS IP packet processing method according to any of the preceding claims 1-7.
10. An electronic device, wherein the FPGA on-chip resources of the electronic device at least include an on-chip high-speed interface, an on-chip processor, a partition cache, and an on-chip memory, where the on-chip high-speed interface is used to receive data, read and write data, and transmit the data in parallel to each data processing module, the partition cache is used to divide and form a data array to generate data reference information, the on-chip memory stores a VHDL instruction program, and the on-chip processor directly runs the VHDL instruction program to perform computation and processing, so as to implement the steps of the method for processing IP packets of the FPGA-based ultra-high definition JPEG-XS according to any one of claims 1 to 7.
11. An FPGA-based ultra-high definition JPEG-XS IP packet processing system, comprising:
One or more on-chip processors;
on-chip storage means for storing one or more programs;
The one or more programs, when executed by the one or more processors, cause the one or more processors to implement the FPGA-based ultra-high definition JPEG-XS IP packet processing method of any of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410465671.9A CN118075207B (en) | 2024-04-18 | 2024-04-18 | FPGA-based ultra-high definition JPEG-XS IP data packet processing method, device and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410465671.9A CN118075207B (en) | 2024-04-18 | 2024-04-18 | FPGA-based ultra-high definition JPEG-XS IP data packet processing method, device and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN118075207A CN118075207A (en) | 2024-05-24 |
CN118075207B true CN118075207B (en) | 2024-06-28 |
Family
ID=91104261
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410465671.9A Active CN118075207B (en) | 2024-04-18 | 2024-04-18 | FPGA-based ultra-high definition JPEG-XS IP data packet processing method, device and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN118075207B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101873632A (en) * | 2009-04-27 | 2010-10-27 | 大唐移动通信设备有限公司 | Data dispatching method, system and device |
CN106603889A (en) * | 2017-02-08 | 2017-04-26 | 广州波视信息科技股份有限公司 | Ultra high-definition VR solid-state delayer based on FPGA chip |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11223870B2 (en) * | 2019-01-25 | 2022-01-11 | Shenzhen Lenkeng Technology Co., Ltd. | Method and device of transmitting and receiving ultra high definition video |
CN113473060B (en) * | 2021-07-02 | 2023-12-15 | 北京格非科技股份有限公司 | 4K ultra-high definition non-compression IP signal processor and processing method |
CN115278210A (en) * | 2022-06-21 | 2022-11-01 | 青岛大学 | High-integration-level ultra-high definition HDMI (high definition multimedia interface) video test signal source system and method |
-
2024
- 2024-04-18 CN CN202410465671.9A patent/CN118075207B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101873632A (en) * | 2009-04-27 | 2010-10-27 | 大唐移动通信设备有限公司 | Data dispatching method, system and device |
CN106603889A (en) * | 2017-02-08 | 2017-04-26 | 广州波视信息科技股份有限公司 | Ultra high-definition VR solid-state delayer based on FPGA chip |
Also Published As
Publication number | Publication date |
---|---|
CN118075207A (en) | 2024-05-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113348645B (en) | System and method for classifying data streams | |
CN109787833B (en) | Network abnormal event sensing method and system | |
JP5270901B2 (en) | Network test apparatus and method | |
CN112532409B (en) | Network parameter configuration method, device, computer equipment and storage medium | |
US8582466B2 (en) | Flow statistics aggregation | |
US10091675B2 (en) | System and method for estimating an effective bandwidth | |
US11392317B2 (en) | High speed data packet flow processing | |
TWI566551B (en) | Packet output processing | |
US12047295B2 (en) | Microburst detection and management | |
CN110505112B (en) | Network performance monitoring method, device and storage medium | |
CN108183917A (en) | DDoS attack cross-layer cooperative detection method based on software defined network | |
CN113472697A (en) | Network information transmission system | |
CN108259364A (en) | A kind of network congestion determines method and device | |
CN113422690A (en) | Service quality degradation prediction method and system | |
CN115037700A (en) | Complex network data packet transmission method, system, terminal and storage medium | |
CN113300916B (en) | Data center network congestion detection system and method based on stream watermarking | |
CN114513467A (en) | Network traffic load balancing method and device of data center | |
CN118075207B (en) | FPGA-based ultra-high definition JPEG-XS IP data packet processing method, device and system | |
CN101984635B (en) | Method and system for flow identification of point to point (P2P) protocol | |
CN111385060B (en) | Data transmission method, device, equipment and storage medium | |
CN108696389B (en) | Network flow and protocol message analysis platform based on mass data | |
CN110297785A (en) | A kind of finance data flow control apparatus and flow control method based on FPGA | |
CN112565457A (en) | General data processing system for Internet of things | |
CN113194071B (en) | Method, system and medium for detecting DDoS (distributed denial of service) based on unsupervised deep learning in SDN (software defined network) | |
CN116614449A (en) | Application layer-oriented self-adaptive flow control device and method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |