Channel knowledge graph construction method of non-cooperative interference source

Abstract

The invention provides a channel knowledge graph construction method of a non-cooperative interference source, which belongs to the field of communication and comprises the following steps: measuring received signal strength at a position # imgabs1 # according to a sampling proportion # imgabs0 #, obtaining a sampling channel graph # imgabs2 #, further obtaining a base station channel graph estimation result # imgabs3 #, obtaining a base station sampling channel graph # imgabs4 # with linear power and a sampling channel graph # imgabs5 # with linear power, and obtaining a base station channel graph estimation result # imgabs3 # with linear power; the method comprises the following steps of: obtaining an interference source channel sampling spectrum estimation result # imgabs6 # of linear power, inputting an obtained negative power indication matrix # imgabs7 # and an interference source channel spectrum reconstruction input # imgabs8 # into a deep learning network model # imgabs9 #, obtaining a normalized interference source channel spectrum estimation result # imgabs10 #, obtaining an interference source position # imgabs12 # according to # imgabs11 #, and obtaining a linear power interference source channel sampling spectrum estimation result # imgabs12 # according to the interference source position # imgabs12 #. And further obtaining an interference source channel knowledge graph estimation result # imgabs13 #, calculating SINRs of all positions based on # imgabs14 # and # imgabs15 #, and obtaining an SINR map # imgabs16 #. According to the method, the channel knowledge graph is accurately constructed under the interference condition that the non-cooperative interference source exists, and the accurate positioning of the interference source is realized.

Description

Technical Field

[0001] The present invention belongs to the field of communications, and in particular relates to a method for constructing a channel knowledge graph of a non-cooperative interference source. Background Art

[0002] With the rapid development of Unmanned Aerial Vehicles (UAV) technology, UAV-assisted communication technology has effectively promoted the 6G network to provide wide coverage and multi-functional communication services with its advantages of high flexibility, strong maneuverability and low manufacturing cost. It is worth noting that the channel environment of the aerial network is very different from that of the ground network. The traditional transmission method based on channel state information feedback is difficult to adapt to the complex and fast-changing channel environment of the UAV network. The channel map establishes a mapping relationship between the device location and the channel state, and its effectiveness has been verified in applications such as UAV network spectrum resource management and aerial base station deployment.

[0003] Existing channel map construction methods are mainly based on the accurate channel status reported by cooperative communication links, relying on the actual received signal strength of the communication link sampled at different locations, and constructing a complete channel map based on spatial relationships or interpolation methods. However, in actual situations, the presence of non-cooperative interference sources will reduce the accuracy of the measurement, resulting in reduced accuracy in channel map construction. Therefore, how to construct a channel map under interference conditions and further locate the interference source is an open problem that requires further research. Summary of the invention

[0004] The purpose of the present invention is to propose a channel knowledge graph construction method for non-cooperative interference sources, to achieve accurate construction of the channel knowledge graph under the interference conditions where non-cooperative interference sources exist, and to achieve accurate positioning of the interference sources.

[0005] The present invention is achieved through the following technical solutions: A method for constructing a channel knowledge graph of a non-cooperative interference source comprises the following steps: Step S1: According to the sampling ratio , at position Measure the received signal strength and obtain the sampling channel spectrum ,in, , is the set of sampling locations; Step S2: Obtain all locations based on environmental information Base station location , determine the communication link LoS / NLoS status , and according to Estimated location Base station receiving signal strength , and then get the base station channel spectrum estimation result ,in, A location set representing the area covered by the constructed channel knowledge graph; Step S3: Base station channel spectrum estimation result According to the sampling position set Sampling to obtain the base station sampling channel spectrum , the base station sampling channel spectrum and sampling channel spectrum Converted into linear power to obtain the linear power base station sampling channel spectrum and linear power of the sampled channel spectrum , and then obtain the linear power interference source channel sampling spectrum estimation result ; Step S4: Estimation result of interference source channel sampling spectrum based on linear power Get the interference source channel sampling spectrum estimation result , Negative Power Indicator Matrix and interference source channel spectrum reconstruction input ; Step S5: Negative power indicator matrix and interference source channel spectrum reconstruction input Input to the deep learning network model , after interpolation, the normalized interference source channel spectrum estimation result is obtained ; Step S6: Detect the normalized interference source channel spectrum estimation result based on the peak detection algorithm The peak position in the interference source is obtained ; Step S7: Estimation result based on interference source channel sampling spectrum And the normalized interference source channel spectrum estimation results , estimate the maximum and minimum values ​​of the interference source channel knowledge graph, and obtain the interference source channel knowledge graph estimation result ; Step S8: Estimation result based on base station channel spectrum And the interference source channel knowledge graph estimation results Calculate the SINR at all locations and get the SINR map .

