CN112804119A - MAC protocol identification method based on convolutional neural network - Google Patents
MAC protocol identification method based on convolutional neural network Download PDFInfo
- Publication number
- CN112804119A CN112804119A CN202110007441.4A CN202110007441A CN112804119A CN 112804119 A CN112804119 A CN 112804119A CN 202110007441 A CN202110007441 A CN 202110007441A CN 112804119 A CN112804119 A CN 112804119A
- Authority
- CN
- China
- Prior art keywords
- time
- convolutional neural
- neural network
- data
- frequency
- 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.)
- Granted
Links
Images
Classifications
-
- 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
- H04L43/18—Protocol analysers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Communication Control (AREA)
Abstract
The invention discloses a MAC protocol identification method based on a convolutional neural network, which comprises the following steps: (1) generating training data: firstly, obtaining labeled original sampling data in the form of a time-power value sequence, and then converting the original sampling data into a time-frequency graph; (2) training a convolutional neural network: inputting the generated time-frequency diagram into a convolutional neural network for training to obtain a trained convolutional neural network model; (3) and MAC protocol identification: and acquiring data of a network to be identified, converting the data into a time-frequency diagram, and inputting the time-frequency diagram into the trained convolutional neural network model to realize protocol identification. The method adopts an imaging recognition idea to convert the original communication data of the communication network into a time-frequency graph, and classifies the time-frequency graph by using the convolutional neural network to realize the recognition of the MAC protocol type of the target network.
Description
Technical Field
The invention relates to the technical field of non-cooperative communication networks, in particular to a MAC protocol identification method based on a convolutional neural network.
Background
In a wireless communication network, a Medium Access Control (MAC) protocol mainly solves a problem of how to allocate and divide channel resources when a plurality of nodes share the same link in the network. The MAC protocols adopted by different types of communication networks are different, and the identification of the MAC protocols has important significance for obtaining communication rules, estimating access parameters in a frequency spectrum hole mode and changing a cognitive radio access mechanism. Specifically, in the field of cognitive radio, after a cognitive user identifies the MAC protocol type of a master user network, access parameters such as duration in a spectrum hole mode can be acquired, the access parameters of the cognitive user network can be adjusted in a self-adaptive mode, and the spectrum utilization rate is improved. In addition, for a network in which the master user is a reaction (for example, in a master user network adopting a CSMA/CA protocol, the transmission of the cognitive user may cause a change in an internal operation state of the master user, such as a size of a backoff window), after the cognitive user determines the MAC protocol type of the master user, an access mechanism of the master user may be changed, thereby avoiding interference with the master user.
At present, most of researches on MAC protocol identification adopt a support vector machine method in machine learning, the identification performance of the method is greatly influenced by traffic load, and when the traffic load is lower, the high protocol identification accuracy rate cannot be achieved. Furthermore, the feature extraction of the protocol requires manual operations, requires a high level of expertise and a high level of effort.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a MAC protocol identification method based on a convolutional neural network, which adopts an imaging identification thought to convert original communication data of a communication network into a time-frequency graph, classifies the time-frequency graph by using the convolutional neural network to realize the identification of the MAC protocol type of a target network, and the researched MAC protocol types are four types: TDMA (time division multiple access), CSMA/CA (carrier sense multiple access with collision amplitude), Slotted ALOHA and Pure ALOHA.
In order to solve the above technical problem, the present invention provides a MAC protocol identification method based on a convolutional neural network, which includes the following steps:
(1) generating training data: firstly, obtaining labeled original sampling data in the form of a time-power value sequence, and then converting the original sampling data into a time-frequency graph;
(2) training a convolutional neural network: inputting the generated time-frequency diagram into a convolutional neural network for training to obtain a trained convolutional neural network model;
(3) and MAC protocol identification: and acquiring data of a network to be identified, converting the data into a time-frequency diagram, and inputting the time-frequency diagram into the trained convolutional neural network model to realize protocol identification.
