CN111082992A - SDN network data packet identification method based on deep learning - Google Patents
SDN network data packet identification method based on deep learning Download PDFInfo
- Publication number
- CN111082992A CN111082992A CN201911338814.5A CN201911338814A CN111082992A CN 111082992 A CN111082992 A CN 111082992A CN 201911338814 A CN201911338814 A CN 201911338814A CN 111082992 A CN111082992 A CN 111082992A
- Authority
- CN
- China
- Prior art keywords
- data packet
- detection model
- data
- deep learning
- identification method
- 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.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/12—Discovery or management of network topologies
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/147—Network analysis or design for predicting network behaviour
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/02—Capturing of monitoring data
- H04L43/028—Capturing of monitoring data by filtering
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention provides a deep learning-based SDN network data packet identification method, which comprises a detection model training step and a data packet identification step by using a detection model, wherein the trained detection model is embedded into a controller, when the data packet is received, the data packet is disassembled, required information is extracted, and then an extracted feature list is transmitted into the trained model. And judging the output result. And the normal flow performs resource scheduling, and the flow which is not normal is judged to be an attack packet and then is discarded. The SDN network data packet identification method based on deep learning enables the control surface of the SDN network not to be in port-to-port transmission but to be in centralized intelligence, the whole network is seen from a higher level, and the network behavior is controlled through software. The zero configuration can be flexibly changed, and online real-time detection of the data packet is realized.
Description
Technical Field
The invention relates to the technical field of network security, in particular to a deep learning-based SDN network data packet identification method.
Background
The arrival of the 5G era brings about not only huge flow, excellent speed, excellent performance, but also uncountable data packets. This becomes particularly important for detection of hazardous data traffic. The availability of conventional network inspection packets is currently known in only two categories: ACL and firewall. ACL (Access Control Lists), an Access Control technique based on packet filtering, can filter the data packets on the interface according to the set conditions, and allow them to pass or drop. The access control list is widely applied to routers and three-layer switches, and by means of the access control list, the access of users to the network can be effectively controlled, so that the network security is guaranteed to the greatest extent. The firewall technology is a technology for protecting the security of user data and information by organically combining various software and hardware devices for security management and screening to help a computer network to construct a relatively isolated protection barrier between an internal network and an external network. The firewall technology has the functions of discovering and processing the problems of security risk, data transmission and the like which may exist during the operation of the computer network in time, wherein the processing measures comprise isolation and protection, and meanwhile, the firewall technology can record and detect various operations in the security of the computer network so as to ensure the operation security of the computer network, ensure the integrity of user data and information and provide better and safer computer network use experience for users.
However, both of the above mentioned packet detections have common disadvantages: the arrangement is troublesome and is not easy to change.
The traditional network is a distributed network from the beginning, and has no central control node, and all devices in the network learn the reachable information of the network by means of oral-oral transmission, and each device determines how to forward the reachable information, which directly results in no holistic concept and cannot regulate and control the flow from the perspective of the whole network. Because of the oral transmission, a language of everyone needs to be used, namely a network protocol. Each equipment supplier can not develop protocols at will, otherwise, different manufacturers can not develop protocols at will, and the network is still obstructed. Thus, a global organization, such as IETF, is born. The RFC is a law of network protocols, which is equivalent to international law, and each equipment provider follows international law practices, so that normal operation of the whole network world is basically guaranteed. The device providers are equivalent to different countries, like the reality, laws are continuously revised and supplemented to adapt to the development of the era, all countries continuously seize mountains which are more beneficial to the countries based on the consideration of own interests, the constraints of the laws which are dominant in other countries on the countries are avoided, and the traditional network continuously advances in the game. This development, constrained by common laws, limits the deployment rate of new services for network operators. Network operators are responsible for providing network access functions for users, the requirements of the users are very different, once an original basic network cannot meet new requirements, the requirements need to be raised to a protocol making and modifying layer, each equipment provider starts to provide various schemes to be determined by IETF, new laws are formed, the implementation is realized by the equipment providers, and all equipment in the basic network is upgraded again to support the deployment of new services. This process is thought to be lengthy, typically over a period of at least half a year, or even years.
Therefore, there is a need for an improvement over existing cumbersome-to-configure packet detection methods.
Disclosure of Invention
In view of this, in order to solve the problems in the prior art, the invention provides a method for identifying an SDN network data packet based on deep learning, which can flexibly change zero configuration and can realize online real-time detection of the data packet. The online extraction of the data packet can be realized based on the SDN, the extracted data packet is transmitted into a deep learning model for detection, and processing is carried out according to the detection result.
