CN107579789B - Large-scale unmanned aerial vehicle relay network channel simulation device and GPU real-time simulation method - Google Patents

Abstract

The invention discloses a large-scale unmanned aerial vehicle relay network channel simulation device and a GPU real-time simulation method, wherein the large-scale unmanned aerial vehicle relay network channel simulation device comprises a network node dynamic topology parameter input unit, a network channel parameter estimation unit, a network Channel modeling and generating unit, network channel combination and superposition unit, network node transmit signal input unit and network node receive signal output unit. The simulation system of the invention is flexible and universal, and supports dynamic adjustment of network scale and topology structure; the signal, interference and noise of each relay link are modeled by an equivalent method, which greatly simplifies the complexity of system simulation; for different communication scenarios, Support different nodes to adopt different channel models and update them in real time, considering the mutual interference between network nodes.

Description

Large-scale UAV relay network channel simulation device and GPU real-time simulation method

Technical field:

The invention relates to a large-scale unmanned aerial vehicle relay network channel simulation device and a GPU real-time simulation method, belonging to the field of wireless information transmission, in particular for the future large-scale unmanned aerial vehicle relay network system, between each communication node and the relay node A wireless channel simulation device, and a real-time simulation method using a GPU platform. .

Background technique:

As the transmission medium of electromagnetic wave, wireless channel directly affects the transmission quality and performance of wireless communication system. The modeling and generation of wireless channels refers to establishing a channel model that is consistent with the actual propagation environment, and accurately and effectively restoring its channel characteristics through computer software simulation or hardware simulation, which is crucial for optimal design, evaluation and verification of wireless communication systems. Important, and can also effectively shorten the system development cycle.

Unmanned Aerial Vehicle (UAV) was born in the 1920s. Limited by the level of early technology, the development of UAVs was relatively slow before the 1970s. Since then, with the rapid development of science and technology such as communications, microelectronics, new materials, and aero-engines, it has now become a research hotspot for military and even civilian use in various countries. The scope of military applications is not limited to traditional fields such as aerial reconnaissance and war situation assessment, but is also widely used in military and police missions such as air combat, attack, interception, and anti-smuggling and anti-terrorism; Tasks such as aerial survey, remote sensing, environmental protection, disaster early warning and assessment, etc. In recent years, some express companies are also conducting drone express experiments. In addition, the Wireless Mobile Ad-Hoc Networks (MANET) based on UAV relays can be independent of any fixed infrastructure, and has the advantages of fast and flexible networking, wide coverage, high reliability and strong invulnerability. Figure 1 shows a typical application scenario of MANET based on UAV relay. In the wars in Afghanistan and Iraq, UAVs, as air relay communication stations, assumed the role of some information network nodes. They could not only transmit the information of enemy ground targets to their own ground combat troops and air fighters, but also through Its own airborne equipment enables mutual communication between ground forces, air fighters and headquarters.

However, the communication environment of the UAV is complex, and the wireless signal will be affected by factors such as terrain, ground objects and atmospheric rainfall, coupled with the change of its own flight attitude, resulting in rapid and random fading of the received signal, resulting in the degradation of the transmission performance of the entire communication network. In order to ensure high-quality command or data information transmission between the UAV and the communication parties, the control center or other aircraft during the flight, it is necessary to conduct in-depth research on the channel transmission characteristics of the UAV relay network and analyze. At the same time, the actual measurement of the UAV relay network channel is difficult, and it is usually necessary to mass-produce a large number of UAVs to deploy the relay network, resulting in high cost, long cycle and low efficiency for communication equipment testing. Therefore, it is particularly important to simulate the channel propagation characteristics of the UAV relay network in the ground laboratory environment, and then complete the simulation test of the UAV communication system.

In the field of mobile communication, there are currently commercial channel simulators abroad, such as PropsimC8 of Elabit, which can support the channel models recommended by 3GPP such as Geometric Model (SCM), SCM Extended Model (SCME), etc., and can support up to 12500km /h Doppler extension, the maximum delay extension is 6.4ms, which can meet the simulation test requirements of mobile communication channels in most scenarios; AEROFLEX's wideband channel simulator CS8007 can simulate Doppler frequency and Doppler acceleration , capable of simulating passband amplitude, phase distortion and various fading, while superimposing accurate white Gaussian noise and phase noise; Spirent's SR5500 can accurately simulate advanced receivers with diverse beamforming and MIMO Complex broadband wireless channel characteristics, its modular architecture can provide a variety of combinations, enabling complex MIMO channel testing. Commercial channel simulators can implement common channel fading models, such as Rayleigh fading, Rice fading, Nakagami fading, log-normal fading and Suzuki fading models. However, such simulators are generally only applicable to cellular mobile communication applications. And often only consider a single statistical model to achieve, can not meet the requirements of large-scale network node communication channel simulation.

