A topology tracking method for a non-cooperative wireless communication network

Abstract

The present application belongs to the technical field of communication network topology reasoning, and particularly relates to a topology structure tracking method of a non-cooperative wireless communication network. The present application is suitable for non-cooperative scenarios and time-varying communication networks, and can quickly find a new signal frequency point of a node when a signal of a target wireless communication network node disappears. In the present application, a linear autoregressive Granger causality analysis method is used to track the target wireless communication network node, which is more simple and efficient than modulation recognition and radiation source identification using nonlinear mapping. The Granger causality analysis matching signal is less susceptible to interference than the modulation recognition and radiation source identification methods, and can efficiently mine new communication relationships possibly existing in the changed node.

Description

Technical Field

[0001] This invention belongs to the field of communication network topology reasoning technology, specifically relating to a method for tracking the topology of a non-cooperative wireless communication network. Background Technology

[0002] In the field of communication network topology reasoning, topology tracking of non-cooperative time-varying wireless communication networks is a crucial research direction. The communication relationships of non-cooperative parties change frequently; alterations in the spatial distribution of non-cooperative units, sudden events, and deliberate avoidance by the target communication network can all cause changes in the communication network topology. This poses a challenge to target communication network reasoning in non-cooperative scenarios. In recent years, numerous experts and scholars have proposed many efficient and reliable methods for topology tracking in time-varying wireless communication networks.

[0003] In 2018, Zhang Suyuan et al. from Nanjing University of Science and Technology proposed a time-varying network link prediction method based on Bayesian theory (application number: CN201811237929.0). The basic idea of ​​this invention is that nodes of the same type are more likely to be related than nodes of different types. First, based on correlation, statistical methods are used to predict the network sequence at future times from the network sequence at past times. Then, the prediction results are corrected using multi-layer network information through Bayesian methods. This invention, which predicts network topology based on correlation, has relatively low accuracy, especially in non-cooperative adversarial scenarios. Only by considering multiple factors and starting from causality can accurate prediction of network topology be achieved.

[0004] In 2020, Wang Jian et al. from Nanjing University proposed an active network topology construction method and system for time-varying networks (application number: CN202011250831.6). This invention aims to solve the problems of the single network topology construction strategy and weak adaptability to time-varying networks in general. First, it extracts network situation features from multiple nodes in the network to construct a situation parameter set. The features include node congestion probability, relative stability of node pairs, average bandwidth, average latency, and average packet loss rate. Then, it shares the situation parameter information of the entire network in a broadcast manner, and finally realizes the aggregation of the entire network situation information, and finally obtains a multi-dimensional network adjacency matrix. This invention constructs the topology structure of time-varying networks from the perspective of the entire network situation. However, many parameters in this invention can only be obtained under cooperative conditions, so this invention cannot be directly used in non-cooperative scenarios.

[0005] In 2022, Song Yehui et al. from the Army Engineering University published an article titled "Topology tracking of dynamic UAV wireless networks" in the Chinese Journal of Aeronautics. This article introduced a method for tracking the topology of dynamic UAV wireless networks. This method models communication events between UAV swarms as multidimensional Hawkes processes, obtains the network topology by solving the multidimensional Hawkes processes, and continuously tracks the network topology using a sliding window mechanism, taking advantage of the gradual changes in UAV networks. However, this method focuses on analyzing the complex communication relationships between multiple communication nodes and does not consider the frequency shifting of communication nodes and subsequent tracking issues. Summary of the Invention

[0006] The purpose of this invention is to provide a method for tracking the topology of a non-cooperative wireless communication network.

[0007] A method for tracking the topology of a non-cooperative wireless communication network includes the following steps:

[0008] Step 1: The sensing network continuously receives data from each node x in the target wireless communication network. i signal {S i (t)};

[0009] Step 2: At the same time T power Within, calculate x for each node in the target wireless communication network. i signal energy When the topology of the target wireless communication network remains stable, record the x of each node. i average signal energy like Then node x i Let x be the node that changes. m Steps 3 to 9 are executed to track the topology of the target wireless communication network.