Preferably, in step (1), the original sampling data is obtained by sampling communication data of a known network (MAC protocol type) in a period of time, so as to obtain a time sequence of signals and a corresponding power value sequence and a communication frequency; and tags the data according to the type of MAC protocol employed by the known network, the four MAC protocols being denoted by 0, 1, 2 and 3, respectively.
Preferably, in step (1), a time-frequency diagram is further generated based on the raw data, including the time-power value sequence and the communication frequency. The abscissa and the ordinate of the time-frequency diagram respectively represent a time domain and a frequency domain, and represent the magnitude of the power value by the color depth, which can embody the time characteristic, the power characteristic and the frequency domain information of the original data.
Preferably, in the step (2), when the convolutional neural network is used for training data, training samples corresponding to the four MAC protocols are classified according to the four labels, after the designed convolutional neural network model is compiled, the training data is input into the network to train the model, and the initial value of the training round is set to be 30.
Preferably, in the step (2), the convolutional neural network model is a deep network structure with a small convolutional filter, and the network mainly comprises 8 hidden layers, wherein convolutional layers and pooling layers are alternately connected; the model is added with a full connection layer and a 4-channel softmax layer at last and is used for outputting the final four-classification result; and in order to overcome the over-fitting problem, a dropout layer is added to the model.
Preferably, in the step (3), the signal of the network to be identified is subjected to data acquisition and converted into a time-frequency diagram, and the time-frequency diagram is input into the trained convolutional neural network model to realize the identification of the protocol, which specifically comprises the following steps:
(31) acquiring original data: acquiring signals of a network to be identified within a period of time to obtain a time sequence of the signals, a corresponding power value sequence and communication frequency, namely original sampling data;
(32) generating time-frequency graph data: converting the original sampling data into a time-frequency graph;
(33) testing a convolutional neural network: and inputting the time-frequency diagram serving as a test set into the trained convolutional neural network model for testing, and outputting a classification result, namely a certain label in '0', 1 ', 2 and 3' by the model.
The invention has the beneficial effects that: the original communication data is presented by a time-frequency graph and directly used as input data of a convolutional neural network, the original communication data is not subjected to further feature selection, and an image containing a large amount of information is used for replacing a numerical value, so that the information loss existing when the support vector machine method is used for manually extracting features is avoided; based on the superiority of the convolutional neural network in the aspect of image processing, and the time-frequency graph is combined with the convolutional neural network, the method can realize the imaging MAC protocol identification with automatic feature extraction and high accuracy.
Drawings
Fig. 1 is a schematic view of a scenario in which the present invention is applied.
Fig. 2 is a time domain diagram of four MAC protocols considered by the method of the present invention.
FIG. 3 is a schematic diagram of the time and frequency required by the method of the present invention.
FIG. 4 is a schematic flow chart of the method of the present invention.
Fig. 5 is a flowchart illustrating the MAC protocol identification step of the present invention.
FIG. 6 is a comparison graph of the recognition effect of the present invention method and the SVM method.
Detailed Description
As shown in fig. 1, the background of the embodiment is that, in a non-cooperative wireless communication network communication scenario, a sensing node is used by a party to identify a MAC protocol type adopted by a target network composed of a plurality of radiation sources, so as to adjust an access parameter. The present embodiment assumes that several radiation sources of the target network all communicate on the same frequency point.
Specifically, the target network communicates using one of four MAC protocols — TDMA, CSMA/CA, Slotted ALOHA and Pure ALOHA. Of these four MAC protocols, TDMA and CSMA/CA protocols are collision-free protocols and Slotted ALOHA and Pure ALOHA are collision-present protocols. Fig. 2 is a time domain diagram of four MAC protocols considered in the present invention. As shown, both protocols have a periodic feature in the time dimension, as TDMA and Slotted ALOHA always send packets at the start of an allocated slot, whereas CSMA/CA and Pure ALOHA do not. Furthermore, from the power dimension, Slotted ALOHA and Pure ALOHA cause power values to overlap when packet collisions occur, whereas TDMA and CSMA/CA protocols do not. In summary, the information including the two layers of time and power can be used as a classification basis for four MAC protocols.