[0006] Furthermore, in step S1, the position Represents the three-dimensional position coordinates in a square area. The length, width and height of the square area are , at position The received signal strength measured at , by measuring All locations within , while the position Corresponding Set to 0 to get the sampling channel spectrum ,in, is the transmission power of the base station, It is n The transmission power of an interference source, N is the number of interference sources, Indicates n The location of the interference source, Indicates the location of the base station The transmitter to the location The channel gain, the logarithmic calculation expression is , represent the path loss exponent and path loss intercept respectively, is the indicator variable of direct link and indirect link, represents the channel shadow fading due to unpredictable scatterers, Represents the channel fading caused by multipath transmission.

[0007] Furthermore, in step S2, the base station receives a signal strength Expressed as , All locations within Form the base station channel spectrum estimation result ,in, , represents the estimated result of the base station received signal power at the location, They are The estimated value of .

[0008] Furthermore, the step S3 specifically includes the following steps: Step S31: In the base station channel spectrum estimation result According to the sampling position set Sampling to obtain the base station sampling channel spectrum , Middle position The element value at The sampling results of the corresponding position in the position Set the element value at 0; Step S32: According to the formula The base station sampling channel spectrum Linearization to obtain the linear power base station sampling channel spectrum , according to the formula Sampling channel spectrum Convert to linear power and get the sampling channel spectrum of linear power , , , where linear power conversion is only for position The corresponding received signal power is carried out, and the position The corresponding sampling result is set to 0. Represents the linear power conversion calculation process; Step S33: According to the formula Calculate the received signal power of the interference source by calculating All locations within , while the position Corresponding Set to 0 to obtain the interference source channel sampling spectrum estimation result of linear power ,in, For location The actual received signal power sampling result is For location The base station received signal power estimation result.

[0009] Furthermore, the step S4 specifically includes the following steps: Step S41: Estimation result of linear power interference source channel sampling spectrum Logarithmization is performed to obtain the interference source channel sampling spectrum estimation result and negative power indicator matrix , where the linear power interference source channel sampling spectrum estimation result is Negative numbers and 0 values ​​in the interference source channel are not logarithmized. The corresponding position is 0, , negative power indicator matrix , corresponding to the set of sampling positions The elements in the positions other than 0 are set to 0, and the other elements correspond to The position of is 0, corresponding to The position of is 1; Step S42: Estimation result of interference source channel sampling spectrum The non-zero values ​​in are normalized to the maximum and minimum values ​​to obtain the interference source channel spectrum reconstruction input .

[0010] Furthermore, in step S5, the deep learning network model It is a convolutional neural network.

[0011] Furthermore, in step S6, the peak detection algorithm is a two-bit constant false alarm probability detection algorithm.

[0012] Furthermore, the step S7 specifically includes the following steps: Step S71: Vectorized interference source channel sampling spectrum estimation result The non-zero value in is represented by ; Step S72: normalize the interference source channel spectrum estimation result The sampling spectrum estimation result of the interference source channel is The elements of the non-zero positions in are vectorized and expressed as ; Step S73: construct the maximum value of the interference source channel knowledge graph and minimum value The estimation optimization problem is expressed as , solving the estimation optimization problem yields ,in, Representation vector The module length, ; Step S74: Denormalize the normalized interference source channel spectrum estimation result , and obtain the interference source channel knowledge graph estimation result The specific calculation expression is ,in, for In Location The element value of For Location The element value of .

