CN108693403A - A kind of virtual densification frequency spectrum situation generation method of wide area - Google Patents

Abstract

The invention discloses a wide-area virtual intensive spectrum situation generation method, which mainly solves the problems of huge sensing equipment required in the prior art and low accuracy of electromagnetic situation inversion. Its technical solution is: 1. Determine and configure complex electromagnetic environment parameters; 2. Obtain sensing data through wide-area virtual densification; 3. Construct sensing node position matrix; 4. Construct path loss matrix; 5. According to sensing node position matrix, path The radiation source is identified by the loss matrix, and the location and radiation power of the radiation source are obtained; 6. According to the identified radiation source, the electromagnetic situation is inverted to generate the electromagnetic spectrum situation. The present invention can perform wide-area virtual densification on sensors under the condition that a small number of sensor positions are randomly distributed, and radiation source positions and radiation power are randomly distributed, so as to realize radiation source identification and then generate electromagnetic situation, which can be used for wide-area virtual densification spectrum situation generate.

Description

A Wide-Area Virtual Densification Spectrum Situation Generation Method

technical field

The invention belongs to the field of communication technology, and relates to spectrum sensing technology, radiation source identification, and electromagnetic spectrum situation, and further relates to a wide-area virtual intensive spectrum situation generation method, which can be used for high-precision spectrum situation generation in a wide-area environment.

Background technique

The rapid development and widespread use of wireless communication technologies have made the electromagnetic environment increasingly complex, spectrum resources increasingly scarce, and the contradiction between supply and demand of electromagnetic spectrum increasingly acute. Driven by the requirements of spectrum sharing, radio order management, and electromagnetic spectrum warfare in future mobile communication systems, electromagnetic spectrum situation awareness and intelligent spectrum management based on spectrum situation have become the core topics of research. The research on the spectrum situation is to map the complex electromagnetic environment into the information space to form a virtual electromagnetic spectrum space, which can be understood as the current state, comprehensive situation and development trend of the electromagnetic environment. Spectrum situational awareness refers to the comprehensive perception of the battlefield through monitoring and detection means, obtaining data, and processing the data to form spectrum information and situation that can be used by me. It is the main means to understand the spectrum situation in a timely and accurate manner. The main goal is to obtain the electromagnetic spectrum action space, working time, working frequency and radiation power, etc. The electromagnetic radiation in the electromagnetic environment mainly comes from various frequency-using equipment, that is, the radiation source. As long as the location, state, working parameters, signal characteristics and other attributes of the radiation source can be correctly obtained, the impact of the radiation source on the environmental area can be approximately calculated. electromagnetic effect. Therefore, constructing environmental data based on radiation source identification is an effective way for spectrum situational awareness.

Spectrum situation generation is to analyze and predict the comprehensive situation and future development trend of spectrum space based on the current state of spectrum space acquired by spectrum situation awareness. 1) The current electromagnetic spectrum detection and analysis system can only obtain the electromagnetic parameters of the location of the detection node, and it is difficult to form a comprehensive understanding of the electromagnetic situation of the entire detection area. 2) The actual measurement method carries out long-term on-site measurement of all locations within the detection coverage, which has the highest spectrum situation construction accuracy, but requires extremely dense detection nodes, heavy workload, long detection time, and is difficult to apply to large-scale detection networks. 3) The propagation model method takes the detection node as the hypothetical emission source, and the electromagnetic parameters measured by the detection node as the emission parameters, and calculates the electromagnetic parameters of any point in the surrounding space through Okumura model, Hata model and other classic electromagnetic propagation models, so as to construct the spectrum situation. The propagation model method has a small amount of calculation and a simple model. It only needs the data of one detection node to estimate the spectrum situation in the surrounding area, but the accuracy of the estimation result is poor. 4) The gridded electromagnetic spectrum broadband real-time detection and analysis system deploys a large number of dense micro-stations, which can effectively increase the coverage area of the electromagnetic spectrum detection function, and extract regular features from the measured data to construct the electromagnetic spectrum situation. However, to deploy detection nodes at fixed locations, it is necessary to cooperate with multiple surrounding detection nodes to participate in the construction of the spectrum situation. In order to ensure that the detection target is approached, the number of detection nodes is large. 5) The ray tracing method calculates the propagation characteristics of each beam on the building and the ground surface through numerical simulation, such as reflection and diffraction, so as to calculate the electromagnetic spectrum parameters of the surrounding locations, and can generate the electromagnetic situation more accurately. However, the ray tracing method is only suitable for simple propagation paths or short-distance communications. For complex electromagnetic environments, a large number of channel measurements are required, and detailed topographic and topographical data need to be mastered, and the amount of calculation is large. When the electromagnetic environment changes, it cannot generate electromagnetic waves in real time. situation. In summary, these methods for realizing electromagnetic situation generation have the following deficiencies:

(1) The modeling time is long, the workload is heavy, and it is difficult to apply to the situation deduction of the electromagnetic environment in a large area;

(2) The accuracy of spectrum situation generation is not high;

(3) It is necessary to deploy dense fixed detection nodes, and jointly participate in the construction of spectrum situation with multiple surrounding detection nodes, which requires the fixed deployment of a large number of sensing devices.

Contents of the invention

The purpose of the present invention is to overcome the above-mentioned shortcomings, and proposes a wide-area virtual dense spectrum situation generation method. The wide-area virtual intensive spectrum situation generation method attaches the spectrum sensor equipment to a small number of mobile bearing platforms such as vehicles and ships, and uses the mobility of the spectrum sensor equipment bearing platform to acquire the electromagnetic spectrum data of the current position multiple times within the monitoring duration. , realize dense virtualization of spectrum sensing nodes, increase spectrum monitoring samples, and then use spectrum situation sparse inversion theory to obtain wide-area high-resolution electromagnetic spectrum situation.

A kind of wide-area virtual intensive spectrum situation generation method proposed by the present invention comprises the following steps:

(1) Determine and configure complex electromagnetic environment parameters: the experimental area adopts N-point grid layout, and K radiation sources, T spectrum sensor devices and their supporting platforms are randomly distributed at the vertices of the grid in the experimental area. The platform is equipped with a GPS module, which is used to synchronously record the location information at the sensing node and the movement route of the spectrum sensor device and its carrying platform, and select the N grid vertices as N reference points.

(2) Wide-area virtual intensive acquisition of sensing data: the T spectrum sensor devices and their bearing platforms are virtualized into T mobile nodes, and each mobile node is moved to densify n sensing nodes for acquiring sensing data. The sensing data includes received signal strength RSS (Received Signal Strength) and location information. The T mobile nodes are co-densified into M=n T sensing nodes, and the received signal strength in the sensing data acquired by the M sensing nodes is RSS constitutes an M-dimensional column vector P _s ∈ R ^M .

(3) According to the position information in the sensing data, the sensing node position matrix is constructed, and the sensing node position matrix Φ can be expressed by the following formula:

Among them, S={s _k |k=1,2,...,M} represents the set of sensing nodes, s _k represents the kth sensing node, k is used to identify the kth sensing node, V={V _j |j=1,2,...,N} represents the set of all reference points, V _j represents the jth reference point, j is used to identify the jth reference point, s _k ∈ V _j represents the kth sensing node located at the jth reference point, Indicates that the kth sensing node is not located at the jth reference point, and the sensing node position matrix [Φ] _kj is an M*N matrix.

(4) According to the electromagnetic propagation model of the electromagnetic environment, construct the path loss matrix Ψ.

(5) Perform radiation source identification according to the sensing node position matrix and the path loss matrix, and obtain the position and radiation power of the radiation source.

(6) According to the identified radiation source, the electromagnetic situation is inverted, and the received signal strength RSS at the N reference points is obtained:

Among them, the column vector P _r ∈ R ^N represents the N-dimensional column vector composed of the received signal strength RSS on the N reference points, and the column vector P _t ∈ R ^N represents the N-dimensional column composed of the radiation power of the radiation source on the N reference points vector, Indicates the additive white Gaussian noise AWGN power.

In some embodiments, in the step (2), after each mobile node moves, it densifies n sensing nodes for acquiring sensing data, the sensing data includes received signal strength RSS and location information, including the following steps: each The mobile node randomly selects a reference point in the test area as a destination, and moves towards the destination. When the destination is reached, the mobile node is called the destination, that is, the sensing node at the reference point, Measure the received signal strength RSS and location information at the reference point to form sensing data, repeat the above steps until the mobile node passes through n reference points and acquires sensing data at the reference point, where n is called densification coefficient.