Further, in this embodiment, identifying the MAC protocol type of the target network requires collecting communication data of the target network to obtain original data, i.e., a time sequence, a power value sequence, and a communication frequency. And converting the sampled data into a time-frequency diagram, and inputting the time-frequency diagram into a trained convolutional neural network model to realize a protocol identification task.
Further, as mentioned above, the time-power sequence includes time and power dimension characteristics of four MAC protocols, which can be used as a basis for classifying the four protocols. In order to fully utilize the original data containing the protocol features, the embodiment further converts the original sampling data into a time-frequency diagram, rather than artificially extracting the features of the original data. Fig. 3 is a time-frequency diagram illustration required by a convolutional neural network-based MAC protocol identification method. The horizontal axis represents time, the vertical axis represents frequency, and the color intensity represents the magnitude of the power value. As can be seen from fig. 3, the time-frequency diagram can reflect the characteristics of the communication data of the target network under different protocols and contains a large amount of information. And by combining the advantage that the convolutional neural network is good at automatically extracting features from a large amount of data, the embodiment can realize high-accuracy identification of the protocol.
Fig. 4 is a specific flowchart of a MAC protocol identification method based on a convolutional neural network provided in the present invention, including the following steps:
s401, generating training data: firstly, a plurality of groups of labeled original sampling data are obtained, and the sampling data are directly converted into a time-frequency graph.
Specifically, the original sampling data may be acquired by acquiring communication data of a known network (MAC protocol type) within a period of time, to obtain a time sequence of signals, a corresponding power value sequence, a communication frequency, and N pairs of time-power values corresponding to the N sampling points. On the basis, the group of data is labeled according to the MAC protocol type of the network, and training data are generated. Multiple sets of training data may be obtained for multiple samples of networks employing different types of protocols.
S402, convolutional neural network training: and inputting the generated multiple groups of training data (time-frequency graphs) into a convolutional neural network for training to obtain a trained convolutional neural network model.
Further, the present invention uses a general convolutional neural network model, i.e., a deep network structure with a small convolutional filter. Specifically, the method comprises 8 hidden layers, namely a convolutional layer and a pooling layer, which are alternately used. In addition, the model also adds a full-link layer and a 4-channel softmax layer. And finally adding a dropout layer in order to overcome the over-fitting problem.
S403, MAC protocol identification: and collecting communication data of a target network to be identified to obtain a time-power value sequence and communication frequency of the signal, converting the time-power value sequence and the communication frequency into a time-frequency diagram, and inputting the time-frequency diagram into a trained convolutional neural network model to realize protocol identification.
Further, as shown in fig. 5, the MAC protocol identification of the present invention specifically includes:
s501, acquiring original data: the method comprises the steps of collecting communication data of a network to be identified within a period of time, and obtaining a time sequence of signals and a corresponding power value sequence, namely original sampling data.
S502, generating time-frequency graph data: and directly converting the original sampling data consisting of the time-power value sequence into a time-frequency graph.
S503, testing a convolutional neural network: and inputting the time-frequency diagram into the trained convolutional neural network model, and outputting the MAC protocol label corresponding to the time-frequency diagram by the model.
The invention has the following beneficial effects: in order to avoid the problem of information loss when the characteristics are manually extracted in the support vector machine method, the invention converts the original communication data into a time-frequency graph containing a large amount of characteristic information; and then, based on the characteristic that the convolutional neural network is good at automatically extracting features from a large amount of data, the time-frequency graph is combined with the convolutional neural network, so that the imaged MAC protocol identification with high accuracy and automatically extracted features is realized. As shown in fig. 6, for comparing the recognition effects of the support vector machine method and the convolutional neural network method provided by the present invention, it is obvious that the method of the present invention has a beneficial effect on MAC protocol recognition.