[0013] Furthermore, in step S8, the position The SINR is expressed as , is the noise power.

[0014] The present invention has the following beneficial effects: 1. The present invention first obtains a sampling channel spectrum And the base station channel spectrum estimation results , and then the sampling channel spectra are And the base station channel spectrum estimation results Linearization to obtain the sampling channel spectrum of linear power Base station sampling channel spectrum with linear power , and then obtain the linear power interference source channel sampling spectrum estimation result , and will Get the negative power indicator matrix and interference source channel spectrum reconstruction input Input deep learning network model , after interpolation, the normalized interference source channel spectrum estimation result is obtained Then, the normalized interference source channel spectrum estimation result is detected based on the peak detection algorithm The peak position in the interference source is obtained , and based on the interference source channel sampling spectrum estimation result And the normalized interference source channel spectrum estimation results , estimate the maximum and minimum values ​​of the interference source channel knowledge graph, and obtain the interference source channel knowledge graph estimation result , and get the SINR map . In order to address the pain point that it is impossible to accurately measure the received signal power in the presence of non-cooperative interference sources, a channel knowledge graph can be accurately constructed under the interference conditions of non-cooperative interference sources, and the interference sources can be accurately located and the interference power can be accurately estimated, and the construction of SINR maps can be supported. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention is further described in detail below with reference to the accompanying drawings.

[0016] Figure 1 It is a flow chart of the present invention.

[0017] Figure 2 are the network parameters of the deep learning network model of the present invention.

[0018] Figure 3 A simulation diagram showing how the error in constructing the knowledge graph of the present invention changes with the sampling ratio.

[0019] Figure 4 This is a simulation diagram of the interference source positioning error changing with the sampling ratio of the present invention.

[0020] Figure 5 It is a simulation diagram of the variation of SINR map construction error with sampling ratio according to the present invention. DETAILED DESCRIPTION

[0021] This embodiment considers a downlink communication system model in which a UAV is used as a receiving end in an urban environment, and there are non-cooperative interference nodes at unknown locations that interfere with UAV communications. Figure 1 As shown, the channel knowledge graph construction method of the non-cooperative interference source is characterized by comprising the following steps: Step S1: According to the sampling ratio , at position Measure the received signal strength and obtain the sampling channel spectrum ,in, , is the set of sampling locations; Specifically, the location Represents the three-dimensional position coordinates in a square area. The length, width and height of the square area are , at position The received signal strength measured at , by measuring All locations within , while the position Corresponding Set to 0 to get the sampling channel spectrum ,in, represents the location set of the area covered by the constructed channel knowledge graph, represents the channel spectrum coverage area at a fixed height, where is the length of the region, is the width of the region, is the transmission power of the base station, It is n The transmission power of an interference source, N is the number of interference sources, Indicates n The location of the interference source, Indicates the location of the base station The transmitter to the location The channel gain, the logarithmic calculation expression is , represent the path loss exponent and path loss intercept respectively, Indicates the variables for Line-of-Sight (LoS) and Non-Line-of-Sight (NLoS), k =1 represents LoS, k =0 represents NloS, It represents the channel shadow fading caused by unpredictable scatterers, and the variance is The standard Gaussian process is used to characterize Represents the channel fading caused by multipath transmission, and the variance is is characterized by a standard Gaussian process.

[0022] In this embodiment, the base station transmit power ; Number of interference sources , the interference source transmission power ; The interference source is located at The channel parameters are defined as: , ; ; .

[0023] Step S2: Obtain all locations based on environmental information Base station location , determine the communication link LoS / NLoS status , and according to Estimated location Base station receiving signal strength , and then get the base station channel spectrum estimation result ,in, A location set representing the area covered by the constructed channel knowledge graph; Base station receiving signal strength Expressed as , by estimating All locations within , the base station channel spectrum estimation result can be obtained ,in, , represents the estimated result of the base station received signal power at the location, They are The estimated value of .