The real-time simulation of large-scale network node communication channel has a large amount of calculation and high real-time requirements. If it is implemented in hardware, it will consume huge hardware resources, making it difficult to implement. If it is implemented by traditional PC-based software, the calculation speed is slow and cannot be implemented. Meet real-time requirements. In recent years, the Graphic Processing Unit (GPU) has been developing at a speed exceeding Moore's Law. Since there are more processing units, its computing power far exceeds that of the CPU, and the memory bandwidth capability also has obvious advantages over the CPU. In the early days, GPUs were mainly used for image rendering. With the substantial increase in computing power, they are more and more used in general computing, such as oil exploration, biomedicine, weather forecasting, fluid mechanics, marine environment simulation, earth science, finance The fields of analysis, big data processing and artificial intelligence have also been used in the application research of communication algorithms in recent years.

Invention content:

The present invention provides a large-scale UAV relay network channel simulation device and a GPU real-time simulation method in order to solve the problems existing in the above-mentioned prior art, which are suitable for real-time testing and real-time testing of large-scale UAV relay and network node equipment. Verify the realm.

The present invention also adopts the following technical scheme: a large-scale UAV relay network channel simulation device, comprising a network node dynamic topology parameter input unit, a network channel parameter estimation unit, a network channel modeling and generation unit, and a network channel combination and superposition unit , a network node transmit signal input unit and a network node receive signal output unit;

The network node dynamic topology parameter input unit is connected to the network channel parameter estimation unit, and is used for the user to input the communication scene parameters of each node of the network;

The network channel parameter estimation unit is used to convert the communication scene parameters of each node in the network node dynamic topology parameter input unit into the model parameters of each node channel of the UAV relay network, and then group the calculated results in the order of discrete time. frame, and sequentially transmitted to the network channel modeling and generation unit based on GPU parallel computing according to the network channel state update interval;

The network channel modeling and generating unit includes a ground transmitting node signal model, a UAV relay and forwarding node receiving signal model, a UAV relay receiving node receiving signal model, a ground receiving node signal model, a ground node interference signal model, The ground node noise model is based on the network sub-channel model parameters of each frame calculated by the network channel parameter estimation unit. The UAV relay network channel is sequentially generated through the above model, and the output data is sequentially transmitted to the network in the FPGA through the PCIE bus. Channel combination superposition unit;

The network channel combination and superposition unit models the network channel and the generation unit generates the UAV relay network channel fading and superimposes the baseband signal input by the network node transmission signal input unit in the FPGA, and sends it to the network node in the FPGA to receive the signal output unit;

The network node transmission signal input unit converts the input intermediate frequency IF or radio frequency RF signal into a complex baseband signal through down-conversion, and transmits it to the network channel combination and superposition unit;

The network node receiving signal output unit converts the complex baseband signal input by the network channel combination and superposition unit after passing through the UAV relay network channel into an intermediate frequency or radio frequency signal through up-conversion.

The present invention also adopts the following technical scheme: a GPU real-time simulation method of a large-scale unmanned aerial vehicle relay network channel simulation device, comprising the following steps:

In the first step, the user inputs the communication scene parameters through the network node dynamic topology parameter input unit, and the communication scene parameters are sent to the network channel parameter estimation unit;

In the second step, the network channel parameter estimation unit calculates the model parameters of each node channel of the UAV relay network according to the input parameters of the user, and then frames the calculated results according to the discrete time sequence, and transmits them to the network channel state update interval in turn. Network channel modeling and generation unit;

In the third step, the network channel modeling and generating unit establishes the signal model received by the UAV relay and forwarding node and the signal model received by the UAV relay receiving node according to the model parameters of each sub-channel of the network input by the network channel parameter estimation unit. , ground node receiving signal model, ground node interference signal model and ground node noise model;

The fourth step is to use the above model to generate the data of each UAV relay network propagation channel through GPU simulation, and transmit the data to the network channel combination and superposition unit in the FPGA in turn through the PCIE bus. At the same time, the intermediate frequency or radio frequency The signal input to the network node in the FPGA transmits the signal input unit, converts it into a baseband signal through down-conversion, and transmits it to the network channel combination and superposition unit;

In the fifth step, the network channel combination and superposition unit simulates the UAV relay network channel superposition process, and superimposes the network channel modeling and generation unit to generate the UAV relay network channel to the baseband signal input by the network node transmission signal input unit, and The network node in the FPGA receives the signal output unit, and the network node receives the signal output unit to convert the baseband signal after passing through the channel into an intermediate frequency or radio frequency signal for output through up-conversion.

Further, in step 3: the ground transmitting node N ₁ , passing through N UAV relay nodes R ₁ to R _N , and finally reaching the ground receiving node N ₂ , the received signal of a relay communication link is modeled as follows: Equivalent model

代表无人机中继节点R_i的转发增益；

分别表示地空、空地和无人机中继节点R_i-1与R_i之间的传播信号损耗因素，将其取值为In the formula,

Represents the forwarding gain of the _UAV relay node Ri;

Represents the propagation signal loss factor between the ground-air, air-ground and UAV relay nodes R _i-1 and R _i respectively, and takes the value of

α=32.44+20lg(f _MHz )+20lg(d _km ) (13)

表示地空和空地两段链路级联衰落，将其建模为一个随机变量，对应概率密度分布为Among them, f _MHz represents the communication frequency, the unit is MHz; d _km represents the communication distance, the unit is km,

Represents the cascading fading of the ground-air and air-ground links, which is modeled as a random variable, and the corresponding probability density distribution is

分别表示地空与空地链路的信道衰落平均功率；In the formula, m _i , m _si , i = 1, 2 respectively reflect the severity of multipath fading and shadow fading of ground-air and air-ground links;

Represent the average power of channel fading of ground-air and air-ground links, respectively;

其中噪声建模为加性高斯噪声，

建模为如下等效模型，The equivalent models of interference and noise are denoted as

where the noise is modeled as additive Gaussian noise,

Modeled as the following equivalent model,

分别表示第k路干扰信道与干扰源信号，服从独立同分布的复高斯分布CN(0,1)；l(d)表示大尺度衰落函数，可表示为In the formula, M represents the number of interference sources; d _k , k=1,...,M represents the distance between the k-th interference source and N ₂ ; P _k ,k=1,...,M represents the k-th interference source signal power;

respectively represent the k-th interference channel and the interference source signal, which obey the complex Gaussian distribution CN(0,1) of the independent and identical distribution; l(d) represents the large-scale fading function, which can be expressed as

In the formula, B _l ~B(1,p _l ) represents a random variable obeying Bernoulli distribution, where p _l represents the probability that the interference source and N ₂ have a line-of-sight path; L _l ~log(0,σ _l ) and L _n ～log(0,σ _n ) represent random variables obeying log-normal distribution, where σ _l , σ _n represent the shadow fading degree under line-of-sight paths and non-line-of-sight paths, respectively; α _l , α _n and β _l , β _n represent the path loss index and intercept under the line-of-sight path and the non-line-of-sight path, respectively.

Further, in step four:

The specific generation method of using GPU to generate channel fading, interference and noise is as follows:

1) First generate a Gaussian random process by the following method

where, N represents the number of indistinguishable scattering branches; σ ² represents the variance; ω _i,n =2πf _i,d cosα _i,n , where f _i,d =f ₀ /vc represents the maximum Doppler frequency, f ₀ , v, c correspond to the carrier frequency, the relative movement speed of the transceiver and the speed of light respectively; α _i,n , φ _i,n refer to the incident angle and initial phase of each scattering branch respectively, and the incident angle α _i,n is set to be in [ 0,2π) at equal intervals, the initial phase φ _i,n is set to be randomly and uniformly distributed within [0,2π), and the characteristics of GPU parallel computing are used to generate a Gaussian random process;

2) Using a set of Gaussian random processes u _i,0 (t)～N(0,1) generated by the method in step 1), perform nonlinear transformation to obtain a random process representing shadow fading, that is,

where σ _x , μ _x are the standard deviation and regional mean of shadow fading, respectively;

3) Using the method of step 1) to generate multiple groups of Gaussian random processes, perform nonlinear transformation, and generate random processes representing multipath fading, that is,

Among them, u _i (t)～N(0,σ ² ) and σ ² =Ω/2m, Ω=E[x ² ] represents the average power of multipath fading; m represents the fading factor, which is used to describe the signal caused by different scattering environments degree of decline;

4) Repeat the generation process of steps 2) and 3), and use the following formula to obtain the equivalent fading that obeys the distribution of formula (3),

与

分别表示地空与空地链路的阴影衰落和多径衰落的随机过程；in,

and

represent the random processes of shadow fading and multipath fading of ground-air and air-ground links, respectively;

利用如下方法计算获得第i时刻干扰源的瞬时数目，5) According to the scenario input by the user, obtain the generation probability λ _G and the average number of interference sources

Use the following method to calculate and obtain the instantaneous number of interference sources at the i-th moment,