The purpose of the invention is realized by the following technical scheme:
the SDN network data packet identification method based on deep learning comprises a detection model training step and a data packet identification step by using the detection model, wherein the trained detection model is embedded into a controller, when the data packet is received, the data packet is decoded, the characteristics of the data packet are extracted, then an extracted characteristic list is transmitted into the detection model, the detection model is used for identification, the normal data packet is used for resource scheduling, and the abnormal data packet is discarded.
Further, the step of training the detection model includes:
s101, collecting training data;
step S102, preprocessing data;
step S103, extracting the characteristics of the preprocessed data;
s104, inputting the extracted features into a CNN algorithm, training and storing a detection model;
and step S105, embedding the test model into the controller.
Further, the step of identifying the data packet by using the detection model comprises:
step S201, receiving a data packet;
step S202, decoding the data packet and extracting required fields;
step S203, the extracted fields are transmitted into a detection model, the detection model is facilitated to judge whether the data packet is a normal data packet, if the data packet is the normal data packet, the step S204 is switched to, and if the data packet is not the normal data packet, the step S205 is switched to;
step S204, a flow table is issued, and the controller guides the forwarding of the data packet;
step S205, discarding the abnormal data packet.
Further, in step S101, a Wireshark packet capturing tool is used to capture packets and collect training data.
Further, in step S103, the characteristics include, but are not limited to, a source IP, a destination IP, a source port, and a destination port.
Further, in step S104, the CNN algorithm is based on the formula: (N +2P-F)/S +1, wherein:
n refers to the size of the feature;
p refers to the size of the padding data;
f refers to the size of the convolution kernel;
s refers to the step size.
Furthermore, the detection model comprises two layers, wherein each layer comprises a convolution layer and a pooling layer, and finally, the convolution layer and the pooling layer are subjected to an average pooling layer and a full connection layer.
The SDN network data packet identification method based on deep learning enables the control surface of the SDN network not to be in port-to-port transmission but to be in centralized intelligence, the whole network is seen from a higher level, and the network behavior is controlled through software. The zero configuration can be flexibly changed, and online real-time detection of the data packet is realized.
Drawings
Figure 1 is a schematic diagram of an SDN network topology;
FIG. 2 is a flowchart of a training procedure for the detection model of the present invention;
FIG. 3 is a flow chart of the present invention for facilitating identification of a data packet by a detection model.
DETAILED DESCRIPTION OF EMBODIMENT (S) OF INVENTION
The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.
The embodiments of the present disclosure are described below with specific examples, and other advantages and effects of the present disclosure will be readily apparent to those skilled in the art from the disclosure in the specification. It is to be understood that the described embodiments are merely illustrative of some, and not restrictive, of the embodiments of the disclosure. The disclosure may be embodied or carried out in various other specific embodiments, and various modifications and changes may be made in the details within the description without departing from the spirit of the disclosure. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict. All other embodiments, which can be derived by a person skilled in the art from the embodiments disclosed herein without making any creative effort, shall fall within the protection scope of the present disclosure.
Example one
Current SDN networks are generally of the form: the system comprises a controller (or a cluster) which is responsible for collecting information such as topology, flow and the like of the whole network, calculating a flow forwarding path, issuing a forwarding table item to a switch through an OpenFlow protocol, and executing forwarding action according to the table item by the switch. The switch that performs the forwarding action corresponding to the controller is generally referred to as a repeater. The control plane is stripped from the single devices of the traditional network and is concentrated on the controller, and the forwarding plane is formed by repeaters.
The control plane of the SDN network is not communicated with the interface, but is centralized and intelligent, the whole network is seen from a higher level, and the network behavior is controlled through software. The software here is no longer provided by the device vendor bound device, but the user can implement it himself. The only standardization of the SDN is that a communication interface between a controller and a repeater is OpenFlow, the communication interface is not a network control protocol in the traditional sense, and only one interface is needed, and the controller and the repeater are realized according to the interface. SDN opens a door to curiosity.
The rapid development of artificial intelligence enables a new mode to be provided for detecting a large number of data packets, people can learn through an artificial intelligence algorithm, train out a model, then input real data into the model for abnormal data detection, and give corresponding alarms. Therefore, the data packets in the network can be monitored anytime and anywhere, and harmful data packets can be filtered out.