[0024] In this embodiment, the environment information is considered to be an urban environment, where buildings are modeled as squares and the building distribution is determined by the parameters Decide, They are specifically defined as: a represents the proportion of area covered by buildings, b Indicates the number of buildings per square kilometer, λ Represents the Rayleigh distribution parameter of the building height, and its specific value is .

[0025] Step S3: Base station channel spectrum estimation result According to the sampling position set Sampling to obtain the base station sampling channel spectrum , the base station sampling channel spectrum and sampling channel spectrum Converted into linear power to obtain the linear power base station sampling channel spectrum and linear power of the sampled channel spectrum , and then obtain the linear power interference source channel sampling spectrum estimation result , specifically including the following steps: Step S31: In the base station channel spectrum estimation result According to the sampling position set Sampling to obtain the base station sampling channel spectrum , Middle position The element value at The sampling results of the corresponding position in the position Set the element value at 0; Step S32: According to the formula The base station sampling channel spectrum Convert to linear to get the linear power base station sampling channel spectrum , according to the formula Sampling channel spectrum Convert to linear power and get the sampling channel spectrum of linear power , , , where linear power conversion is only for position The corresponding received signal power is carried out, and the position The corresponding sampling result is set to 0. Represents the linear power conversion calculation process; Step S33: According to the formula Calculate the received signal power of the interference source by calculating All locations within , while the position Corresponding Set to 0 to obtain the interference source channel sampling spectrum estimation result of linear power ,in, For location The actual received signal power sampling result is For location The base station received signal power estimation result.

[0026] Step S31: In the base station channel spectrum estimation result According to the sampling position set Sampling to obtain the base station sampling channel spectrum ,exist Middle, Location The corresponding sampling results are The corresponding element value, position The corresponding sampling result is set to 0; Step S32: According to the formula The base station sampling channel spectrum Convert to linear to get the linear power base station sampling channel spectrum , according to the formula Sampling channel spectrum Convert to linear power and get the sampling channel spectrum of linear power , , , where linear power conversion is only for position The corresponding received signal power is carried out, and the position The corresponding sampling result is set to 0. Represents the linear power conversion calculation process; Step S33: According to the formula Calculate the received signal power of the interference source by calculating All locations within , while the position Corresponding Set to 0 to obtain the interference source channel sampling spectrum estimation result of linear power ,in, For location The actual received signal power sampling result is For location The base station receives the signal power.

[0027] Step S4: Estimation result of interference source channel sampling spectrum based on linear power Get the interference source channel sampling spectrum estimation result , Negative Power Indicator Matrix and interference source channel spectrum reconstruction input , specifically including the following steps: Step S41: Estimation result of linear power interference source channel sampling spectrum Logarithmization is performed to obtain the interference source channel sampling spectrum estimation result and negative power indicator matrix , where the linear power interference source channel sampling spectrum estimation result is Negative numbers and 0 values ​​in the interference source channel are not logarithmized. The elements at the corresponding positions are 0, and the negative power indicator matrix , corresponding to the set of sampling positions The elements in the positions other than 0 are set to 0, and the other elements correspond to The position of is 0, corresponding to The position of is 1; Step S42: Estimation result of interference source channel sampling spectrum The non-zero values ​​in are normalized to the maximum and minimum values ​​to obtain the interference source channel spectrum reconstruction input ; The specific calculation expression is: ,in, Represents the interference source channel sampling spectrum estimation result The set of 0 elements in .

[0028] Step S5: Negative power indicator matrix and interference source channel spectrum reconstruction input Input to the deep learning network model , after interpolation, the normalized interference source channel spectrum estimation result is obtained ; Specifically, the deep learning network model is a convolutional neural network, and its network parameters are as follows Figure 2 shown.

[0029] Step S6: Detect the normalized interference source channel spectrum estimation result based on the peak detection algorithm The peak position in the interference source is obtained ; Specifically, the peak detection algorithm is a known two-bit constant false alarm probability detection algorithm.