In some embodiments, in the step (4), according to the electromagnetic propagation model of the electromagnetic environment, the path loss matrix Ψ is constructed, including the following steps:

(4a) The propagation model of electromagnetic waves between the i-th reference point and the j-th reference point in two-dimensional free space is:

Among them, i,j∈{1,2,...,N}, i is used to identify the i-th reference point, j is used to identify the j-th reference point, and P _jr represents the received power of the j-th reference point; P _it represents the transmitting power of the i-th reference point; G _jr represents the receiving antenna gain of the j-th reference point, G _it represents the transmitting antenna gain of the i-th reference point; λ is the working wavelength of the electromagnetic wave; d _ij represents the i-th The distance between the transmitting antenna of the first reference point and the receiving antenna of the jth reference point, G _jr , G _it are all known constants, then the propagation model can be simplified as:

in, Indicates the loss coefficient between the i-th reference point and the j-th reference point.

(4b) The path loss matrix Ψ can be expressed by the following formula:

Wherein, the path loss matrix [Ψ] _ij is an N*N matrix.

In some embodiments, in the step (5), the radiation source is identified according to the sensing node position matrix and the path loss matrix, and the position and radiation power of the radiation source are obtained, including the following steps:

(5a) Calculate the sensing matrix Q according to the sensing node position matrix Φ and the path loss matrix Ψ:

Q＝ΦΨ

(5b) There is the following relationship between the M-dimensional column vector P _s formed by the received signal strength at M sensing nodes and the N-dimensional column vector P _t formed by the radiation power of the radiation source on N reference points:

in, is the additive Gaussian white noise AWGN power, the column vector P _t ∈ R ^N represents the N-dimensional column vector composed of the radiation power of the radiation source on N reference points, R ^N represents the N-dimensional vector space, and P _t ∈ R ^N represents P _t is an N-dimensional vector, the column vector ε∈RM represents the measurement error of the sensor, R ^M represents the ^M -dimensional vector space, and ε∈RM represents that ε is an ^M -dimensional vector.

(5c) Construct preprocessing data P _proc :

(5d) According to the minimum L1-norm, solve the N-dimensional column vector P _t formed by the position of the radiation source and the radiation power of the radiation source on N reference points:

min||P _t ||,st||P _proc -QP _t || ₂ ≤ μ

Among them, ||·|| represents the 1-norm, which means the sum of the moduli of all elements in the vector, and ||·|| ₂ represents the 2-norm, which means the sum of the squares of the moduli of all elements in the vector and then the square root , μ is the convergence accuracy, min means minimization, st is the abbreviation of subject to means "constrained to", the meaning of the whole equation is that under the condition of satisfying the constraint condition of ||P _proc -QP _t || ₂ ≤ μ, such that The sum of the moduli of all elements of P _t is the smallest, and there is a radiation source at the reference point corresponding to the non-zero elements of the N-dimensional column vector P _t , and its value represents the power of the radiation source.

The present invention has the following advantages:

1. The present invention adopts a wide-area virtual intensive spectrum situation generation method, and attaches spectrum sensor equipment to a small number of mobile bearing platforms such as vehicles and ships, which expands the spatial range of spectrum sensing and can be used for spectrum situation generation in large areas.

2. The present invention uses the received signal strength RSS measured by the sensor to identify the radiation source, and then generates the electromagnetic situation of the entire environment according to the electromagnetic environment propagation model, which improves the breadth and accuracy of situation generation.

3. In the generation of the electromagnetic situation, the experimental area also adopts an N-point grid layout. In the prior art, a sensor needs to be set at each grid vertex to measure the received signal strength at the grid vertex. However, the present invention adopts a wide The domain virtual intensive spectrum situation generation method only needs a very small number of sensors to realize electromagnetic situation generation, T is much smaller than N, and the invention can effectively reduce the number of sensing devices.

4. Since the present invention only needs a small number of samples (received signal strengths of M sensing nodes) for algorithm implementation, the calculation complexity is low and the time is short, which can meet the real-time requirements of electromagnetic situation inversion.

Description of drawings

Other features, objects, technical processes and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments with reference to the following drawings:

Fig. 1 is the realization flowchart of the present invention;

Fig. 2 is a simulation diagram of the moving route of the sensor device and its carrying platform;

Fig. 3 is under the present invention, the relative error between the identification of the radiation source and the radiation power at the corresponding position of the actual radiation source;

Fig. 4 is under the condition of K=8 radiation sources and T=10 sensors, radiation source identification performance diagram under different densification parameters;

Fig. 5 is under the condition of K=8 radiation sources, the radiation source identification performance of the present invention is compared with the radiation source identification performance under the fixed sensor condition;

Fig. 6 is a comparison diagram of the simulation of the radiation power of the actual radiation source and the radiation power of the identified radiation source under the present invention; a comparison diagram of the simulation of the actual electromagnetic situation and the generation of the electromagnetic situation.