Claims (6)
1. A MAC protocol identification method based on a convolutional neural network is characterized by comprising the following steps:
(1) generating training data: firstly, obtaining labeled original sampling data in the form of a time-power value sequence, and then converting the original sampling data into a time-frequency graph;
(2) training a convolutional neural network: inputting the generated time-frequency diagram into a convolutional neural network for training to obtain a trained convolutional neural network model;
(3) and MAC protocol identification: and acquiring data of a network to be identified, converting the data into a time-frequency diagram, and inputting the time-frequency diagram into the trained convolutional neural network model to realize protocol identification.
2. The convolutional neural network-based MAC protocol identification method of claim 1, wherein in step (1), the raw sampling data is obtained by sampling communication data of a known network for a period of time, so as to obtain a time sequence of signals and a corresponding power value sequence and communication frequency; and tags the data according to the type of MAC protocol employed by the known network, the four MAC protocols being denoted by 0, 1, 2 and 3, respectively.
3. The convolutional neural network-based MAC protocol identification method of claim 1, wherein in step (1), a time-frequency map is further generated based on the raw data including the time-power value sequence and the communication frequency.
4. The convolutional neural network-based MAC protocol recognition method of claim 1, wherein in step (2), when the convolutional neural network is used to train data, the training samples corresponding to the four MAC protocols are classified according to the four labels, after the designed convolutional neural network model is compiled, the training data is input into the network to train the model, and the initial value of the training round is set to 30.
5. The convolutional neural network-based MAC protocol identification method of claim 1, wherein in step (2), the convolutional neural network model is a deep network structure with a small convolutional filter, the network is mainly composed of 8 hidden layers, and the convolutional layers and the pooling layers are alternately connected; the model is added with a full connection layer and a 4-channel softmax layer at last and is used for outputting the final four-classification result; in addition, a dropout layer is added to the model.
6. The convolutional neural network-based MAC protocol recognition method of claim 1, wherein in step (3), the signal of the network to be recognized is subjected to data acquisition and converted into a time-frequency diagram, and the time-frequency diagram is input into a trained convolutional neural network model to realize the protocol recognition, specifically comprising the following steps:
(31) acquiring original data: acquiring signals of a network to be identified within a period of time to obtain a time sequence of the signals, a corresponding power value sequence and communication frequency, namely original sampling data;
(32) generating time-frequency graph data: converting the original sampling data into a time-frequency graph;
(33) testing a convolutional neural network: and inputting the time-frequency diagram serving as a test set into the trained convolutional neural network model for testing, and outputting a classification result, namely a certain label in '0', 1 ', 2 and 3' by the model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110007441.4A CN112804119B (en) | 2021-01-05 | 2021-01-05 | MAC protocol identification method based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110007441.4A CN112804119B (en) | 2021-01-05 | 2021-01-05 | MAC protocol identification method based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112804119A true CN112804119A (en) | 2021-05-14 |
| CN112804119B CN112804119B (en) | 2022-06-14 |
Family
ID=75808171
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110007441.4A Active CN112804119B (en) | 2021-01-05 | 2021-01-05 | MAC protocol identification method based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112804119B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113221863A (en) * | 2021-07-12 | 2021-08-06 | 深圳市永达电子信息股份有限公司 | Encryption protocol identification method, system and medium based on electromagnetic space period statistics |
| CN113364647A (en) * | 2021-06-03 | 2021-09-07 | 上海天旦网络科技发展有限公司 | Rapid protocol stack identification method and system based on multitask network |