[0030] Step S7: Estimation result based on interference source channel sampling spectrum And the normalized interference source channel spectrum estimation results , estimate the maximum and minimum values ​​of the interference source channel knowledge graph, and obtain the interference source channel knowledge graph estimation result ; The specific steps include: Step S71: Vectorized interference source channel sampling spectrum estimation result The non-zero value in is represented by ; Step S72: normalize the interference source channel spectrum estimation result The sampling spectrum estimation result of the interference source channel is The elements of the non-zero positions in are vectorized and expressed as ; Step S73: construct the maximum value of the interference source channel knowledge graph and minimum value The estimation optimization problem is expressed as , solving the estimation optimization problem yields ,in, Representation vector The module length, ; Step S74: Denormalize the normalized interference source channel spectrum estimation result , and obtain the interference source channel knowledge graph estimation result The specific calculation expression is ,in, for In Location The element value of For Location The element value of .

[0031] Step S8: Estimation result based on base station channel spectrum And the interference source channel knowledge graph estimation results Calculate the SINR (Signal-to-Interference-Rate) of all locations and get the SINR map ; Among them, the location The SINR is expressed as , is the noise power.

[0032] Figure 3 The horizontal axis is the sampling ratio, and the vertical axis is the estimation result of the interference source channel knowledge graph constructed. Compared with the normalized mean square error of the true value, it can be seen that the error of the present invention is smaller than that of the Kriging method (Krige) and the inverse distance weighted method (IDW); Figure 4 The horizontal axis is the sampling ratio, and the vertical axis is the estimated interference source location. Compared with the normalized mean square error of the true value, it can be seen that the error of the present invention is smaller than that of the Kriging method (Krige) and the inverse distance weighted method (IDW); Figure 5 The horizontal axis is the sampling ratio, and the vertical axis is the normalized mean square error between the reconstructed SINR map and the true value. It can be seen that compared with the Kriging method (Krige) and the inverse distance weighted method (IDW), the error of the present invention is smaller.

[0033] The above description is only a preferred embodiment of the present invention, and therefore cannot be used to limit the scope of implementation of the present invention. That is, equivalent changes and modifications made according to the scope of the patent application of the present invention and the contents of the specification should still fall within the scope covered by the patent of the present invention.

Claims

1. A channel knowledge graph construction method for non-cooperative interference sources, characterized in that: The steps include: Step S1: According to the sampling ratio , at position Measure the received signal strength and obtain the sampling channel spectrum ,in, , is the set of sampling locations; Step S2: Obtain all locations based on environmental information With base station location , determine the communication link LoS / NLoS status , and according to Estimated location Base station receiving signal strength , and then get the base station channel spectrum estimation result ,in, A location set representing the area covered by the constructed channel knowledge graph; Step S3: Base station channel spectrum estimation result According to the sampling position set Sampling to obtain the base station sampling channel spectrum , the base station sampling channel spectrum and sampling channel spectrum Converted into linear power to obtain the linear power base station sampling channel spectrum and linear power of the sampled channel spectrum , and then obtain the linear power interference source channel sampling spectrum estimation result ; Step S4: Estimation result of interference source channel sampling spectrum based on linear power Get the interference source channel sampling spectrum estimation result , Negative Power Indicator Matrix and interference source channel spectrum reconstruction input ; Step S5: Negative power indicator matrix and interference source channel spectrum reconstruction input Input to the deep learning network model , after interpolation, the normalized interference source channel spectrum estimation result is obtained ; Step S6: Detect the normalized interference source channel spectrum estimation result based on the peak detection algorithm The peak position in the interference source is obtained ; Step S7: Estimation result based on interference source channel sampling spectrum And the normalized interference source channel spectrum estimation results , estimate the maximum and minimum values ​​of the interference source channel knowledge graph, and obtain the interference source channel knowledge graph estimation result ; Step S8: Estimation result based on base station channel spectrum And the interference source channel knowledge graph estimation results Calculate the SINR at all locations and get the SINR map .