Detailed ways

The invention is used for spectrum situation generation, performs wide-area virtual densification on sensors, realizes the measurement of received signal strength at grid vertices, preprocesses the received signal strength, realizes the identification of radiation sources, and finally generates electromagnetic situation.

Referring to Fig. 1, it shows the implementation flowchart 100 of the present invention, and the specific steps are as follows:

Step 101, determine and configure complex electromagnetic environment parameters.

The actual electromagnetic environment is ever-changing, and it is impossible to simulate it with a universally applicable and accurate mathematical model. To this end, the following reasonable assumptions and simplifications are made: the experimental area adopts an N-point grid layout. In this experimental area, a certain number of radiation sources need to be set in the experimental area according to the scale of the battlefield. Here, the number of radiation sources is set to K , K radiation sources are randomly distributed at N grid vertices, the number and position of radiation sources are unknown, the radiation power of radiation sources is randomly distributed, and the type of radiation sources is not limited, which can be communication equipment, jammers, transmitters one or more of the devices. In the experimental area, T mobile carrying platforms such as vehicles and ships are set up. This mobile platform is equipped with sensors and GPS (Global Positioning System, Global Positioning System) modules, which are used to simultaneously record the position information and spectrum sensor equipment at the sensing nodes and their carrying platforms. For the moving route, the number T of sensors is much smaller than the number N of grid vertices.

Step 102, wide-area virtual densification acquires perception data.

Wide-area virtual intensive acquisition of sensing data: initially T spectrum sensor devices and their bearing platforms are randomly distributed at the reference point of the experimental area, and the T spectrum sensor devices and their bearing platforms are virtualized into T mobile nodes , each mobile node densifies n sensing nodes after moving to obtain sensing data, the sensing data includes received signal strength RSS and location information, and the T mobile nodes densify M=n·T sensing nodes , the received signal strength RSS in the sensing data acquired by M sensing nodes constitutes an M-dimensional column vector P _s ∈ R ^M , where R ^M represents an M-dimensional vector space, and P _s ∈ R ^M represents that P _s is an M-dimensional of vectors.

Wherein, each mobile node densifies n sensing nodes after moving to obtain sensing data, the sensing data includes received signal strength RSS and location information, including the following steps:

Each mobile node randomly selects a reference point in the test area as a destination, and moves towards the destination. node, the spectrum sensor and GPS module of the sensing node respectively measure the received signal strength RSS and location information at the reference point to form the sensing data; after that, the mobile node continues to randomly select a reference point as a new destination and continues toward When moving to the destination and arriving at a new destination, measure the received signal strength RSS and location information of the new destination, that is, the reference point, and repeat the above steps until the mobile node passes through n reference nodes and obtains the location information of the reference node. Sensing data of , that is, a mobile node densifies n sensing nodes after moving, where n is called densification coefficient.

In some embodiments, initially each spectrum sensor device and its carrying platform, that is, a mobile node, is randomly distributed at a reference point in the experimental area, and the mobile node at the reference point can also be recorded as a sensing node, and measure the Received signal strength RSS and location information on the reference point.

Step 103, constructing a sensing node position matrix according to the position information in the sensing data.

It can be seen from step 102 that the GPS module is used to record the location information of each sensing node, and construct the sensing node location matrix according to the location information of M sensing nodes, then the sensing node location matrix Φ can be expressed by the following formula:

Among them, S={s _k |k=1,2,...,M} represents the set of sensing nodes, s _k represents the kth sensing node, k is used to identify the kth sensing node, V={V _j |j=1,2,...,N} represents the set of all reference points, V _j represents the jth reference point, j is used to identify the jth reference point, s _k ∈ V _j represents the kth sensing node located at the jth reference point, Indicates that the kth sensing node is not located on the jth reference point, and the sensing node position matrix [Φ] _kj is an M*N matrix.

Step 104, constructing a path loss matrix according to the electromagnetic propagation model of the electromagnetic environment.

Electromagnetic waves usually propagate in irregular and non-uniform environments. When estimating the path loss, it is necessary to consider the terrain and landforms on the propagation path, as well as obstacles such as buildings, trees, and utility poles. Choose a different path transfer model. Commonly used outdoor electromagnetic propagation models include Okumura model, Hata model, etc. The present invention uses a free-space path loss model.

(4a) The propagation model of electromagnetic waves between the i-th reference point and the j-th reference point in two-dimensional free space is:

Among them, i,j∈{1,2,...,N}, i is used to identify the i-th reference point, j is used to identify the j-th reference point, and P _jr represents the received power of the j-th reference point; P _it represents the transmitting power of the i-th reference point; G _jr represents the receiving antenna gain of the j-th reference point, G _it represents the transmitting antenna gain of the i-th reference point; λ is the working wavelength of the electromagnetic wave; d _ij represents the i-th The distance between the transmitting antenna of the first reference point and the receiving antenna of the jth reference point, G _jr , G _it are all known constants, then the propagation model can be simplified as:

in, Indicates the loss coefficient between the i-th reference point and the j-th reference point.

(4b) The path loss matrix Ψ can be expressed by the following formula:

Wherein, the path loss matrix [Ψ] _ij is an N*N matrix.

Step 105, identify the radiation source according to the sensing node position matrix and the path loss matrix, and obtain the position and radiation power of the radiation source.

Radiation source identification is performed according to the sensing node position matrix and path loss matrix, and the position and radiation power of the radiation source are obtained, including the following steps:

(5a) Calculate the sensing matrix Q according to the sensing node position matrix Φ and the path loss matrix Ψ:

Q＝ΦΨ

(5b) There is the following relationship between the M-dimensional column vector P _s formed by the received signal strength at M sensing nodes and the N-dimensional column vector P _t formed by the radiation power of the radiation source on N reference points:

in, is the additive white Gaussian noise AWGN (Additive White Gaussian Noise) power, the column vector P _t ∈ R ^N , R ^N represents the N-dimensional vector space, P _t ∈ R ^N represents the P _t is an N-dimensional vector, and the column vector ε∈ ^RM represents the measurement error of the sensor, ^RM represents the M-dimensional vector space, and ε∈RM represents ε as the ^M -dimensional vector.

(5c) Construct preprocessing data P _proc :

(5d) According to the minimum L1-norm, solve the N-dimensional column vector P _t formed by the position of the radiation source and the radiation power of the radiation source on N reference points:

min||P _t ||,st||P _proc -QP _t || ₂ ≤ μ

Among them, ||·|| represents the 1-norm, which means the sum of the moduli of all elements in the vector, and ||·|| ₂ represents the 2-norm, which means the sum of the squares of the moduli of all elements in the vector and then the square root , μ is the convergence accuracy, min means minimization, st is the abbreviation of subject to means "constrained to", the meaning of the whole equation is that under the condition of satisfying the constraint condition of ||P _proc -QP _t || ₂ ≤ μ, such that The sum of the moduli of all elements of P _t is the smallest, and there is a radiation source at the reference point corresponding to the non-zero elements of the N-dimensional column vector P _t , and its value represents the power of the radiation source.

Step 106, according to the identified radiation sources, the electromagnetic situation is inverted to obtain the received signal strengths RSS at the N reference points.

Calculate the vector P _r formed by the received signal strength RSS at the N reference points according to the following formula:

Among them, the column vector P _r ∈ R ^N represents the N-dimensional vector composed of the received signal strength RSS on the N reference points, that is, the received power on the N reference points, and the column vector P _t ∈ R ^N represents the radiation on the N reference points An N-dimensional vector composed of the radiation power of the source, Represents the AWGN power of additive white Gaussian noise, R ^N represents an N-dimensional vector space, and P _r ∈ R ^N , P _t ∈ R ^N represent P _r , P _t are N-dimensional vectors.

Effect of the present invention can be further illustrated by following simulation:

A. Simulation conditions

The experimental area is arranged on a 200m*200m square, which is divided into 20*20 grids, each grid has an area of 100m ² , and the total number of grid vertices is N=400. T sensors and K radiation sources are randomly distributed on 400 grid vertices. The above T sensors are respectively placed on T spectrum sensor equipment carrying platforms, and the spectrum sensor equipment carrying platforms are equipped with a GPS module, which is virtualized as a mobile node. During the simulation process, each mobile node randomly selects a reference point in the test area as the destination, and moves towards the destination. When the destination is reached, the received signal strength and position information on the destination, that is, the reference point, are measured. Afterwards, the mobile node continues to randomly select a reference point as a new destination and continues to move towards the destination. When arriving at the new destination, it measures the received signal strength at the new destination, that is, the reference point, and repeats. In the data acquisition time, assuming that each mobile node passes through n destinations, that is, a mobile node moves and densifies n sensing nodes, then T mobile nodes pass through M=n·T destinations in total, and M Received signal strength and location information at a reference point. The received signal strengths RSS of these M reference points constitute an M-dimensional column vector P _s ∈ R ^M . Assuming that the radiation frequency is 3MHz, the possible value of the radiation power is an integer multiple of P0, that is, the transmission power is randomly distributed in the power set {P0,2P0,...,P _m }, where P0 is the reference power, and P _m represents the power maximum value. The performance of radiated power reconstruction is expressed as relative error. When the radiation power is reconstructed, the number and position of radiation sources are unknown.

B. Simulation content and results

Simulation 1: In a certain experiment, sensor T=5, densification coefficient n=6, sensing nodes M=n·T=30, the simulation diagram of the moving route of sensor equipment and its carrying platform is shown in Figure 2.

It can be seen from Figure 2 that each mobile node moves and densifies 6 sensing nodes, M _ij represents the jth sensing node virtualized by the i-th sensor, and the mobile nodes move randomly, making the dense sensing nodes Spatial distribution is also random.

Simulation 2: Under the condition that the radiation power of K=8 radiation sources is randomly distributed in a power set, the recognition performance of the radiation source of the present invention is simulated. The performance of radiation source identification is expressed by the relative error PowE, which is calculated by taking the ratio of the absolute value of the difference between the real radiation power vector of the radiation source and the corresponding element of the identified radiation power vector to the reference radiation power P0:

Among them, P _t is an N*1-dimensional vector composed of the real radiation power of the radiation source, and P _t (i) represents the real radiation power of the radiation source on the i-th reference point, is an N*1-dimensional vector composed of the identified radiation power of the radiation source, Indicates the identified radiation power of the radiation source on the i-th reference point, and P0 is the reference power. In each experiment, the radiation source is randomly distributed on 400 grid vertices, the radiation power is randomly distributed in the power set, the sensor T=5, the densification coefficient n=20, the experiment is repeated 100 times, the identification of the radiation source Performance simulation results are shown in Figure 3.

It can be seen from Fig. 3 that in 100 tests, the relative error remained above 10 ⁻¹² , and the relative error was particularly small and relatively stable, indicating that the identification performance of the radiation source of the present invention is superior, and the location of the radiation source, sensor equipment and The moving path of the carrying platform and the radiation power of the radiation source are both robust.

Simulation 3: K=8 radiation sources are randomly distributed on 400 grid vertices, T=10 sensors are randomly distributed in the experimental area, and the radiation power is randomly distributed in the power set. In each experiment, the densification coefficient is changed, and the simulation results of the identification performance of the radiation source are shown in Fig. 4 .

It can be seen from Fig. 4 that the present invention carries out wide-area virtual densification of sensors, and the number of sensors remains unchanged. With the increase of the virtual densification coefficient, the relative error of identifying the radiation source becomes smaller and smaller, that is, the identification of the radiation source Performance just keeps getting better and better. When the densification coefficient n=4, that is, the number of virtual sensing nodes M=40, the relative error is already close to zero, that is, a very high radiation source identification performance has been achieved.

Simulation 4: If the sensors are not placed on the mobile platform, the sensors are randomly distributed at the vertices of the grid during initialization, and the positions of the sensors are fixed during the experiment, that is, under the condition of fixed sensors, only the reference points at T sensor positions can be measured The received signal strength is used as the received signal strength at the sensing node, and the number of sensing nodes is equal to the number of sensors at this time. In the present invention, K=8 radiation sources are randomly distributed on 400 grid vertices, the radiation power is randomly distributed in the power set, and the densification coefficient n=2. In each experiment, the number of sensors is changed, and the number of sensing nodes is also changed accordingly. The simulation results of radiation source recognition performance are shown in Fig. 5.

Curve 1 in Fig. 5 shows the radiation source identification performance under the condition of a fixed sensor, and curve 2 shows the radiation source identification performance of wide-area virtual densification.

It can be seen from Figure 5 that with the increase of the number of sensors, the relative error of radiation source identification under the condition of fixed sensors and wide-area virtual densification is getting smaller and smaller, that is, the performance of radiation source identification is getting better and better; Curve 1 and Curve 1 2. It can be seen from the comparison that when the number of sensors is the same, the radiation source recognition performance of the wide-area virtual intensive is higher than that of the fixed sensor, which reflects the superiority of the radiation source recognition of the present invention.