| CN115277513A (en) * | 2022-03-23 | 2022-11-01 | 电子科技大学 | MAC layer protocol identification method based on data time sequence characteristics |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109274621A (en) * | 2018-09-30 | 2019-01-25 | 中国人民解放军战略支援部队信息工程大学 | Communication protocol signals recognition methods based on depth residual error network |
| CN111562842A (en) * | 2020-01-21 | 2020-08-21 | 北京航空航天大学 | Virtual keyboard design method based on electromyographic signals |
-
2021
- 2021-01-05 CN CN202110007441.4A patent/CN112804119B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109274621A (en) * | 2018-09-30 | 2019-01-25 | 中国人民解放军战略支援部队信息工程大学 | Communication protocol signals recognition methods based on depth residual error network |
| CN111562842A (en) * | 2020-01-21 | 2020-08-21 | 北京航空航天大学 | Virtual keyboard design method based on electromyographic signals |
Non-Patent Citations (1)
| Title |
|---|
| XUTONG ZHANG 等: "A MAC Protocol Identification Approach Based on Convolutional Neural Network", 《2020 INTERNATIONAL CONFERENCE ON WIRELESS COMMUNICATIONS AND SIGNAL PROCESSING (WCSP)》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113364647A (en) * | 2021-06-03 | 2021-09-07 | 上海天旦网络科技发展有限公司 | Rapid protocol stack identification method and system based on multitask network |
| CN113221863A (en) * | 2021-07-12 | 2021-08-06 | 深圳市永达电子信息股份有限公司 | Encryption protocol identification method, system and medium based on electromagnetic space period statistics |
| CN113221863B (en) * | 2021-07-12 | 2021-10-22 | 深圳市永达电子信息股份有限公司 | Encryption protocol identification method, system and medium based on electromagnetic space period statistics |
| CN115277513A (en) * | 2022-03-23 | 2022-11-01 | 电子科技大学 | MAC layer protocol identification method based on data time sequence characteristics |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112804119B (en) | 2022-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112804119B (en) | MAC protocol identification method based on convolutional neural network | |
| Schmidt et al. | Wireless interference identification with convolutional neural networks | |
| CN111935721B (en) | Method and system for realizing coexistence of heterogeneous networks | |
| Bitar et al. | Wireless technology identification using deep convolutional neural networks | |
| Zhang et al. | Multi-agent deep reinforcement learning-based cooperative spectrum sensing with upper confidence bound exploration | |
| Shahid et al. | A convolutional neural network approach for classification of lpwan technologies: Sigfox, lora and ieee 802.15. 4g | |
| CN105978644B (en) | Star based on cuckoo searching algorithm ground cognitive system frequency spectrum access method | |
| CN108683526B (en) | Method for identifying competitive MAC protocol | |
| CN111355616B (en) | Tactical communication network key node identification method based on physical layer data | |
| CN109787928B (en) | Decision tree TWSVM-based high-speed optical signal modulation format recognition method | |
| CN110022182A (en) | Interference source prediction model method for building up, method for interference source identification, device and terminal device | |
| CN107231427B (en) | MAC protocol identification method based on support vector machine | |
| Qin et al. | Pathway to intelligent radio | |
| CN113922899A (en) | Spectrum sensing access method based on neural network and Welch transformation | |
| CN105553574A (en) | Support-vector-machine-based MAC protocol identification method in cognitive radio | |
| Ya et al. | Modulation recognition of digital signal based on deep auto-ancoder network | |
| Zhang et al. | A mac protocol identification approach based on convolutional neural network | |
| EP4026360A1 (en) | Predicting a radio spectrum usage | |
| CN111865848A (en) | Signal modulation format identification method and system | |
| Yang et al. | Improving medium access efficiency with intelligent spectrum learning | |
| CN113259289B (en) | Single-channel aliasing signal modulation mode identification method based on residual error neural network | |
| CN116094594A (en) | Optical communication method, optical communication device, electronic equipment and storage medium | |
| Xu et al. | High-frequency band automatic mode recognition using deep learning | |
| CN115276858A (en) | Dynamic spectrum multi-domain anti-interference method and system based on cognitive anti-interference model | |
| CN114501344A (en) | Massive LoRa node fast ad hoc network communication method based on wireless ranging technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |