CN102869064A - Cluster modulation identification method based on feature level and strategy level combined fusion - Google Patents
Cluster modulation identification method based on feature level and strategy level combined fusion Download PDFInfo
- Publication number
- CN102869064A CN102869064A CN2012102626716A CN201210262671A CN102869064A CN 102869064 A CN102869064 A CN 102869064A CN 2012102626716 A CN2012102626716 A CN 2012102626716A CN 201210262671 A CN201210262671 A CN 201210262671A CN 102869064 A CN102869064 A CN 102869064A
- Authority
- CN
- China
- Prior art keywords
- fusion
- modulation
- level
- 2ask
- bunch
- 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
Abstract
The invention aims to provide a cluster modulation identification method based on feature level and strategy level combined fusion. The method is used for carrying out feature level and strategy level secondary fusion recognition to a modulation signal, overcomes the disadvantage of low recognition rate under low single-to-noise ratio, and provides the robustness of modulation recognition. The method comprises the steps that multiple sensor nodes in a wireless sensor network are divided into a plurality of clusters; respective feature parameter of each user in the corresponding cluster is extracted, the feature parameters are transmitted to converging node cluster heads, each cluster head transmits a feature vector constructed by feature parameters respectively extracted by users in the corresponding cluster to a trained SVM (support vector machine) classifier, the received modulation signal is subjected to feature level fusion, and the recognition result of the corresponding cluster is obtained; and finally, the cluster heads transmit the recognition results of the clusters to a fusion center, then the recognition results are subjected to strategy fusion, the fusion center carries out voting judgment on the recognition result from each cluster according to the voting fusion rule, so that the final recognition result of the modulation signal corresponding to the modulation type is obtained.
Description
Technical field
The present invention relates in a kind of multisensor network to unite based on feature level and decision level the implementation of the sub-clustering Modulation Identification of fusion, belong to communication technical field.
Background technology
Along with the development of the communication technology, the day by day complexity of wireless communications environment effectively utilizes band resource more, and the modulation system of signal of communication is more varied, and is simultaneously also more complicated.The automatic identification technology of modulation is the identification of the signal of communication that receives being carried out modulation system, and the vital task of bearing threat assessment and disturbing identification is one of key measure that guarantees legitimate correspondence, the significant and effect in the civilian and military field.
At present, the research method of automatic recognition mainly can be divided into two classes: based on the decision theory method with based on statistical pattern recognition method.To utilize the leaf Bayesian theory in the hypothesis testing to solve classification problem based on the decision theory method, according to the statistical property of signal, obtain test statistics by theory analysis and derivation, with suitable threshold value comparison, form decision rule, so the setting of threshold value is core and the emphasis of the method.Based on statistical pattern recognition method, threshold value need not be set, comprise modulation signal is carried out feature extraction and according to grader modulation signal carried out these two parts of Classification and Identification.Feature extracting method is commonly used signal transient characteristic quantity, Higher Order Cumulants, wavelet transformation etc.Be used for the grader of Modulation Identification, mainly comprise artificial neural net, SVMs, cluster and some other mode identification method, but major part only limit to the research of single node.
Wherein, a kind of mode identification method that SVMs grows up based on Statistical Learning Theory, it at first is mapped to a higher dimensional space by the nonlinear transformation of kernel function definition with input vector, transfer nonlinear problem to linear problem, at this higher dimensional space structure optimum linearity classification hyperplane, realize optimal classification, possess good Generalization Ability.
Along with wireless communications environment is day by day complicated, the development of multisensor Data Fusion technology, various data anastomosing algorithms are applied to each side, comprise being applied in the cooperation Modulation Identification.Data fusion technique is applied in the Automatic Modulation Recognition technology, overcome single node and finished separately the problem that the judgement of the extraction that receives signal and receiving signal type processed may exist, such as problems such as the deep fade that exists in the radio communication, shadow effect and concealed nodes, so that the Modulation Identification performance is more accurately reliable, has robustness.For Modulation Identification, more commonly data fusion, feature level merge and decision level fusion.Pixel-based fusion is the fusion of minimum level, is the fusion that the most original data are carried out, and requires fusion center that higher error correcting capability is arranged, and requires high to channel capacity.It is to extract characteristic parameter in original modulation signal that feature level merges, and the composition characteristic parameter vector send fusion center to integrate and processes, and fusion center can be to carry out fusion recognition with integration technologies such as cluster, SVMs or neural nets.Feature level merges widely compressed information amount, than the pixel-based fusion transmission capacity require low, little than decision level fusion information loss amount, but the burden of fusion center is heavier.Decision level fusion is to have finished identification mission before merging, and fusion center makes a policy according to certain criterion, obtains recognition result.Existing decision level fusion research algorithm has the Voting Fusion criterion, maximum posteriori criterion and DS evidence theory etc.Each node of Decision fusion needs to judge separately modulation system, and amount of calculation merges large than feature level, but the amount of calculation of fusion center is less, requires lower to transmission bandwidth.Existing most research is that independent fusion method is applied in the Modulation Identification, because the complementarity of the whole bag of tricks two, consideration will be planted and two or more methods is united, and can maximize favourable factors and minimize unfavourable ones, and than a kind of method of independent employing more excellent recognition performance be arranged.
Summary of the invention
Technical problem:The object of the present invention is to provide in a kind of multisensor network based on the sub-clustering Modulation Identification method of feature with decision-making associating fusion, the method is carried out the secondary fusion recognition of Fusion Features and Decision fusion to modulation signal, overcome independent employing Fusion Features than the low and independent employing Decision fusion of discrimination under the high s/n ratio shortcoming low than discrimination under the low signal-to-noise ratio, so that modulation signal has better recognition performance than the Modulation Identification method of independent employing feature level fusion and the Modulation Identification method of decision level fusion from low to high in signal to noise ratio.
Technical scheme:The invention provides a kind of method of the new Modulation Identification that merges based on sub-clustering, the method is based on the associating of feature level fusion and decision level fusion, modulation signal is carried out the secondary fusion recognition, effectively overcome independent employing Fusion Features than the low and independent employing Decision fusion of discrimination under the high s/n ratio shortcoming low than discrimination under the low signal-to-noise ratio.The method has been chosen 2ASK, BPSK, and QASK, 4PSK, 4FSK, these several typical modulation systems of OFDM are identified.
The present invention is based on feature level and decision level unites the sub-clustering Modulation Identification method of fusion and comprises following steps:
Step 5. M bunch head respectively will bunch in the recognition result that merges of feature level deliver to respectively fusion center;
Step 6. fusion center adopts the Voting Fusion criterion to carry out decision level fusion identification, if it is 2ASK that the modulation type of a bunch of head judgement modulation signal is arranged, throw so 2ASK one ticket, also be to adopt identical way for other modulation types, behind the poll closing, add up this voting results, if there is a certain modulation type poll to surpass
, so fusion center with this modulation type as last court verdict.
Described support vector machine classifier, its structure may further comprise the steps:
Step 5 is for { 2ASK, 4ASK }, according to parameter
,
With
, structure level V grader is divided into { 2ASK } and { 4ASK } with { 2ASK, 4ASK }.
Beneficial effect:For the wireless sensor network of multinode, the invention provides a kind of new sub-clustering Modulation Identification method based on feature and decision-making associating fusion, have following beneficial effect:
1. the present invention combines the advantage of feature level fusion and decision level fusion, overcome independent employing Fusion Features than the low and independent employing Decision fusion of discrimination under the high s/n ratio shortcoming low than discrimination under the low signal-to-noise ratio, so that modulation signal has better recognition performance than the Modulation Identification method of independent employing feature level fusion and the Modulation Identification method of decision level fusion from low to high in signal to noise ratio.
2. the present invention only utilizes 3 parameters to modulation signal, guarantees correctly to carry out reliably Modulation Mode Recognition, and parameter extraction once just can repeatedly use in grader simultaneously, reduces amount of calculation, saves the energy consumption of sensor node.
3. svm classifier device design utilizes 3 parameters to adopt stage divisions, has overcome in the SVMs classification problem the long and large shortcoming of class classifier number one to one of a pair of multiclass training time, high-class efficient.
Description of drawings
Fig. 1 is the schematic diagram that the multisensor network carries out feature and decision-making associating fusion recognition modulation signal,
Fig. 2 is the classification svm classifier device schematic diagram based on decision Binary Tree,
The feature level of Fig. 3 in being bunch merges schematic diagram,
The decision level fusion schematic diagram of Fig. 4 between being bunch.
Embodiment
Pixel-based fusion has kept the most original information of signal, but high to the disposal ability requirement of fusion center, and is also high to the requirement of transmission bandwidth, is difficult for realizing for the multisensor network; Feature level merges its characteristic parameter of signal extraction, is combined into the characteristic parameter vector and is sent to fusion center integration processing, and each sensor node is shared amount of calculation, and is also relatively low to the requirement of fusion center and bandwidth; Each node of decision level fusion identification requirement rules out separately modulation type, recognition result is sent fusion center judgement again, to bandwidth require lowly, the amount of calculation of each node is large.What the present invention proposed carries out sub-clustering with multinode in the multisensor network, adopt the feature level fusion recognition in bunch, adopts decision level fusion identification between bunch, in conjunction with having come the advantage of feature level fusion with decision level fusion.The sub-clustering Modulation Recognition based on feature and decision-making associating fusion that the present invention proposes comprises three parts: the one, and the characteristic parameter of extraction modulation signal; The 2nd, the feature level fusion recognition in bunch; The 3rd, the decision level fusion identification between bunch.
Based on the system model of feature and the sub-clustering Modulation Identification method of decision-making associating fusion as shown in Figure 1, among the figure, T is for sending modulation signal in the multisensor network that the present invention proposes, C is fusion center, 0,1,2,3 ... N, N+1, N+2, N+3, N+4 is respectively sensor node, and wherein 2,, N+1 is aggregation node bunch head.Sensor node is dispersed in different geographical position, and each transducer independently receives the modulation signal that sends over, transducer in each bunch comprises aggregation node---bunch head, comprised the svm classifier device in the aggregation node, be used for modulation system is identified, the specific implementation process is as follows: (1) modulation signal sends in each sensor network nodes, and the multinode in the sensor network is divided into some bunches, and several node transducers are arranged in each bunch.Each node receives modulation signal in the sensor network, and all has enough energy and extract characteristic parameter and it is sent to aggregation node bunch head; (2) bunch a transducer will bunch in the characteristic parameter sent here of each other node transducers be combined into characteristic vector, send into the svm classifier device that has trained and carry out feature level and merge, obtain the recognition result of modulation signal; (3) each bunch head will be separately recognition result deliver to fusion center and carry out decision level fusion identification, utilize the judgement of Voting Fusion criterion, obtain the last court verdict of modulation signal.
The first based on feature and the sub-clustering Modulation Recognition flow process of decision-making associating fusion that the present invention proposes is exactly the pretreatment stage of signal, namely the user of each in each bunch extracts respectively characteristic parameter separately, characteristic quantity and building method thereof based on the feature extraction classics mainly comprise: the signal transient characteristic quantity, wavelet transformation, Higher Order Cumulants etc., the present invention has chosen based on 3 characteristic parameter combinations in the instantaneous characteristic quantity, is respectively
,
With
The below provides its definition and calculating:
1) standard deviation of the non-weak signal section of zero center instantaneous phase nonlinear component absolute value
, definition is:
Wherein,
An amplitude decision threshold level judging the weak signal section,
TIn the gross sample data
NIn belong to the number of non-weak signal value,
The nonlinear component of instantaneous phase after the zero center processing, can by
Calculate, wherein
,
It is instantaneous phase.
2) standard deviation of the non-weak signal section of zero center instantaneous phase nonlinear component
, definition is:
Wherein the connotation of variable with
In identical.It with
Difference only be:
The standard deviation of phase place absolute value, and
It is the standard deviation of Direct Phase.
3) standard deviation of normalize and center instantaneous amplitude absolute value
, definition is:
Wherein,
NExpression receives the sample points of burst,
The normalize and center instantaneous amplitude, it can by
Calculate, in
, and
Be instantaneous amplitude
A (i)Mean value, coming instantaneous amplitude is carried out normalized purpose with mean value is to eliminate the impact of channel gain.
Emulation of the present invention 2ASK, BPSK, QASK, 4PSK, 4FSK, 3 the characteristic parameter figures of characteristic parameter under different signal to noise ratios can find, for different modulation signals in these several modulation signals of OFDM, the value of each characteristic parameter has obvious difference, signal to be identified can be carried out different classes of differentiation, design different svm classifier devices according to different characteristic parameters, these six kinds of modulation systems are identified.In the document based on the instantaneous characteristic value of feature extraction in the past, find that more grader is to have adopted
,
,
,
,
These several characteristic parameters, consider in the multisensor network, the design of grader should consider to guarantee correctly to carry out reliably the parameter of Modulation Mode Recognition, just consider again simultaneously to extract once can be in grader nonexpondable parameter, reduce amount of calculation, save the energy consumption of sensor node.The present invention has designed a kind ofly to be only had
,
,
The grader of 3 parameters such as Fig. 2, has not only guaranteed recognition performance but also has saved the energy consumption of each Node extraction characteristic parameter; Simultaneously, during SVMs identification multiclass problem, a pair of multiclass commonly used reaches one to one two kinds of algorithms of class.In the present invention, the stage division that is based on binary tree that the design of grader is adopted has overcome the long and large shortcoming of class classifier number one to one of a pair of multiclass training time.
By Fig. 2, can see, utilize
,
,
The svm classifier device of these three characteristic parameter Combination Design has been used 5 classifiers, wherein parameters
Be used to four times, so characteristic parameter
Bunch in different transducers to repeat to extract.By
Parameter Map as can be known, the differentiation of these two kinds of modulation systems of 4FSK and OFDM is fairly obvious, both indicatrixes relatively far apart.But
Parameter Map in the differentiation of 2ASK and these two kinds of debud modes of 4ASK and not obvious, both indicatrixes are relatively approaching, by simulation result as can be known, the debud mode that discrimination is lower is 2ASK and 4ASK, so characteristic parameter
Bunch in different transducers to repeat to extract.Therefore, in the sub-clustering Modulation Identification method based on feature and decision-making associating fusion that the present invention proposes, multisensor is divided into 6 bunches, 5 sensor nodes are arranged in each bunch, the allocative decision of extracting for these 5 sensor node parameters is: in 5 sensor nodes, each sensor node only extracts a characteristic parameter, wherein has two transducers respectively to extract characteristic parameter
Once, there are two transducers respectively to extract characteristic parameter
Once, there is a sensor node to extract characteristic parameter
Once.
The feature level of the second portion based on feature and the sub-clustering Modulation Recognition flow process of decision-making associating fusion that the present invention proposes in being exactly bunch merges, in the multisensor network, multisensor is divided into some bunches, in each bunch, carry out the feature level fusion recognition, such as Fig. 3, some sensor nodes 1,2 are arranged in each bunch ... N
Represent i the modulation signal that each transducer receives, although reception is the modulation signal that same transmitting terminal sends, each transducer is in different geographical position, and the channel of process is different, and last reception signal also is different.The present invention supposes that channel is shadow fading, and the Signal-to-Noise obeys logarithm normal distribution that receives of each transducer.Sensor node 1,2 ..., the characteristic parameter that N extracts is respectively
, being sent to simultaneously a bunch head, a bunch head is combined into characteristic vector with it
, send the identification of svm classifier device.
SVM is mapped to input vector the feature space of a higher-dimension, the problem of input space linearly inseparable can be converted into the linear separability problem solve in higher dimensional space, the classifying face of an optimum of structure in high-dimensional feature space.Requiring to classify the plane not only can be faultless separately with two class samples, and will make the distance between two classes maximum.The decision function that can try to achieve the optimal classification hyperplane is:
SVM utilizes the decision function that has trained to characteristic vector
Calculate, obtain
Value, thereby solved classification problem.
In the feature level fusion recognition, a plurality of sensor nodes in the multisensor network are divided into some bunches, than adopting separately the feature level blending algorithm, total sensor node number is constant, but once participates in the sensor node number minimizing that feature level merges in the cluster-dividing method in every bunch, and the characteristic parameter of transmission is few, transmission bandwidth is required to reduce, reduce simultaneously the amount of calculation of fusion center, alleviated the burden of fusion center grader, reduced the training time of grader.
The decision level fusion of the third part based on feature and the sub-clustering Modulation Recognition flow process of the associating fusion of making a strategic decision that the present invention proposes between being exactly bunch, each bunch head bunch in obtain recognition result by the feature level fusion recognition, send fusion center to utilize decision rule to carry out decision level fusion identification its result, such as Fig. 4, in the multisensor each bunch 1,2 ... M
Represent that M bunch head to the recognition result of the i time modulation signal, is 2ASK, BPSK, 4ASK, 4PSK, 4FSK, one of these six kinds of modulation systems of OFDM.The specific practice of Voting Fusion criterion of the present invention is: total M bunch head, if the modulation type that has a bunch of head to adjudicate the i time modulation signal is 2ASK, throw so 2ASK one ticket, and also be to adopt identical way for other modulation types.Behind the poll closing, add up this voting results, if there is a certain modulation type poll to surpass
, so fusion center with this modulation type as last court verdict.
Adopt separately the method for decision level fusion identification, it is the characteristic parameter that each sensor node independently extracts the modulation signal that receives, the svm classifier device that uses the present invention to propose is judged the modulation type of the modulation signal of this node, recognition result with modulation type is sent to fusion center again, and fusion center utilizes the recognition result of the modulation signal that the Voting Fusion criterion sends here each node to make conclusive judgement.The amount of calculation of each node is large in the method, but requires lower to transmission bandwidth.The present invention is in the decision level fusion identification of carrying out on the feature level fusion recognition basis, compare with independent employing decision level fusion identification, be not to be that the amount of calculation of each node is large, the amount of calculation of bunch head of just carrying out the feature level fusion recognition is larger, and the amount of calculation that adopts bunch head in the blending algorithm of sub-clustering participates in feature level fusion recognition, computation amount with respect to the node of all sensors network.Combine the advantage of feature level fusion recognition and decision level fusion identification, modulation signal is carried out secondary identification, with independent employing special level fusion recognition and decision level fusion identification more excellent recognition performance is arranged.
Claims (2)
1. sub-clustering Modulation Identification method of uniting fusion based on feature level and decision level is characterized in that the method includes the steps of:
Step 1. is divided into some bunches with the multisensor node in the wireless sensor network, 5 sensor nodes are arranged in each bunch, and each sensor node independently receives respectively modulation signal and extracts characteristic parameter: the standard deviation of the non-weak signal section of zero center instantaneous phase nonlinear component absolute value
, the non-weak signal section of zero center instantaneous phase nonlinear component standard deviation
Standard deviation with normalize and center instantaneous amplitude absolute value
Step 2. is for 5 sensor nodes of each bunch, and each sensor node only extracts a characteristic parameter, wherein has two transducers respectively to extract characteristic parameter
Each once has two transducers respectively to extract characteristic parameter
Each once has a sensor node to extract characteristic parameter
Once;
Step 3. bunch in, the corresponding characteristic parameter of each Node extraction, and characteristic parameter is sent to a bunch head;
Step 4. bunch head will bunch in the characteristic parameter composition characteristic vector that extracts respectively of each sensor node, deliver to the support vector machine classifier that trains, modulation signal is carried out the feature level fusion recognition, obtain the recognition result of the modulation type of each bunch head;
Step 5. M bunch head respectively will bunch in the recognition result that merges of feature level deliver to respectively fusion center;
Step 6. fusion center adopts the Voting Fusion criterion to carry out decision level fusion identification, if it is 2ASK that the modulation type of a bunch of head judgement modulation signal is arranged, throw so 2ASK one ticket, also be to adopt identical way for other modulation types, behind the poll closing, add up this voting results, if there is a certain modulation type poll to surpass
, so fusion center with this modulation type as last court verdict.
2. unite as claimed in claim 1 the sub-clustering Modulation Identification method of fusion based on feature level and decision level, it is characterized in that described support vector machine classifier, its structure may further comprise the steps:
Step 1, based on the multi-class support vector machine of decision Binary Tree according to parameter
Structure first order grader is divided into modulation system { 4PSK} and { 2PSK, 2ASK, 4ASK, 4FSK, OFDM};
Step 2, for 2PSK, and 2ASK, 4ASK, 4FSK, OFDM} is still according to parameter
Structure second level grader is divided into modulation system { 4FSK, OFDM } and { 2PSK, 2ASK, 4ASK};
Step 3 is for { 4FSK, OFDM }, according to parameter
Structure third level grader is divided into { 4FSK } and { OFDM } with { 4FSK, OFDM };
Step 4 is for { 4ASK} is according to parameter for 2PSK, 2ASK
With
Structure fourth stage grader is with { 2PSK, 2ASK, 4ASK} are divided into { 2PSK } and { 2ASK, 4ASK };