[0010] Step 3: Obtain the changing node x in the target wireless communication network m Former adjacent node {x 1}, extract T granger Duration comes from node {x 1} radiation signal From at equal intervals Extracting communication events

[0011] Step 4: Select a frequency point with no valid signal and no interference signal to receive T. power Calculate pure noise energy from duration signals.

[0012] If the node x changes m signal energy Greater than pure noise energy Then start from the changing node x m Original frequency Extract communication event S m (n), for S m (n) and Perform Granger analysis, if S m (n) and If Granger causality exists, the topology of the target wireless communication network is determined to remain unchanged; otherwise, proceed to step 5.

[0013] If the node x changes m signal energy Less than or equal to pure noise energy Then proceed directly to step 5;

[0014] Step 5: From the lowest frequency point f low Let the current frequency f begin analysis. p The lowest frequency point f low ;

[0015] Step 6: Extract T granger The current frequency point f of duration p signal At equal intervals from Extracting communication events

[0016] Step 7: [Regarding...] and Perform Granger analysis; if and If Granger causality exists, then the frequency point f p For node x m The new frequency point is denoted as Proceed to step 8; otherwise, switch to the next frequency and return to step 6.

[0017] If until f p For the highest frequency point f high Node x was still not detected. m If the frequency is new, it is determined that the topology of the target wireless communication network remains unchanged;

[0018] Step 8: Extract T granger Frequency of duration Signal At equal intervals from Extracting communication events

[0019] Step 9: From node x mThe second-order adjacency nodes are traversed to x m The set of the highest-order adjacent nodes in the target wireless communication network topology is denoted as x. n ={x 2 ,x 3 ...x n}, x n x represents m The nth order of adjacent nodes; analysis during traversal. With x n China Communications Event S n Granger causality between (t) and node x, discovering the node x m Are there any new connections? If node x is found during the traversal... m If a new communication relationship is established, the target wireless communication network topology is immediately updated, and set x is also updated. n .

[0020] Furthermore, the specific steps of the Granger analysis are as follows: (Based on step 7...) and Taking Granger analysis as an example, the specific steps are as follows:

[0021] Step 7.1: For Perform linear autoregression;

[0022]

[0023]

[0024] Where τ is the autoregression order; ε1(n) and ε0(n) are the residual noise of the regression fit; a 1,k ,b 1,k ,a 0,k These are the regression parameters of the autoregressive model;

[0025] Step 7.2: Calculation In autoregressive models Sum of squared residuals calculate In autoregressive models Sum of squared residuals

[0026] Step 7.3: Calculation As a measure of causality, Indicates by Participation in inference Granger causality in the case of autoregression, where var(R1) and var(R0) are the variances of multiple R1 and R0 values ​​obtained from multiple autoregressions;

[0027] Step 7.4: Repeat steps 7.1 to 7.2 for calculation. With S noise (n) The value is used as a threshold for determining connectivity; where S noise (n) represents the pure noise S noise Communication events extracted at equal intervals in (t);

[0028] Step 7.5: If the current frequency point f p Calculated Value greater than Then determine the current frequency point f p The changing node x m The new transmission signal frequency; otherwise, determine the current frequency f. p and changing node x m No relation.

[0029] The beneficial effects of this invention are as follows:

[0030] This invention is applicable to non-cooperative scenarios and time-varying communication networks. It can quickly locate the new transmission frequency of a target wireless communication network node when its signal disappears. The tracking of target wireless communication network nodes in this invention employs a linear autoregressive Granger causality analysis method, which is simpler and more efficient than modulation identification and radiation source identification using nonlinear mapping. Matching signals using Granger causality analysis is less susceptible to interference compared to modulation identification and radiation source identification methods, thus more efficiently uncovering potential new communication relationships between changing nodes. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the present invention.

[0032] Figure 2 This is the overall flowchart of the present invention.

[0033] Figure 3 This is a detailed flowchart of the present invention.

[0034] Figure 4 This is a schematic diagram of the adjacent nodes of each order of the changing node in this invention. Detailed Implementation

[0035] The present invention will now be further described with reference to the accompanying drawings.

[0036] This invention provides a method for topology tracking in non-cooperative wireless communication networks, relating to the field of communication network topology reasoning. The method includes: detecting changes in the target communication network topology based on the energy of the transmitted signals from nodes; analyzing the Granger causality between communication events of the first-order adjacent nodes of the changed node and all events across the entire frequency band of the target communication network to search for new frequency points of the changed node; extracting communication events from the new frequency points of the changed node and performing Granger causality analysis with communication events of other nodes in the network to mine potential new connectivity relationships; and updating the connectivity relationships of the target communication network. This invention is applicable to topology tracking in non-cooperative time-varying wireless communication networks where the frequency points of node transmitted signals change along with changes in node connectivity relationships. It features high tracking efficiency, and the tracking process eliminates the need for modulation identification or radiation source identification of the signal.

[0037] The applicable scenarios and characteristics of this invention are as follows:

[0038] 1. Changes in the target communication network topology include communication node silence, changes in the frequency of radiated signals, and changes in communication relationships. These correspond to the behavior of non-cooperative communication nodes in real-world scenarios when they realize they are being monitored and try to evade surveillance by remaining silent, switching channels, or changing their contact partners.

[0039] 2. Changes in the communication relationships between nodes in the target communication network are accompanied by changes in the frequency of the corresponding node's transmitted signal;

[0040] 3. Target communication network node x in the scenario of this invention m Even after switching frequencies, it still maintains communication with one to all of its previous neighboring nodes.

[0041] This invention includes the following steps:

[0042] (1) Detection of changes in the topology of the target communication network (see appendix) Figure 2 -S100

[0043] The sensing system receives electromagnetic radiation signals {S} from the target wireless communication network. i (t)}.

[0044] The topology W of the target communication network can be obtained using well-known methods such as solving a Multidimensional Hawkes process or performing Granger causality analysis between pairs of nodes. uu' .

[0045] Detect the signal energy of each node in the target communication network Determine whether the target communication network topology has changed.

[0046] If a frequency signal that was previously present disappears, it indicates that the topology of the target communication network has changed and tracking is required.

[0047] If the target network node xm Radio frequency point Upper signal energy If it becomes smaller, then node x m It may have stopped transmitting signals, however, due to ambient noise S noise The frequency of influence of (t) Signal energy detected A value greater than 0 may also indicate that the node has reduced its transmitted signal power; in this case, node x should be used. m First-order adjacent node {x 1 Communication events Communication events for this node Perform Granger causality analysis.

[0048] The existence of Granger causality indicates that the target network node x m It is still transmitting signals; otherwise, the network topology has changed and tracking would be necessary.

[0049] (2) Search the target communication network across the entire frequency band using the communication events of the neighboring nodes of the changing nodes, see Appendix Figure 2 -S200

[0050] Find the past and changed nodes x m Nodes {x} with connectivity 1}, from node {x 1 Extract communication events

[0051] The full frequency band where the target communication network signal may exist. low →f high The above is based on the duration T granger intercept signal For signal Perform equal-interval sampling to extract communication events.

[0052] The target network searches for the lowest frequency point f across the entire frequency band. low Set the frequency to be lower than the known minimum and maximum frequency of the target communication network. high Set a frequency higher than the highest known target communication network frequency.

[0053] For the current frequency point f p The events on With the changing node x m First-order adjacency node communication events Granger causality analysis was performed to identify new frequency points that might exist at the points of change.

[0054] If the current frequency point f p The events on With the changing node xm First-order adjacency node communication events If Granger causality exists, then the current frequency f p For the changing node x m new frequency Otherwise, repeat the above steps for the next frequency point.

[0055] (3) Update the target communication network connectivity, see appendix. Figure 2 -S300

[0056] Target communication network node x m When switching frequencies, the communication relationship may be switched, that is, to communicate with a more secure node.

[0057] Find a new frequency point for the node Signal extraction is then required. Communication events in Granger causality analysis is performed on each of the remaining nodes in the target network.

[0058] Generally speaking, lower-order adjacent nodes of a node are closer in space and have a stronger connection, while higher-order adjacent nodes of a node are farther in space and have a weaker connection.