2. The method for constructing a channel knowledge graph of a non-cooperative interference source according to claim 1, characterized in that: In step S1, the position Represents the three-dimensional position coordinates in a square area. The length, width and height of the square area are , at position The received signal strength measured at , by measuring All locations within , while the position Corresponding Set to 0 to get the sampling channel spectrum ,in, is the transmission power of the base station, It is n The transmission power of an interference source, N is the number of interference sources, Indicates n The location of the interference source, Indicates the location of the base station The transmitter to the location The channel gain, the logarithmic calculation expression is , represent the path loss exponent and path loss intercept respectively, is the indicator variable of direct link and indirect link, represents the channel shadow fading due to unpredictable scatterers, Represents the channel fading caused by multipath transmission.

3. The method for constructing a channel knowledge graph of a non-cooperative interference source according to claim 2, characterized in that: In step S2, the base station receives the signal strength Expressed as , All locations within Form the base station channel spectrum estimation result ,in, , represents the estimated result of the base station received signal power at the location, They are The estimated value of .

4. The method for constructing a channel knowledge graph of a non-cooperative interference source according to claim 3, characterized in that: The step S3 specifically includes the following steps: Step S31: In the base station channel spectrum estimation result According to the sampling position set Sampling to obtain the base station sampling channel spectrum , Middle position The element value at The sampling results of the corresponding position in the position Set the element value at 0; Step S32: According to the formula The base station sampling channel spectrum Linearization to obtain the linear power base station sampling channel spectrum , according to the formula Sampling channel spectrum Convert to linear power and get the sampling channel spectrum of linear power , , , where linear power conversion is only for position The corresponding received signal power is carried out, and the position The corresponding sampling result is set to 0. Represents the linear power conversion calculation process; Step S33: According to the formula Calculate the received signal power of the interference source by calculating All locations within , while the position Corresponding Set to 0 to obtain the interference source channel sampling spectrum estimation result of linear power ,in, For location The actual received signal power sampling result is For location The base station received signal power estimation result.

5. The method for constructing a channel knowledge graph of a non-cooperative interference source according to claim 4, characterized in that: The step S4 specifically includes the following steps: Step S41: Estimation result of linear power interference source channel sampling spectrum Logarithmization is performed to obtain the interference source channel sampling spectrum estimation result and negative power indicator matrix , where the linear power interference source channel sampling spectrum estimation result is Negative numbers and 0 values ​​in the interference source channel are not logarithmized. The corresponding position is 0, , negative power indicator matrix , corresponding to the set of sampling positions The elements in the positions other than 0 are set to 0, and the other elements correspond to The position of is 0, corresponding to The position of is 1; Step S42: Estimation result of interference source channel sampling spectrum The non-zero values ​​in are normalized to the maximum and minimum values ​​to obtain the interference source channel spectrum reconstruction input .

6. The method for constructing a channel knowledge graph of a non-cooperative interference source according to claim 5, characterized in that: In step S5, the deep learning network model It is a convolutional neural network.

7. The method for constructing a channel knowledge graph of a non-cooperative interference source according to claim 6, characterized in that: In step S6, the peak detection algorithm is a two-bit constant false alarm probability detection algorithm.

8. The method for constructing a channel knowledge graph of a non-cooperative interference source according to claim 7, characterized in that: The step S7 specifically includes the following steps: Step S71: Vectorized interference source channel sampling spectrum estimation result The non-zero value in is represented by ; Step S72: normalize the interference source channel spectrum estimation result The sampling spectrum estimation result of the interference source channel is The elements of the non-zero positions in are vectorized and expressed as ; Step S73: construct the maximum value of the interference source channel knowledge graph and minimum value The estimation optimization problem is expressed as , solving the estimation optimization problem yields ,in, Representation vector The module length, ; Step S74: Denormalize the normalized interference source channel spectrum estimation result , and obtain the interference source channel knowledge graph estimation result The specific calculation expression is ,in, for In Location The element value of For Location The element value of .

9. The method for constructing a channel knowledge graph of a non-cooperative interference source according to claim 8, characterized in that: In step S8, the position The SINR is expressed as , is the noise power.

Images