CN110601764A - Radio frequency modulation format identification method based on optical assistance - Google Patents
Radio frequency modulation format identification method based on optical assistance Download PDFInfo
- Publication number
- CN110601764A CN110601764A CN201910868856.3A CN201910868856A CN110601764A CN 110601764 A CN110601764 A CN 110601764A CN 201910868856 A CN201910868856 A CN 201910868856A CN 110601764 A CN110601764 A CN 110601764A
- Authority
- CN
- China
- Prior art keywords
- radio frequency
- optical
- hidden layer
- frequency modulation
- signals
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/25—Arrangements specific to fibre transmission
- H04B10/2575—Radio-over-fibre, e.g. radio frequency signal modulated onto an optical carrier
Landscapes
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Digital Transmission Methods That Use Modulated Carrier Waves (AREA)
Abstract
The invention discloses a radio frequency modulation format identification method based on optical assistance, which comprises the following steps: the method comprises the steps that five common radio frequency modulation formats are modulated onto an optical carrier to achieve light intensity modulation, modulated optical signals are divided into two paths of completely identical signals through an optical coupler, only one path of the two paths of signals is subjected to optical time delay processing, the two paths of optical signals are input into a polarization beam combiner, the two paths of optical signals are orthogonal, and radio frequency signals are recovered through a low-speed photoelectric detector, so that the characteristic extraction of the radio frequency signals is achieved; sampling the recovered radio frequency signals by using an oscilloscope, preprocessing the sampled data into a statistical histogram, inputting the statistical histogram into a neural network, and finally identifying and classifying the five common modulation formats through the neural network. On the basis of identifying five radio frequency modulation formats with high accuracy, the method greatly improves the identification accuracy of common radio frequency modulation formats under the condition of low signal to noise ratio, and enriches the methods for identifying the radio frequency modulation formats.
Description
Technical Field
The invention relates to the fields of optical fiber communication, radio frequency photonics and electronic countermeasure, in particular to a radio frequency signal identification and classification technology. In particular to a radio frequency modulation format identification method based on optical assistance.
Background
Modulation format identification (MFR) is a process of automatically identifying the modulation type of an unknown intercepted signal, ensuring that the received signal can be correctly demodulated and accurately recovered. MFR is an important area of research in communication systems, and has found widespread use both in civilian and military applications. The signal monitoring, signal interception and signal identification in military electronic countermeasure and the parallel transmission of a large amount of broadband data in civil use can not be distinguished from modulation format. At present, wireless communication environments are various, modulation formats are different, and how to correctly identify modulation types of communication signals in different occasions and different wireless communication environments is an important research topic for military and civil use at present.
Methods of MFR fall into two main categories: likelihood function (LB) based methods and Feature (FB) based methods. The LB method achieves MFR by using probabilities and hypothesis test parameters, requiring correct hypotheses and setting appropriate threshold values with minimal probability of misidentification, providing an optimal solution in bayesian sense. However, the LB method involves a high computational complexity and is difficult to implement. In the FB method, in contrast, it is necessary to extract specific features from the signal and then classify the signal by using a classification method. Although the recognition result of FB is not as good as that of LB, FB has a definite structure, low complexity and is easy to realize. The reasonably designed FB method can infinitely approximate the recognition result of the LB method. In recent years, deep learning has been rapidly developed, and a combination of MFR and deep learning has come to work. The combination scheme mainly takes the characteristics of the signal to be detected as the input of the neural network, and realizes the classification of different modulation formats by utilizing a deep learning tool.
Currently, the MFR scheme using deep learning is mainly implemented under the condition of high signal-to-noise ratio, and the recognition accuracy deteriorates rapidly with low signal-to-noise ratio. However, in electronic countermeasure, the signal under test is often buried in noise, and the common solution is that the identification accuracy of these low signal-to-noise ratio signal under test cannot be maintained high.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an optical assistance-based radio frequency modulation format identification method.
The invention relates to an optical assistance-based radio frequency modulation format identification method, which comprises the following steps:
step 1: five common radio frequency modulation formats are generated, and radio frequency signals are modulated onto optical carriers by using an electro-optical modulator.
Step 2: the modulated optical signal is split into two identical optical signals using an optical coupler.
And step 3: any one path of the two paths of optical signals is input into the optical time delay device, and the other path of the two paths of optical signals is not subjected to time delay processing.
And 4, step 4: and combining the two paths of optical signals together by using a polarization beam combiner, and recovering a radio frequency signal by using a photoelectric detector, thereby realizing the feature extraction.
And 5: and sampling the recovered radio frequency signal by using an oscilloscope.
Step 6: and carrying out data preprocessing on the sampling signal to generate input data suitable for the neural network.
And 7: five common radio frequency modulation formats are identified and classified by using a neural network.
Further, step 6 specifically includes: firstly, performing Hilbert transform on radio frequency signals sampled by an oscilloscope, then performing smooth filtering, then performing power normalization processing, and finally generating a statistical histogram as the input of a neural network.
Further, step 7 adopts a deep neural network DNN, and the building steps are as follows:
(1) firstly, a back propagation neural network BP only containing a hidden layer is built, and the number of nodes of an input layer and the number of nodes of an output layer are determined.
(2) And initializing a weight and a bias, wherein the value range of the weight initialization is uniform distribution values of [ -0.05, 0.05], and the bias initialization is all 0.
(3) Determining a hidden layer activation function, an output layer activation function, an optimizer and an error calculation function: in DNN, a ReLU function is adopted as a hidden layer activation function, a Softmax function is adopted as an output layer activation function, an Adam algorithm is adopted by an optimizer, and a cross binary entropy is adopted as an error calculation function.
(4) Determining the number of hidden layer nodes: the hidden layer node number value range is a value between an output neuron and an input neuron, all values in the range are sequentially used as hidden layer node numbers, and the values with less node numbers and high identification accuracy are used as the hidden layer node numbers of the network by comparing the identification accuracy of the final training data.
(5) Determining the number of hidden layers in the DNN: a basic BP frame is built through the previous steps, an Epoch value used for observing the recognition accuracy of the training data and the obvious change of a loss function is set, the number of layers of the hidden layer is gradually increased, the number of nodes of the hidden layer of each layer is the same as that of the nodes of the first hidden layer, and the number of layers of the hidden layer with few layer values and high recognition accuracy of the training data is determined by observing the change conditions of the recognition accuracy of the training data and the loss function.
(6) Determination of the Epoch value in DNN: setting an Epoch value for the DNN in advance, observing a loss function curve of the training data and a loss function curve of the verification data, and taking the Epoch value of the turning point as a final Epoch value of the DNN when the trend of the loss curve of the verification data is not close to zero any more but begins to increase gradually.
Compared with the prior art, the invention has the beneficial technical effects that:
the invention greatly improves the identification accuracy of the radio frequency signal under the condition of low signal-to-noise ratio while realizing the high identification accuracy of the common radio frequency modulation format under the condition of high signal-to-noise ratio.
Drawings
Fig. 1 is a block diagram of an implementation of the method for identifying a radio frequency modulation format based on optical assist according to the present invention.
FIG. 2 is a flow chart of data preprocessing.
Fig. 3 is a histogram for five modulation formats with a step size of 5 dB.
Fig. 4 is a diagram of a deep neural network architecture.
FIG. 5 is a diagram of a confusion matrix generated by a deep neural network.
Fig. 6 is a graph of recognition accuracy of six neural networks under different signal-to-noise ratios.
Detailed Description
The invention is described in further detail below with reference to the figures and the detailed description.
The invention provides an optical assistance-based radio frequency modulation format identification method, as shown in fig. 1, comprising the following steps:
step 1: five common radio frequency modulation formats are generated, and simultaneously, Gaussian white noise is added into the radio frequency signals, so that the signal-to-noise ratio adjusting range of the signals is controlled. Modulating the radio frequency signal to an optical carrier by using an electro-optical modulator in an intensity modulation mode to obtain an optical signal EIM(t)：
Wherein j is an imaginary unit, a1To the first order sideband amplitude, omegamFor the frequency of the radio-frequency signal, omegacE is a carrier frequency and is a constant.
Step 2: the modulated optical signal is split into two identical optical signals using a 3dB optical coupler.
And step 3: any one path of the two optical signals is input into the optical time delay device, and the other path is not subjected to time delay processing, so that the envelope of the two optical signals has a half-code element difference, namely
Wherein τ is relative time delay, Γ (τ) is time delay formula, Γ (τ) EIM(t)＝EIM(t-τ)。
And 4, step 4: the two paths of optical signals are combined together by utilizing a polarization beam combiner, and radio frequency signals are recovered by a photoelectric detector:
I1＝|EIM(t)|2＝|1+2a1 cos(ωmt)|2＝(2a1 2+1)+4a1 cos(ωmt) (3)
I2＝|EIM(t-τ)|2＝|1+2a1 cos[ωm(t-τ)]|2＝(2a1 2+1)+4a1 cos[ωm(t-τ)] (4)
wherein, I1For non-delayed signals, I2For the time-delayed path signal, θ is the combined phase of the two paths and θ is arctan (cot (ω)mτ/2))。
And 5: and sampling the recovered radio frequency signal by using an oscilloscope.
Step 6: the sampling signal is subjected to data preprocessing to generate input data suitable for the neural network, as shown in fig. 2, the processing procedure is as follows: firstly, performing Hilbert transform on radio frequency signals sampled by an oscilloscope, then performing smooth filtering, then performing power normalization processing, and finally generating a statistical histogram as the input of a neural network. The generated statistical histogram is shown in fig. 3, and it can be seen that the shape features of the five modulation formats are different greatly under the condition of high signal-to-noise ratio, and are similar under the condition of low signal-to-noise ratio, which is also a main reason why the different modulation formats are difficult to distinguish under the condition of low signal-to-noise ratio.
And 7: and taking the ordinate of all the statistical histograms as the input of the neural network, and taking the abscissa as the number of the neurons of the input layer. The collected data is divided into three parts: the data processing system comprises a training data set, a verification data set and a testing data set, wherein the training data set accounts for 80% of total data, the rest 20% of the total data are used as the testing data set, and half of the testing data set is used as the verification data set.
The neural network of the invention adopts a deep neural network DNN, the DNN is shown as figure 4, and the construction steps are as follows:
(1) firstly, a back propagation neural network BP only containing a hidden layer is built, and the number of nodes of an input layer and the number of nodes of an output layer are determined.
(2) And initializing a weight and a bias, wherein the value range of the weight initialization is uniform distribution values of [ -0.05, 0.05], and the bias initialization is all 0.
(3) Determining a hidden layer activation function, an output layer activation function, an optimizer and an error calculation function: in DNN, a ReLU function is adopted as a hidden layer activation function, a Softmax function is adopted as an output layer activation function, an Adam algorithm is adopted by an optimizer, and a cross binary entropy is adopted as an error calculation function:
ReLU: (x) max (x,0), where x is the neuron that inputs the hidden layer node,
Softmax：wherein xiK is the total number of neurons in the last hidden layer.
(4) The number of hidden layer nodes is determined. The hidden layer node number value range is a value between an output neuron and an input neuron, all values in the range are sequentially used as hidden layer node numbers for better determining the neural network, and the values with relatively few node numbers and the highest identification accuracy are used as the hidden layer node numbers of the network by comparing the identification accuracy of final training data.
(5) The number of hidden layer layers in the DNN is determined. A basic BP frame is built through the previous steps, an Epoch value used for observing the recognition accuracy of the training data and the obvious change of a loss function is set, the number of layers of the hidden layer is gradually increased, the number of nodes of the hidden layer of each layer is the same as that of the nodes of the first hidden layer, and the number of layers of the hidden layer with relatively few values and the highest recognition accuracy of the training data is determined by observing the change conditions of the recognition accuracy of the training data and the loss function.
(6) The Epoch value in the DNN is determined. And setting a relatively large Epoch value for the DNN in advance, observing a loss function curve of the training data and a loss function curve of the verification data, and taking the Epoch value of the turning point as a final Epoch value of the DNN when the trend of the loss function curve of the verification data is not close to zero any more but begins to increase gradually.
In order to verify the superiority of the scheme, six neural network models, namely BP, DNN, convolutional neural network CNN, long-time memory neural network LSTM, probabilistic neural network PNN and CNN + LSTM, are used for identifying and classifying five common radio frequency modulation formats.
The recognition accuracy of five radio frequency modulation formats using six neural networks in the range of-10 dB to 15dB in signal to noise ratio is shown in fig. 5. DNN, CNN, LSTM and CNN + LSTM still keep higher recognition accuracy under the condition of lower signal-to-noise ratio. Wherein the DNN has the optimal recognition performance, and the average recognition accuracy is 96.058%. The confusion matrix obtained by using DNN for five modulation formats is shown in fig. 6, and it can be clearly seen that the recognition rate of the common five modulation formats is higher than 95.481%.
Claims (3)
1. An optical assistance-based radio frequency modulation format identification method is characterized by comprising the following steps:
step 1: generating five common radio frequency modulation formats, and modulating a radio frequency signal onto an optical carrier by using an electro-optical modulator;
step 2: dividing the modulated optical signal into two identical optical signals by using an optical coupler;
and step 3: inputting any one path of the two paths of optical signals into an optical time delay device, and not performing time delay processing on the other path of the two paths of optical signals;
and 4, step 4: combining two paths of optical signals together by using a polarization beam combiner, and recovering a radio frequency signal by using a photoelectric detector so as to realize feature extraction;
and 5: sampling the restored radio frequency signal by using an oscilloscope;
step 6: carrying out data preprocessing on the sampling signal to generate input data suitable for a neural network;
and 7: five common radio frequency modulation formats are identified and classified by using a neural network.
2. The method according to claim 1, wherein the step 6 specifically comprises: firstly, performing Hilbert transform on radio frequency signals sampled by an oscilloscope, then performing smooth filtering, then performing power normalization processing, and finally generating a statistical histogram as the input of a neural network.
3. The method for identifying the radio frequency modulation format based on the optical assistance as claimed in claim 1, wherein a Deep Neural Network (DNN) is adopted in the step 7, and the construction steps are as follows:
(1) firstly, building a back propagation neural network BP only containing a layer of hidden layer, and determining the number of nodes of an input layer and the number of nodes of an output layer;
(2) initializing a weight and bias, wherein the value range of the weight initialization is a uniform distribution value of [ -0.05, 0.05], and the bias initialization is all 0;
(3) determining a hidden layer activation function, an output layer activation function, an optimizer and an error calculation function: in DNN, a ReLU function is adopted as a hidden layer activation function, a Softmax function is adopted as an output layer activation function, an Adam algorithm is adopted by an optimizer, and a cross binary entropy is adopted as an error calculation function;
(4) determining the number of hidden layer nodes: the hidden layer node number value range is a value between an output neuron and an input neuron, all values in the range are sequentially used as hidden layer node numbers, and the values with less node numbers and high identification accuracy are used as the hidden layer node numbers of the network by comparing the identification accuracy of the final training data;
(5) determining the number of hidden layers in the DNN: establishing a basic BP frame through the previous steps, setting an Epoch value for observing the recognition accuracy of the training data and the obvious change of a loss function, gradually increasing the number of layers of the hidden layer, ensuring that the number of the hidden layer nodes of each layer is the same as that of the first hidden layer node, and determining the number of the hidden layer with few layer values and high recognition accuracy of the training data by observing the change conditions of the recognition accuracy of the training data and the loss function;
(6) determination of the Epoch value in DNN: setting an Epoch value for the DNN in advance, observing a loss function curve of the training data and a loss function curve of the verification data, and taking the Epoch value of the turning point as a final Epoch value of the DNN when the trend of the loss curve of the verification data is not close to zero any more but begins to increase gradually.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910868856.3A CN110601764A (en) | 2019-09-16 | 2019-09-16 | Radio frequency modulation format identification method based on optical assistance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910868856.3A CN110601764A (en) | 2019-09-16 | 2019-09-16 | Radio frequency modulation format identification method based on optical assistance |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110601764A true CN110601764A (en) | 2019-12-20 |
Family
ID=68859647
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910868856.3A Pending CN110601764A (en) | 2019-09-16 | 2019-09-16 | Radio frequency modulation format identification method based on optical assistance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110601764A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111541484A (en) * | 2020-04-20 | 2020-08-14 | 苏州大学 | Optical fiber communication system optical signal-to-noise ratio monitoring method based on time delay sampling |
| CN111970050A (en) * | 2020-07-14 | 2020-11-20 | 电子科技大学 | System for jointly monitoring modulation format and optical signal-to-noise ratio based on anomaly detection |
| CN113285762A (en) * | 2021-02-25 | 2021-08-20 | 广西师范大学 | Modulation format identification method based on relative entropy calculation |
| CN113723650A (en) * | 2020-05-25 | 2021-11-30 | 中国石油化工股份有限公司 | Chemical process abnormity monitoring system based on semi-supervised model and model optimization device |
| CN114679232A (en) * | 2022-04-06 | 2022-06-28 | 西南交通大学 | Data-driven light-load wireless transmission link modeling method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016180410A1 (en) * | 2015-05-13 | 2016-11-17 | Ls Telcom Ag | Method and device for automatic classification of the modulation format of a digitally modulated signal |
| CN106779069A (en) * | 2016-12-08 | 2017-05-31 | 国家电网公司 | A kind of abnormal electricity consumption detection method based on neutral net |
| CN108427987A (en) * | 2018-03-08 | 2018-08-21 | 四川大学 | A kind of Modulation Mode Recognition method based on convolutional neural networks |
| CN108880692A (en) * | 2018-06-05 | 2018-11-23 | 西南交通大学 | A kind of identification of coherent optical communication system modulation format and optical signal to noise ratio monitoring method |
| CN109685240A (en) * | 2018-10-24 | 2019-04-26 | 国网浙江省电力有限公司 | A kind of prediction of 95598 traffic work orders and unusual fluctuation method for early warning based on LSTM deep learning |
| CN110048781A (en) * | 2019-04-17 | 2019-07-23 | 武汉邮电科学研究院有限公司 | A kind of recognition methods of optical signal modulation format and device |
-
2019
- 2019-09-16 CN CN201910868856.3A patent/CN110601764A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016180410A1 (en) * | 2015-05-13 | 2016-11-17 | Ls Telcom Ag | Method and device for automatic classification of the modulation format of a digitally modulated signal |
| CN106779069A (en) * | 2016-12-08 | 2017-05-31 | 国家电网公司 | A kind of abnormal electricity consumption detection method based on neutral net |
| CN108427987A (en) * | 2018-03-08 | 2018-08-21 | 四川大学 | A kind of Modulation Mode Recognition method based on convolutional neural networks |
| CN108880692A (en) * | 2018-06-05 | 2018-11-23 | 西南交通大学 | A kind of identification of coherent optical communication system modulation format and optical signal to noise ratio monitoring method |
| CN109685240A (en) * | 2018-10-24 | 2019-04-26 | 国网浙江省电力有限公司 | A kind of prediction of 95598 traffic work orders and unusual fluctuation method for early warning based on LSTM deep learning |
| CN110048781A (en) * | 2019-04-17 | 2019-07-23 | 武汉邮电科学研究院有限公司 | A kind of recognition methods of optical signal modulation format and device |
Non-Patent Citations (1)
| Title |
|---|
| JIA YE 等: "Photonic-Assisted Modulation Classification for RF signals Using Probabilistic Neural Network", 《2019 OPTICAL FIBER COMMUNICATIONS CONFERENCE AND EXHIBITION (OFC)》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111541484A (en) * | 2020-04-20 | 2020-08-14 | 苏州大学 | Optical fiber communication system optical signal-to-noise ratio monitoring method based on time delay sampling |
| CN113723650A (en) * | 2020-05-25 | 2021-11-30 | 中国石油化工股份有限公司 | Chemical process abnormity monitoring system based on semi-supervised model and model optimization device |
| CN111970050A (en) * | 2020-07-14 | 2020-11-20 | 电子科技大学 | System for jointly monitoring modulation format and optical signal-to-noise ratio based on anomaly detection |
| CN111970050B (en) * | 2020-07-14 | 2022-03-15 | 电子科技大学 | System for jointly monitoring modulation format and optical signal-to-noise ratio based on anomaly detection |
| CN113285762A (en) * | 2021-02-25 | 2021-08-20 | 广西师范大学 | Modulation format identification method based on relative entropy calculation |
| CN113285762B (en) * | 2021-02-25 | 2022-08-05 | 广西师范大学 | Modulation format identification method based on relative entropy calculation |
| CN114679232A (en) * | 2022-04-06 | 2022-06-28 | 西南交通大学 | Data-driven light-load wireless transmission link modeling method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110601764A (en) | Radio frequency modulation format identification method based on optical assistance | |
| CN109802905B (en) | CNN convolutional neural network-based digital signal automatic modulation identification method | |
| CN112702294B (en) | Modulation recognition method for multi-level feature extraction based on deep learning | |
| CN113435247B (en) | Intelligent recognition method, system and terminal for communication interference | |
| CN112818891B (en) | Intelligent identification method for communication interference signal type | |
| CN111970050B (en) | System for jointly monitoring modulation format and optical signal-to-noise ratio based on anomaly detection | |
| WO2016082562A1 (en) | Signal identification method and system based on radio signal spectrum feature template | |
| CN109309640B (en) | Blind signal format identification method based on machine learning | |
| CN116866129A (en) | Wireless communication signal detection method | |
| CN109587091B (en) | Modulation format identification method of coherent optical communication system based on logistic regression algorithm | |
| Zhou et al. | A novel radio frequency fingerprint identification method using incremental learning | |
| CN113542172B (en) | Elastic optical network modulation format identification method and system based on improved PSO clustering | |
| CN114925720A (en) | Small sample modulation signal identification method based on space-time mixed feature extraction network | |
| CN114615118A (en) | Modulation identification method based on multi-terminal convolution neural network | |
| Wang et al. | A new method of automatic modulation recognition based on dimension reduction | |
| Gai et al. | Spectrum sensing method based on residual cellular network | |
| Su et al. | Robust and efficient modulation recognition with pyramid signal transformer | |
| CN113364519B (en) | Optical performance monitoring method and device, electronic equipment and storage medium | |
| Hao et al. | Contrastive self-supervised clustering for specific emitter identification | |
| Zhao et al. | A spatiotemporal multi-channel learning framework for PCMA signal detection and recognition algorithm | |
| Zhao et al. | A study of individual identification of radiation source based on feature extraction and deep learning | |
| Lu et al. | Spectrum Sensing Algorithm Based on LSTM and Its Implementation of Multiple USRP | |
| Ren et al. | Modulation recognition method of communication signal based on cnn-lstm dual channel | |
| Wen et al. | RF transmitter identification and classification based on deep residual shrinkage network | |
| Motwani et al. | Analysis of various neural network architectures for automatic modulation techniques |
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: 20191220 |