As shown in fig. 1, which is a simple sdn network topology, the controller of sdn is at the top, we embed our trained detection model in the controller of sdn, the switch of sdn is in the middle, and the hosts connected to the sdn switch are on the left and right sides, respectively, at this time, the controller will detect the data flow in the network in real time. At this time, the data packet can be received in the controller, the data packet is disassembled, the required information (characteristics) is extracted, and then the extracted characteristic list is transmitted into the trained model. And judging the output result. And the normal flow performs resource scheduling, and the flow which is not normal is judged to be an attack packet and then is discarded. And finally, a GUI (graphical user interface) is made for displaying the real-time data monitoring condition.
Such as: if the prediction result is an illegal data packet, the sdn controller is used for issuing a corresponding flow table, the data packet of the type is discarded or stored as log, so that the data packet can be conveniently checked and analyzed later, and if the data packet is predicted to be a normal data packet, the data packet is normally forwarded.
The SDN network data packet identification method based on deep learning comprises a step of training a detection model and a step of identifying a data packet by using the detection model.
The step of training the detection model is shown in fig. 2, and includes:
and step S101, collecting training data.
And (3) grabbing the packet by using a packet grabbing tool to acquire a larger data volume as much as possible, so that the training is facilitated, grabbing the packet by using Wireshark, and storing the collected data. Wireshark (formerly Ethereal) is a piece of network packet analysis software. The function of the network packet analysis software is to capture the network packets and display the most detailed network packet data as possible.
And step S102, preprocessing the data.
And processing the collected data, deleting some error data, and removing repeated data, wherein the data can influence the training result.
Since the data collected by using the Wireshark packet-grabbing tool is stored in the form of text, useful data in the text must be extracted, and useful information is the field of each layer of protocol.
And step S103, performing feature extraction on the preprocessed data.
The collected data are all data of some text types, cannot be directly input into an artificial intelligence algorithm, and needs to be subjected to feature extraction.
And the IP layer extracts a source IP, a destination IP and a protocol used by an upper layer according to the hierarchy display IP layer.
The Tcp layer extracts zone bit information such as source port, destination port, fin, syn, rest, push, ack, urg and the like.
The Udp layer only extracts the source port and the destination port.
The statistical data was processed by Numpy (a python implemented scientific computational library to store and process large matrices) and pandas (Numpy based tools created to solve data analysis) to become model processable data. Each preprocessed field is a characteristic of the model, such as a source IP and a destination IP, which are two characteristics of the model, and the fields extracted from each data packet form a line which is a sample of the model, and a line of data is a sample.
In the field of deep learning, feature extraction does not need to be completed manually, a deep learning algorithm can help us to complete automatically, and in the field of machine learning, a large amount of domain knowledge is needed in the process of feature extraction, manual summary is carried out, and then extracted rules are converted into digital features to be transmitted into the algorithm.
And S104, inputting the extracted features into a CNN algorithm, and training and storing the detection model.
Inputting the characteristics into a CNN algorithm (convolutional neural network), training results, continuously adjusting parameters to obtain an optimal result, wherein the common parameter adjustment method comprises cross validation and grid search, the model needs to be evaluated after the model is trained, the accuracy, precision, recall rate and the like of the model are mainly considered, and the model is stored after the model is available for evaluation.
The charm of deep learning is that the training process does not need manual intervention, only processed data need to be transmitted into an algorithm, and the algorithm can automatically extract features in the data for training.
CNN algorithm: since the data is in text form, the data is processed with a one-dimensional convolution of the CNN. According to the formula: (N +2P-F)/S + 1.
Wherein:
n refers to the size of the latitude of the input data, here is the size of the characteristic, and there are 16 characteristics in this text;
p is the size of the padding data, and is 0 by default;
f refers to the size of the convolution kernel;
s refers to the step size, and is 1 by default.
For example: one network architecture may be:
data 16161 is input, where N is 16.
The first layer of convolution: f is 2, 32 filters, S is 2, P is 0, and the output size is 8832.
Second layer convolution: f is 2, 64 filters, S is 2, P is 0, and the output size is 4464.
And a third layer of convolution: f is 2, 128 filters, S is 2, P is 0, and the output size is 22128.
And a fourth layer of convolution: f is 2, 512 filters, S is 2, P is 0, the output size is 11512.
And outputs bit 512. The final result can then be output via a full connection.
The model can be built most quickly by using the method, and the effect is ideal.
The inspection model uses two layers, each layer comprising a convolutional layer and a pooling layer. And finally, passing through an average pooling layer and a full-connection layer.