Simulation 5: Under simulation conditions, T=20 sensors, K=8 radiation sources are randomly distributed on 400 grid vertices, and the densification coefficient n=5. The radiation power of the identified radiation source and the inversion of the electromagnetic situation in the process of generating the wide-area virtual dense spectrum situation of the present invention are simulated, and compared with the simulation diagram under the above-mentioned fixed sensor condition. The simulation results are shown in Figure 6.

Figure 6(a) Actual radiation source power; 6(b) Reconstruction of radiation source power by fixed sensor; 6(c) Wide-area virtual densification identification of radiation source power; 6(d) Actual electromagnetic situation; 6(e) Fixed sensor Electromagnetic situation inversion; 6(f) wide-area virtual densification electromagnetic situation inversion.

It can be seen from Figure 6 that the depth of the color can represent the power level, and the contour line represents the coverage of the radiation source. It can be seen intuitively the position of the radiation source, the radiation power, and the electromagnetic situation of each point. Fig. 6 (a), Fig. 6 (b), Fig. 6 (c) radiation source power diagram comparison, when the present invention identifies the radiation source, the size and position of its reconstructed power are basically the same as the power sum of the actual radiation source radiation The positions are consistent, and the accuracy of the reconstruction radiation source power of the present invention is higher than the accuracy of the fixed sensor reconstruction radiation source power; Fig. 6 (d), Fig. 6 (e), Fig. 6 (f) electromagnetic situation diagram comparison, electromagnetic Situation diagram refers to the visualized diagram of the received signal strength RSS (received power) on 400 reference points in the experimental area, and the inversion diagram of the electromagnetic situation of the present invention is basically consistent with the actual electromagnetic situation diagram, which is the reference of the inversion of the present invention The size and position of the received signal strength RSS (received power) on the point are basically consistent with the actual received signal strength RSS, and the accuracy of the electromagnetic situation inversion of the present invention is higher than that of the fixed sensor electromagnetic situation inversion. It shows that the present invention can more accurately realize the identification of the radiation source, and then realize the inversion of the electromagnetic situation.

Based on the above simulation analysis, the present invention can perform wide-area virtual densification on sensors under the condition that a small number of sensor positions are randomly distributed, and radiation source positions and radiation power are randomly distributed, so as to realize radiation source identification, and then realize wide-area virtualization of spectrum situation. generate.

Claims (4)

1. A wide area virtual intensive spectrum situation generation method is characterized in that spectrum sensor equipment is attached to a small number of loading platforms such as vehicles and ships, spectrum sensing node intensive virtualization is achieved by means of mobility of the loading platforms of the spectrum sensor equipment and through obtaining electromagnetic spectrum data of a current position for multiple times within monitoring duration, spectrum monitoring samples are increased, and a wide area high-resolution electromagnetic spectrum situation is generated by means of a spectrum situation sparse inversion theory, and the method comprises the following steps:

(1) determining and configuring complex electromagnetic environment parameters: the experimental region adopts N-point grid layout, K radiation sources, T spectrum sensor devices and a bearing platform thereof are randomly distributed at grid top points of the experimental region, the bearing platform carries a GPS module and is used for synchronously recording position information at sensing nodes, the spectrum sensor devices and a bearing platform moving route thereof, and the N grid top points are selected as N reference points;

(2) wide area virtual densification acquisition perception data: virtualizing the T spectrum sensor devices and a bearing platform thereof into T mobile nodes, wherein each mobile node is densified after moving to obtain n sensing nodes for obtaining sensing data, the sensing data comprises Received Signal Strength RSS (Received Signal Strength) and position information, the T mobile nodes are collectively densified to obtain M ngT sensing nodes, and the Received Signal Strength RSS in the sensing data obtained by the M sensing nodes forms an M-dimensional column vector P_s∈R^M；

(3) According to the position information in the sensing data, a sensing node position matrix is constructed, and the sensing node position matrix phi can be represented by the following formula:

wherein S ═ { S ═ S_k1, 2.. M } represents a set of sensing nodes, s_kDenotes a k-th sensing node, k is used to identify the k-th sensing node, V ═ V_j1, 2.. N } represents the set of all reference points, V_jDenotes the jth reference point, j being used to identify the jth reference point, s_k∈V_jIndicating that the kth sensing node is located at the jth reference point,indicating that the kth sensing node is not positioned on the jth reference point, and the sensing node position matrix [ phi ]]_kjIs an M x N matrix;