[0059] When checking if there are new connections between nodes, first analyze the communication events. With the changing node x m The former second-order adjacency node {x 2 Whether Granger causality exists, further analysis of the communication events. With the changing node x m The former third-order adjacency node {x 3 Whether a Granger causality exists is determined by traversing to the highest-order adjacent node in the target communication network.

[0060] Example 1:

[0061] A method for tracking the topology of a non-cooperative wireless communication network includes the following steps:

[0062] Step 1: The sensing network continuously receives data from each node x in the target wireless communication network. i signal {S i (t)};

[0063] Step 2: At the same time T power Within, calculate x for each node in the target wireless communication network. i signal energy When the topology of the target wireless communication network remains stable, record the x of each node. i average signal energy like Then node x i Let x be the node that changes. m Steps 3 to 9 are executed to track the topology of the target wireless communication network.

[0064] Step 3: Obtain the changing node x in the target wireless communication network m Former adjacent node {x 1}, extract T granger Duration comes from node {x 1} radiation signal At equal intervals from Extracting communication events

[0065] Step 4: Select a frequency point with no valid signal and no interference signal to receive T. power Calculate the pure noise energy of a duration signal.

[0066] If the node x changes m signal energy Greater than pure noise energy Then start from the changing node x m Original frequency Extract communication event S m (n), for S m (n) and Perform Granger analysis, if S m (n) and If Granger causality exists, the topology of the target wireless communication network is determined to remain unchanged; otherwise, proceed to step 5.

[0067] If the node x changes m signal energy Less than or equal to pure noise energy Then proceed directly to step 5;

[0068] Step 5: From the lowest frequency point f low Let the current frequency f begin analysis. p The lowest frequency point f low ;

[0069] Step 6: Extract T granger The current frequency point f of duration p signal At equal intervals from Extracting communication events

[0070] Step 7: [Regarding...] and Perform Granger analysis; if and If Granger causality exists, then the frequency point f p For node x m The new frequency point is denoted as Proceed to step 8; otherwise, switch to the next frequency and return to step 6.

[0071] If until f p For the highest frequency point f high Node x was still not detected. m If the frequency is new, it is determined that the topology of the target wireless communication network remains unchanged;

[0072] Step 8: Extract T granger Frequency of duration Signal At equal intervals from Extracting communication events

[0073] Step 9: From node x m The second-order adjacency nodes are traversed to x m The set of the highest-order adjacent nodes in the target wireless communication network topology is denoted as x. n ={x 2 ,x 3 ...x n}, x n x represents m The nth order of adjacent nodes; analysis during traversal. With x n China Communications Event S n Granger causality between (t) and node x, discovering the node x m Are there any new connections? If node x is found during the traversal... m If a new communication relationship is established, the target wireless communication network topology is immediately updated, and set x is also updated. n .

[0074] The specific steps of Granger analysis are as follows: (Based on step 7...) and Taking Granger analysis as an example, the specific steps are as follows:

[0075] Step 7.1: For Perform linear autoregression;

[0076]

[0077]

[0078] Where τ is the autoregression order; ε1(n) and ε0(n) are the residual noise of the regression fit; a 1,k ,b 1,k ,a0,k These are the regression parameters of the autoregressive model;

[0079] Step 7.2: Calculation In autoregressive models Sum of squared residuals calculate In autoregressive models Sum of squared residuals

[0080] Step 7.3: Calculation As a measure of causality, Indicates by Participation in inference Granger causality in the case of autoregression, where var(R1) and var(R0) are the variances of multiple R1 and R0 values ​​obtained from multiple autoregressions;

[0081] Step 7.4: Repeat steps 7.1 to 7.2 for calculation. With S noise (n) The value is used as a threshold for determining connectivity; where S noise (n) represents the pure noise S noise Communication events extracted at equal intervals in (t);

[0082] Step 7.5: If the current frequency point f p Calculated Value greater than Then determine the current frequency point f p It is the changing node x m The new transmission signal frequency; otherwise, determine the current frequency f. p and changing node x m No relation.

[0083] Example 2:

[0084] 1. The sensing system receives data from each target network communication node x. i signal {S i (t)}, see appendix Figure 3 -S110;

[0085] 2. Calculate the signal energy to determine if the signal has decreased; see appendix. Figure 3 -S120, the specific steps are steps 3 to 7;

[0086] 3. Calculate the same duration T power Signals S of each node i Energy of (t)

[0087] 4. Select a frequency point with no valid signals or interference signals to receive T.power Duration signal used to calculate pure noise energy

[0088] 5. Record the x of each node when the target communication network topology remains stable. i Mean signal energy

[0089] 6. When formally detecting changes in the target communication network topology, compare... and Numerical size;

[0090] 7. If the current signal energy Significantly smaller than in the past Then the subsequent tracking process will be initiated;

[0091] 8. Match the current communication events of previously adjacent nodes of the changed node across the entire frequency band of the target communication network, as shown in the appendix. Figure 3 -S201, the specific steps are steps 9 to 27;

[0092] 9. Select past nodes x m Nodes {x} with connectivity 1};

[0093] 10. Extract T granger Duration comes from node {x 1} radiation signal

[0094] 11. With equal intervals of time T s from Extracting communication events

[0095] 12. If Greater than pure noise energy First extract the communication event S. m (n) Analysis and The Granger relationship between them is attached. Figure 3 -S202 to Appendix Figure 3 -S203;

[0096] 13. Regarding S m (n) using the formula Perform linear autoregression;

[0097] 14. Then, regarding S... m (n) using the formula Perform linear autoregression; where τ is the autoregression order, ε1(n), ε0(n) are the residual noise of the regression fit, and a 1,k ,b 1,k ,a 0,kThese are the regression parameters of the autoregressive model;

[0098] 15. Calculate in S m (n) in the autoregressive model Sum of squared residuals

[0099] 16. Calculate in S m (n) in the autoregressive model Sum of squared residuals

[0100] 17. with As a measure of causality, Indicates by Participating in inference S m Granger causality in case (n), where var(R1) and var(R0) are the variances of multiple R1 and R0 values ​​from multiple autoregressions;

[0101] 18. Regarding communication events With noise S noise Repeat steps 11 to 15 between (n) to obtain the threshold.

[0102] 19. Regarding communication events Communication events with its connected nodes Repeat steps 11 through 15 to obtain reference values.

[0103] 20. Comparison and The size between, if The value is much greater than Explain node x m With node x 1 There is still a communication relationship between them, target communication network node x m The network topology remains unchanged while employing silent methods such as reducing the number of communications or lowering the power of transmitted signals to evade surveillance.

[0104] 21. Otherwise, in step 10 Greater than pure noise energy This is caused by interference signals generated by other radiation sources;

[0105] 22. Proceed to step 10, if Approximate pure noise energy Explain node x m The original frequency point had no signal transmission; at this time, the target communication network could communicate across the entire frequency band f. low →f high Find node x on i New frequency, flow It must be less than the known minimum frequency of the target communication network node, f high It must be greater than the highest known frequency of the target communication network nodes, see appendix. Figure 3 -S204;

[0106] 23. From f low Start capturing T granger Duration and current frequency f p Signal See attached Figure 3 -S205;

[0107] 24. With equal intervals of time T s from Extracting events See attached Figure 3 -S206;

[0108] 25. Regarding the event Communication events Repeat steps 11 through 15 to obtain...

[0109] 26. If there exists any Much larger Explain the frequency point f p For node x m The new frequency is denoted as Otherwise, switch to the next frequency point f. p+1 See attached Figure 3 -S208, repeat steps 23 to 25;

[0110] 27. If f p =f high Node x was still not detected. m New frequency point description node x m They adopted a silent approach by ceasing to transmit signals to evade surveillance;

[0111] 28. Extract T granger Duration and Frequency Signal

[0112] 29. Extracting Communication Events See attached Figure 3 -S301;

[0113] 30. From node x m The second-order adjacency nodes are traversed to x m The set of the highest-order adjacent nodes in the target network topology is denoted as x. n ={x 2 ,x 3 ...x n}, x n x represents m The nth order adjacent nodes, see appendix. Figure 3 -S302 and its appendices Figure 4 ;

[0114] 31. Analysis during traversal With x n China Communications Event S n Granger causality between (t) and node x, discovering the node x m See attached document for details regarding any new communication relationships. Figure 3 -S303;

[0115] 32. During the traversal, node x was found. m If a new communication relationship is established, the target communication network topology W will be updated immediately. uu' For W′ uu' and update set x n For (x) n (See attached) Figure 3 -S304;

[0116] 33. According to (x) n Repeat steps 28 to 29.

[0117] The advantages of this invention are: 1. It is applicable to non-cooperative scenarios and time-varying communication networks; 2. It can quickly find the new transmission frequency of a node when the target wireless communication network node's signal disappears; 3. The tracking of target wireless communication network nodes uses a linear autoregressive Granger causality analysis method, which is simpler and more efficient than modulation identification and radiation source identification using nonlinear mapping; 4. Using Granger causality analysis to match signals is less susceptible to interference compared to modulation identification and radiation source identification methods; 5. It can efficiently discover new communication relationships that may exist between changing nodes.

[0118] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for tracking the topology of a non-cooperative wireless communication network, characterized in that, Includes the following steps: Step 1: The sensing network continuously receives data from each node in the target wireless communication network. signal ; Step 2: At the same time Internally, it calculates the number of nodes in the target wireless communication network. signal energy When the topology of the target wireless communication network remains stable, record the data of each node. average signal energy ;like Then the node denoted as the change node Steps 3 to 9 are executed to track the topology of the target wireless communication network. Step 3: Obtain changing nodes in the target wireless communication network Former adjacent nodes , cut Duration comes from nodes radiation signal At equal intervals from Extracting communication events ; Step 4: Select a frequency point with no valid signals or interference signals to receive signals. Calculate pure noise energy from duration signals. ; If the node changes signal energy Greater than pure noise energy Then start with the changing nodes. Original frequency Extracting communication events ,right and Perform Granger analysis, if and If Granger causality exists, then the topology of the target wireless communication network is determined to remain unchanged. Otherwise, proceed to step 5; If the node changes signal energy Less than or equal to pure noise energy If so, proceed directly to step 5; Step 5: From the lowest frequency point Begin analysis, setting the current frequency point The lowest frequency ; Step 6: Extract Current frequency of duration signal At equal intervals from Extracting communication events ; Step 7: [Regarding...] and Perform Granger analysis; if and If Granger causality exists, then the frequency point For nodes The new frequency point is denoted as Proceed to step 8; otherwise, switch to the next frequency and return to step 6. If until The highest frequency point No node was detected at that time. If the frequency is new, it is determined that the topology of the target wireless communication network remains unchanged; The pair and The specific steps for performing Granger analysis are as follows: Step 7.1: For Perform linear autoregression; in, The order of autoregression; This represents residual noise from the regression fitting. These are the regression parameters of the autoregressive model; Step 7.2: Calculation In autoregressive models Sum of squared residuals ;calculate In autoregressive models Sum of squared residuals ; Step 7.3: Calculation As a measure of causality, Indicates by Participation in inference Granger causality in the case of and For multiple autoregression processes and The variance; Step 7.4: Repeat steps 7.1 to 7.2 for calculation. and of The value serves as a threshold for determining connectivity; where, To reduce pure noise Communication events extracted at equal intervals; Step 7.5: If the current frequency point Calculated Value greater than Then determine the current frequency point It is a change node The new transmission frequency; otherwise, the current frequency is determined. and change nodes No relation; Step 8: Extract Frequency of duration Signal At equal intervals from Extracting communication events ; Step 9: From node The second-order adjacency nodes are traversed to The set of the highest-order adjacent nodes in the target wireless communication network topology is denoted as . , express of Rank adjacency nodes; analysis during traversal and communication events Granger causality between them, discovering nodes Are there any new connections? If a node is found during the traversal... If a new communication relationship is found, the target wireless communication network topology is immediately updated, and the set is also updated. .

Images