Step 5 is for { 2ASK, 4ASK }, according to parameter
,
With
, structure level V grader is divided into { 2ASK } and { 4ASK } with { 2ASK, 4ASK }.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012102626716A CN102869064A (en) | 2012-07-27 | 2012-07-27 | Cluster modulation identification method based on feature level and strategy level combined fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012102626716A CN102869064A (en) | 2012-07-27 | 2012-07-27 | Cluster modulation identification method based on feature level and strategy level combined fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102869064A true CN102869064A (en) | 2013-01-09 |
Family
ID=47447586
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012102626716A Pending CN102869064A (en) | 2012-07-27 | 2012-07-27 | Cluster modulation identification method based on feature level and strategy level combined fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102869064A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103152790A (en) * | 2013-02-03 | 2013-06-12 | 南京邮电大学 | Two-stage integrated modulation identification method based on relevant clusters |
| CN103647591A (en) * | 2013-12-27 | 2014-03-19 | 中国电子科技集团公司第五十四研究所 | Cooperative interference detection method based on support vector machine |
| CN106371103A (en) * | 2016-10-21 | 2017-02-01 | 复旦大学 | Voting-fusion-based multi-laser-sensor data fusion method |
| CN107634923A (en) * | 2017-09-21 | 2018-01-26 | 佛山科学技术学院 | A kind of distributed communication signal modulate method |
| CN108154164A (en) * | 2017-11-15 | 2018-06-12 | 上海微波技术研究所（中国电子科技集团公司第五十研究所） | Signal of communication modulation classification system and method based on deep learning |
| CN108270703A (en) * | 2016-12-30 | 2018-07-10 | 中国航天科工集团八五研究所 | A kind of signal of communication digital modulation type recognition methods |
| CN110166389A (en) * | 2019-06-12 | 2019-08-23 | 西安电子科技大学 | Modulation Identification method based on least square method supporting vector machine |
| CN111353450A (en) * | 2020-03-06 | 2020-06-30 | 北京波尔通信技术股份有限公司 | Target identification system and method based on heterogeneous electromagnetic perception information fusion |
| CN113033486A (en) * | 2021-04-21 | 2021-06-25 | 上海交通大学 | Signal feature extraction and modulation type identification method based on generalized fractal theory |
| CN113364715A (en) * | 2021-04-30 | 2021-09-07 | 电子科技大学 | Collaborative automatic modulation classification method based on credit voting mechanism |
| US11373116B2 (en) * | 2015-11-16 | 2022-06-28 | Huawei Technologies Co., Ltd. | Model parameter fusion method and apparatus |
| CN114915526A (en) * | 2022-04-19 | 2022-08-16 | 中国人民解放军战略支援部队信息工程大学 | Communication signal modulation identification method, device and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008049175A1 (en) * | 2006-10-24 | 2008-05-02 | Katholieke Universiteit Leuven | High discriminating power biomarker diagnosing |
| CN101917369A (en) * | 2010-07-30 | 2010-12-15 | 中国人民解放军信息工程大学 | Method for identifying modulation mode of communication signal |
-
2012
- 2012-07-27 CN CN2012102626716A patent/CN102869064A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008049175A1 (en) * | 2006-10-24 | 2008-05-02 | Katholieke Universiteit Leuven | High discriminating power biomarker diagnosing |
| CN101917369A (en) * | 2010-07-30 | 2010-12-15 | 中国人民解放军信息工程大学 | Method for identifying modulation mode of communication signal |
Non-Patent Citations (3)
| Title |
|---|
| 刘爱声，等: "多传感器节点分布式协作调制识别算法", 《信号处理》 * |
| 陈美，等: "基于改进神经网络的自动调制识别研究", 《重庆邮电大学学报(自然科学版)》 * |
| 龚晓洁，等: "衰落信道下基于支持向量机的调制识别方法", 《信号处理》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103152790B (en) * | 2013-02-03 | 2016-07-27 | 南京邮电大学 | Two-graded fusion Modulation Identification method based on dependency sub-clustering |
| CN103152790A (en) * | 2013-02-03 | 2013-06-12 | 南京邮电大学 | Two-stage integrated modulation identification method based on relevant clusters |
| CN103647591A (en) * | 2013-12-27 | 2014-03-19 | 中国电子科技集团公司第五十四研究所 | Cooperative interference detection method based on support vector machine |
| CN103647591B (en) * | 2013-12-27 | 2016-06-08 | 中国电子科技集团公司第五十四研究所 | A kind of multipoint cooperative interference detection method based on SVMs |
| US11373116B2 (en) * | 2015-11-16 | 2022-06-28 | Huawei Technologies Co., Ltd. | Model parameter fusion method and apparatus |
| CN106371103A (en) * | 2016-10-21 | 2017-02-01 | 复旦大学 | Voting-fusion-based multi-laser-sensor data fusion method |
| CN108270703A (en) * | 2016-12-30 | 2018-07-10 | 中国航天科工集团八五研究所 | A kind of signal of communication digital modulation type recognition methods |
| CN107634923A (en) * | 2017-09-21 | 2018-01-26 | 佛山科学技术学院 | A kind of distributed communication signal modulate method |
| CN108154164A (en) * | 2017-11-15 | 2018-06-12 | 上海微波技术研究所（中国电子科技集团公司第五十研究所） | Signal of communication modulation classification system and method based on deep learning |
| CN110166389A (en) * | 2019-06-12 | 2019-08-23 | 西安电子科技大学 | Modulation Identification method based on least square method supporting vector machine |
| CN110166389B (en) * | 2019-06-12 | 2021-06-25 | 西安电子科技大学 | Modulation identification method based on least square support vector machine |
| CN111353450A (en) * | 2020-03-06 | 2020-06-30 | 北京波尔通信技术股份有限公司 | Target identification system and method based on heterogeneous electromagnetic perception information fusion |
| CN111353450B (en) * | 2020-03-06 | 2023-12-26 | 北京波尔通信技术股份有限公司 | Target recognition system and method based on heterogeneous electromagnetic perception information fusion |
| CN113033486A (en) * | 2021-04-21 | 2021-06-25 | 上海交通大学 | Signal feature extraction and modulation type identification method based on generalized fractal theory |
| CN113364715A (en) * | 2021-04-30 | 2021-09-07 | 电子科技大学 | Collaborative automatic modulation classification method based on credit voting mechanism |
| CN114915526A (en) * | 2022-04-19 | 2022-08-16 | 中国人民解放军战略支援部队信息工程大学 | Communication signal modulation identification method, device and system |
| CN114915526B (en) * | 2022-04-19 | 2023-05-26 | 中国人民解放军战略支援部队信息工程大学 | Communication signal modulation identification method, device and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102869064A (en) | Cluster modulation identification method based on feature level and strategy level combined fusion | |
| Lee et al. | Deep cooperative sensing: Cooperative spectrum sensing based on convolutional neural networks | |
| Shi et al. | Deep learning-based automatic modulation recognition method in the presence of phase offset | |
| CN102647391A (en) | Cooperative modulation signal identifying method based on data fusion of decision layer | |
| CN101459445B (en) | Cooperative spectrum sensing method in cognitive radio system | |
| CN102571230A (en) | Distributed collaborative signal identification method based on blind estimation of higher order statistics and signal to noise ratio | |
| CN108540202A (en) | A kind of satellite communication signals Modulation Mode Recognition method, satellite communication system | |
| CN108768907A (en) | A kind of Modulation Identification method based on temporal characteristics statistic and BP neural network | |
| CN103067325A (en) | Cooperative modulation identification method based on multi-class characteristic parameters and evidence theory | |
| CN107612867A (en) | A kind of order of modulation recognition methods of MQAM signals | |
| Kokalj-Filipovic et al. | Adversarial examples in RF deep learning: detection of the attack and its physical robustness | |
| CN104994045A (en) | Platform and method for automatically identifying digital modulation mode based on USRP platform | |
| CN101895494B (en) | Stochastic resonance preprocessing-based digital modulation mode automatic identification method | |
| CN108052956A (en) | Wireless light communication subcarrier modulation constellation recognition methods under a kind of atmospheric turbulance | |
| CN105656826A (en) | Modulation recognizing method and system based on order statistics and machine learning | |
| CN103384174A (en) | Method based on cooperation of multiple users and multiple antennas for optimizing spectrum sensing detection probability | |
| CN107451605A (en) | A kind of simple target recognition methods based on channel condition information and SVMs | |
| Agadakos et al. | Deep complex networks for protocol-agnostic radio frequency device fingerprinting in the wild | |
| Subbarao et al. | Automatic modulation recognition in cognitive radio receivers using multi-order cumulants and decision trees | |
| CN114422311B (en) | Signal modulation recognition method and system combining deep neural network and expert priori features | |
| CN117813913A (en) | Method and system for source coding using neural networks | |
| Seo et al. | Communication-efficient and personalized federated lottery ticket learning | |
| CN110166389A (en) | Modulation Identification method based on least square method supporting vector machine | |
| CN111510232B (en) | Vehicle networking combined spectrum sensing method based on neural network and application thereof | |
| Essai et al. | Automatic Modulation Classification: Convolutional Deep Learning Neural Networks Approaches |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130109 |