(4) constructing a path loss matrix psi according to an electromagnetic propagation model of an electromagnetic environment;

(5) identifying a radiation source according to the sensing node position matrix and the path loss matrix to obtain the position and the radiation power of the radiation source;

(6) according to the identified radiation source, electromagnetic situation inversion is carried out, and received signal strength RSS on N reference points is obtained:

wherein the column vector P_r∈R^NAn N-dimensional column vector, a column vector P, representing received signal strengths RSS at N reference points_t∈R^NAn N-dimensional column vector formed by the radiation power of the radiation source on N reference points,representing additive white gaussian noise AWGN power.

2. The wide-area intensive spectrum situation generating method according to claim 1, wherein in step (2), after each mobile node moves, n sensing nodes are intensively used for acquiring sensing data, which includes received signal strength RSS and location information, comprising the following steps:

and each mobile node randomly selects a reference point in the test area as a destination, moves towards the destination, is called the destination, namely a sensing node at the reference point when reaching the destination, measures the Received Signal Strength (RSS) and the position information at the reference point to form sensing data, and repeats the steps until the mobile node approaches n reference points and acquires the sensing data at the reference point, wherein n is called a densification coefficient.

3. The wide-area dense spectrum situation generating method as claimed in claim 1, wherein said step (4) of constructing the path loss matrix Ψ according to the electromagnetic propagation model of the electromagnetic environment comprises the following steps:

(4a) the propagation model of the electromagnetic wave between the ith reference point and the jth reference point in the two-dimensional free space is as follows:

wherein, i, j is belonged to {1, 2.,. N }, i is used for identifying the ith reference point, j is used for identifying the jth reference point, and P is used for identifying the jth reference point_jrRepresents the received power of the jth reference point; p_itRepresenting the radiated power of the ith reference point; g_jrThe gain of the receiving antenna, G, of the jth reference point_itA transmit antenna gain representing the ith reference point; λ is the operating wavelength of the electromagnetic wave; d_ijRepresenting the distance, G, between the transmitting antenna of the ith reference point and the receiving antenna of the jth reference point_jr、G_itAll are known constants, the propagation model can be simplified as:

wherein,representing the loss coefficient between the ith reference point and the jth reference point;

(4b) the path loss matrix Ψ can be represented by the following equation:

wherein the path loss matrix [ psi ]]_ijIs an N x N matrix.

4. The wide-area dense spectrum situation generating method according to claim 1, wherein in the step (5), a radiation source is identified according to the sensing node position matrix and the path loss matrix, and a position and a radiation power of the radiation source are obtained, comprising the following steps:

(5a) calculating a sensing matrix Q according to the position matrix phi of the sensing node and the path loss matrix psi:

Q＝ΦΨ

(5b) m-dimensional column vector P formed by received signal strengths at M sensing nodes_sN-dimensional column vector P formed by radiation power of radiation source on N reference points_tThere are the following relationships between:

wherein,for Additive White Gaussian Noise AWGN (Additive White Gaussian Noise) power, column vector P_t∈R^N，R^NRepresenting a vector space of dimension N, P_t∈R^NRepresents P_tIs a vector of dimension N, the column vector epsilon ∈ R^MIndicating the measurement error of the sensor, R^MRepresents a vector space of dimension M, epsilon ∈ R^MRepresenting a vector with epsilon as dimension M;

(5c) constructing preprocessed data P_proc：

(5d) Solving an N-dimensional column vector P formed by the position of the radiation source and the radiation power of the radiation source on N reference points according to the minimum L1-norm_t：

min||P_t||,s.t.||P_proc-QP_t||₂≤μ

Wherein | · | | represents a 1-norm, meaning is the sum of the moduli of all elements in the vector, | · | | | caltropy₂Expressing 2-norm, meaning the sum of the squares of all the element norm values in the vector, reopening, mu is convergence precision, min represents minimization, s.t. the abbreviation of subject to represents that the constraint is', and the meaning of the whole equation is that the constraint condition is satisfied to be | | P_proc-QP_t||₂Mu or less, so that P is_tIs the smallest sum of all element modulus values, N-dimensional column vector P_tStoring at reference point corresponding to non-zero elementAt the radiation source, the value represents the magnitude of the radiation source power.