An algorithm tuning mode is as follows:
(1) a linear learning rate decay strategy (decreasing learning rate parameters) is used.
(2) The sum of the average and maximum pooling layers is used.
(3) A mini-batch size of approximately 128(0.005) to 256(0.01) is used. If this is too large for the GPU, it is sufficient to scale down the learning rate to this size.
(4) The convolutional layer is used instead of the linear layer in the previous MLP and is predicted with the average pooling layer.
(5) When studying to increase the training set size, it is necessary to determine the balance point of the data set to the performance improvement.
(6) The quality of the data is more important than the data size.
And step S105, embedding the test model into the controller.
The detection model is successfully embedded into the controller, so that real-time processing is realized, and the step is very important. Without this step, real-time processing of the data packets cannot be achieved even if the trained detection model is good. Then the detection of the packet can only be done locally and this will be of no consequence. Each data packet is transmitted through the controller, and the controller analyzes the data packet to extract the characteristic information of the data packet. The data is pre-processed and then placed into a model for prediction. And after a prediction result is obtained, performing corresponding operation on the prediction result.
The step of identifying the data packet using the detection model includes:
step S201, receiving a data packet.
Step S202, decoding the data packet, and extracting required fields, such as source IP, destination IP, source port, destination port, and the like.
Step S203, the extracted fields are transmitted into a detection model, the detection model judges whether the data packet is a normal data packet, if the data packet is the normal data packet, the step S204 is carried out, and if the data packet is not the normal data packet, the step S205 is carried out.
And step S204, issuing a flow table, and guiding the forwarding of the data packet by the controller.
Step S205, discarding the abnormal data packet.
The above description is for the purpose of illustrating embodiments of the invention and is not intended to limit the invention, and it will be apparent to those skilled in the art that any modification, equivalent replacement, or improvement made without departing from the spirit and principle of the invention shall fall within the protection scope of the invention.
Claims (7)
1. The SDN network data packet identification method based on deep learning is characterized by comprising the following steps: the method comprises the steps of training a detection model and identifying a data packet by using the detection model, wherein the trained detection model is embedded into a controller, when the data packet is received, the data packet is decoded, the characteristics of the data packet are extracted, then the extracted characteristic list is transmitted into the detection model, the detection model is used for identification, a normal data packet is used for resource scheduling, and an abnormal data packet is discarded.
2. The deep learning based SDN network packet identification method of claim 1, wherein:
the step of training the detection model comprises:
s101, collecting training data;
step S102, preprocessing data;
step S103, extracting the characteristics of the preprocessed data;
s104, inputting the extracted features into a CNN algorithm, training and storing a detection model;
and step S105, embedding the test model into the controller.
3. The deep learning based SDN network packet identification method of claim 1, wherein:
the step of identifying the data packet using the detection model includes:
step S201, receiving a data packet;
step S202, decoding the data packet and extracting required fields;
step S203, the extracted fields are transmitted into a detection model, the detection model is facilitated to judge whether the data packet is a normal data packet, if the data packet is the normal data packet, the step S204 is switched to, and if the data packet is not the normal data packet, the step S205 is switched to;
step S204, a flow table is issued, and the controller guides the forwarding of the data packet;
step S205, discarding the abnormal data packet.
4. The deep learning based SDN network packet identification method of claim 1, wherein: in the step S101, a Wireshark packet capturing tool is used to capture packets and collect training data.
5. The deep learning based SDN network packet identification method of claim 1, wherein: in step S103, the characteristics include, but are not limited to, a source IP, a destination IP, a source port, and a destination port.
6. The deep learning based SDN network packet identification method of claim 1, wherein: in step S104, the CNN algorithm is based on the following formula: (N +2P-F)/S +1, wherein:
n refers to the size of the feature;
p refers to the size of the padding data;
f refers to the size of the convolution kernel;
s refers to the step size.
7. The deep learning based SDN network packet identification method of claim 1, wherein: the detection model comprises two layers, wherein each layer comprises a convolution layer and a pooling layer, and finally passes through an average pooling layer and a full-connection layer.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911338814.5A CN111082992A (en) | 2019-12-23 | 2019-12-23 | SDN network data packet identification method based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911338814.5A CN111082992A (en) | 2019-12-23 | 2019-12-23 | SDN network data packet identification method based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111082992A true CN111082992A (en) | 2020-04-28 |
Family
ID=70316847
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911338814.5A Pending CN111082992A (en) | 2019-12-23 | 2019-12-23 | SDN network data packet identification method based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111082992A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111935063A (en) * | 2020-05-28 | 2020-11-13 | 国网电力科学研究院有限公司 | System and method for monitoring abnormal network access behavior of terminal equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108123931A (en) * | 2017-11-29 | 2018-06-05 | 浙江工商大学 | Ddos attack defence installation and method in a kind of software defined network |
CN108921198A (en) * | 2018-06-08 | 2018-11-30 | 山东师范大学 | commodity image classification method, server and system based on deep learning |
CN109768981A (en) * | 2019-01-20 | 2019-05-17 | 北京工业大学 | A kind of network attack defence method and system under SDN framework based on machine learning |
CN109981691A (en) * | 2019-04-30 | 2019-07-05 | 山东工商学院 | A kind of real-time ddos attack detection system and method towards SDN controller |
CN110224990A (en) * | 2019-07-17 | 2019-09-10 | 浙江大学 | A kind of intruding detection system based on software definition security architecture |
-
2019
- 2019-12-23 CN CN201911338814.5A patent/CN111082992A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108123931A (en) * | 2017-11-29 | 2018-06-05 | 浙江工商大学 | Ddos attack defence installation and method in a kind of software defined network |
CN108921198A (en) * | 2018-06-08 | 2018-11-30 | 山东师范大学 | commodity image classification method, server and system based on deep learning |
CN109768981A (en) * | 2019-01-20 | 2019-05-17 | 北京工业大学 | A kind of network attack defence method and system under SDN framework based on machine learning |
CN109981691A (en) * | 2019-04-30 | 2019-07-05 | 山东工商学院 | A kind of real-time ddos attack detection system and method towards SDN controller |
CN110224990A (en) * | 2019-07-17 | 2019-09-10 | 浙江大学 | A kind of intruding detection system based on software definition security architecture |
Non-Patent Citations (2)
Title |
---|
李传煌等: "SDN下基于深度学习混合模型的DDoS攻击检测与防御", 《通信学报》 * |
杨垠坦: "基于卷积神经网络的 SDN 入侵检测技术研究", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111935063A (en) * | 2020-05-28 | 2020-11-13 | 国网电力科学研究院有限公司 | System and method for monitoring abnormal network access behavior of terminal equipment |
CN111935063B (en) * | 2020-05-28 | 2023-11-21 | 国网电力科学研究院有限公司 | Abnormal network access behavior monitoring system and method for terminal equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102420701B (en) | Method for extracting internet service flow characteristics | |
Holland et al. | New directions in automated traffic analysis | |
CN106464577B (en) | Network system, control device, communication device and communication control method | |
CN104506484B (en) | A kind of proprietary protocol analysis and recognition methods | |
Braga et al. | Lightweight DDoS flooding attack detection using NOX/OpenFlow | |
CN101714952B (en) | Method and device for identifying traffic of access network | |
CN102271090B (en) | Transport-layer-characteristic-based traffic classification method and device | |
CN108282497A (en) | For the ddos attack detection method of SDN control planes | |
CN107426059B (en) | DPI equipment feature library automatic updating method and system, DPI equipment and cloud server | |
CN102315974A (en) | Stratification characteristic analysis-based method and apparatus thereof for on-line identification for TCP, UDP flows | |
CN110401624A (en) | The detection method and system of source net G system mutual message exception | |
CN106797328A (en) | Collect and analyze selected network traffics | |
CN111935063B (en) | Abnormal network access behavior monitoring system and method for terminal equipment | |
CN109274673A (en) | A kind of detection of exception of network traffic and defence method | |
CN109379341A (en) | A kind of Recall remote control Trojan network flow detection method of Behavior-based control analysis | |
CN109271793A (en) | Internet of Things cloud platform device class recognition methods and system | |
CN103997439A (en) | Flow monitoring method, device and system | |
CN111294342A (en) | Method and system for detecting DDos attack in software defined network | |
CN108011865A (en) | SDN flow paths method for tracing, apparatus and system based on flowing water print and stochastical sampling | |
CN104917628A (en) | Automatic diagnosis method of Ethernet router/switch packet loss fault | |
CN113382039B (en) | Application identification method and system based on 5G mobile network flow analysis | |
CN111953504B (en) | Abnormal flow detection method and device, and computer readable storage medium | |
CN107707549A (en) | A kind of device and method automatically extracted using feature | |
CN106131153B (en) | Business recognition method and device based on intelligent gateway | |
CN111082992A (en) | SDN network data packet identification method based on deep learning |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200428 |