Among them, B _r ~ B(1, P _r ) is a random variable obeying Bernoulli distribution, P _r represents the probability of each interference source surviving within Δt,

表示地面接收节点N₂的速度；

P_F,

分别表示所有干扰源移动的平均概率与平均速度；D_c表示相干距离；in,

Represents the speed of the ground receiving node N ₂ ;

P _F ,

Represents the average probability and average speed of all interference sources moving; D _c represents the coherence distance;

6) Calculate each parameter separately and use formula (5) to obtain l(d _k ), k=1,...,M, and then generate 4M Gaussian random process u _k (t)～N( 0,0.5), k=1,2,...,4M, and then M interference source signals are obtained;

7) Generate two-way Gaussian random process u _i (t)～N(0,σ ² ) by formula (6), i=1,2, where σ ² =P _N /2, P _N represents the complex noise power, and the two The paths correspond to the real and imaginary parts of the complex noise, respectively.

The present invention has the following beneficial effects:

(1) The simulation system is flexible and universal, and supports dynamic adjustment of network scale and topology structure;

(2) The signal, interference and noise of each relay link are modeled by the equivalent method, which greatly simplifies the complexity of the system simulation;

(3) For different communication scenarios, support different nodes to adopt different channel models and update them in real time, considering the mutual interference between network nodes.

Description of drawings:

Figure 1 shows a typical application scenario of MANET based on UAV relay.

Figure 2 shows the simulation scheme of the UAV relay network propagation environment.

Figure 3 is a real-time simulation of a GPU-based Gaussian stochastic process.

Figure 4 is a real-time simulation of GPU-based channel fading.

Detailed ways:

The present invention will be further described below in conjunction with the accompanying drawings.

The invention discloses a large-scale unmanned aerial vehicle relay network channel simulation device, comprising a network node dynamic topology parameter input unit 1-1, a network channel parameter estimation unit 1-2, a network channel modeling and generation unit 1-3, a network The channel combination and superposition unit 1-4, the network node transmit signal input unit 1-5 and the network node receive signal output unit 1-6.

The network node dynamic topology parameter input unit 1-1 is connected to the network channel parameter estimation unit 1-2, and is used for the user to input the scene parameters of each node communication in the network, mainly including the initial position and moving speed of the ground node and the UAV relay. , propagation environment parameters, noise parameters, etc.

The network channel parameter estimation unit 1-2 is used to convert the node communication scene parameters in the network node dynamic topology parameter input unit 1-1 into the model parameters of each node channel of the UAV relay network, including delay, path loss, Shadow fading, multipath fading, noise power and other parameters, and then the calculated results are framed according to the discrete time sequence, and transmitted to the network channel modeling and generation unit 1-3 based on GPU parallel computing according to the network channel state update interval. .

The network channel modeling and generating units 1-3 include the ground transmitting node signal model, the UAV relay forwarding node receiving signal model, the UAV relay receiving node receiving signal model, the ground receiving node signal model, and the ground node interference signal model. Model and ground node noise model, according to the network sub-channel model parameters of each frame calculated by the network channel parameter estimation unit 1-2, the UAV relay network channel is sequentially generated through the above model, and the output data is sequentially transmitted through the PCIE bus To the network channels in the FPGA combine overlay units 1-4.

The network channel combination and superposition unit 1-4 generates the network channel modeling and generation unit 1-3 to generate the UAV relay network channel fading and superimpose the baseband signal input by the network node transmission signal input unit 1-5 in the FPGA, and sends The network nodes in the FPGA receive signal output units 1-6.

The network node transmit signal input unit 1-5 converts the input intermediate frequency (IF) or radio frequency (RF) signal into a complex baseband signal through down-conversion, and transmits it to the network channel combination and superposition unit 1-4.

The network node receiving signal output unit 1-6 converts the complex baseband signal input by the network channel combination and superposition unit 1-4 after passing through the UAV relay network channel into an intermediate frequency or radio frequency signal through up-conversion.

The GPU real-time simulation method of the large-scale UAV relay network channel simulation device of the present invention comprises the following steps:

In the first step, the user inputs parameters such as communication scenarios through the network node dynamic topology parameter input unit 1-1, mainly including the initial position and moving speed of the ground node and the UAV relay, propagation environment parameters, noise parameters, etc. These parameters are It is sent to the network channel parameter estimation unit 1-2.

In the second step, the network channel parameter estimation unit 1-2 calculates the model parameters of each node channel of the UAV relay network according to the user input parameters, which mainly include channel parameters such as path loss, shadow fading, and multipath fading, as well as delay, noise and other channel parameters. power and other parameters, and then the calculated results are grouped into frames according to the discrete time sequence, and are sequentially transmitted to the network channel modeling and generation units 1-3 according to the network channel state update interval.

In the third step, the network channel modeling and generation unit 1-3 establishes the model parameters of each sub-channel network in each frame of the network input by the network channel parameter estimation unit 1-2, and establishes a model of the receiving signal of the UAV relay and forwarding node, and the model of the UAV in the UAV Following the receiving node receiving signal model, the ground node receiving signal model, the ground node interference signal model and the ground node noise model.

Without loss of generality, take a relay communication link from the ground transmitting node N ₁ , passing through N UAV relay nodes R ₁ to R _N , and finally reaching the ground receiving node N ₂ as an example, to illustrate the signal, interference, and noise at the receiving end. and noise modeling method:

1) This patent models the received signal of the relay communication link as the following equivalent model,

代表无人机中继节点R_i的转发增益；

分别表示地空、空地和无人机中继节点R_i-1与R_i之间的传播信号损耗因素，本案将其取值为In the formula,

Represents the forwarding gain of the _UAV relay node Ri;

Represents the propagation signal loss factor between ground-air, air-ground and UAV relay nodes R _i-1 and R _i respectively, which is taken as the value in this case

α=32.44+20lg(f _MHz )+20lg(d _km ) (24)

表示地空和空地两段链路级联衰落，本案将其建模为一个随机变量，对应概率密度分布为Among them, f _MHz represents the communication frequency, the unit is MHz; d _km is the communication distance, the unit is km.

Represents the cascading fading of the ground-air and air-ground links, which is modeled as a random variable in this case, and the corresponding probability density distribution is

分别表示地空与空地链路的信道衰落平均功率。In the formula, m _i , m _si , i = 1, 2 respectively reflect the severity of multipath fading and shadow fading of ground-air and air-ground links;

are the channel fading average powers of the ground-air and air-ground links, respectively.

其中噪声建模为加性高斯噪声，

建模为如下等效模型，2) In this patent, the equivalent models of interference and noise are recorded as

where the noise is modeled as additive Gaussian noise,

Modeled as the following equivalent model,

分别表示第k路干扰信道与干扰源信号，服从独立同分布的复高斯分布CN(0,1)；l(d)表示大尺度衰落函数，可表示为In the formula, M represents the number of interference sources; d _k , k=1,...,M represents the distance between the k-th interference source and N ₂ ; P _k ,k=1,...,M represents the k-th interference source signal power;

respectively represent the k-th interference channel and the interference source signal, which obey the complex Gaussian distribution CN(0,1) of the independent and identical distribution; l(d) represents the large-scale fading function, which can be expressed as

In the formula, B _l ~B(1,p _l ) represents a random variable obeying Bernoulli distribution, where p _l represents the probability that the interference source and N ₂ have a line-of-sight path; L _l ~log(0,σ _l ) and L _n ～log(0,σ _n ) represent random variables obeying log-normal distribution, where σ _l , σ _n represent the shadow fading degree under line-of-sight paths and non-line-of-sight paths, respectively; α _l , α _n and β _l , β _n represent the path loss index and intercept under the line-of-sight path and the non-line-of-sight path, respectively.

In the fourth step, using the above equivalent model, the data of each UAV relay network propagation channel is generated by GPU simulation, and the data is sequentially transmitted to the network channel combination and superposition units 1-4 in the FPGA through the PCIE bus. At the same time, the intermediate frequency or radio frequency signal is input to the network node transmission signal input unit 1-5 in the FPGA, converted into a baseband signal through down-conversion, and transmitted to the network channel combination and superposition unit 1-4. Among them, the specific generation method of using GPU to generate channel fading, interference and noise is as follows:

1) First generate a Gaussian random process by the following method

where, N represents the number of indistinguishable scattering branches; σ ² represents the variance; ω _i,n =2πf _i,d cosα _i,n , where f _i,d =f ₀ v/c represents the maximum Doppler frequency, f ₀ , v, c correspond to the carrier frequency, the relative movement speed of the transceiver and the speed of light respectively; α _i,n , φ _i,n respectively refer to the incident angle and initial phase of each scattering branch, this patent sets the incident angle α _i,n to set In order to take values at equal intervals within [0, 2π), the initial phase φ _i,n is set to be randomly and uniformly distributed within [0, 2π). Using the characteristics of GPU parallel computing, the specific scheme for generating Gaussian random process is shown in Figure 3. In this patent, the number of scattering branches is 32, and each thread block is allocated 32x8 threads. Each thread warp performs correlation calculation on each scattering branch at the same time, and performs a reduction and summation at the end to obtain a Gaussian random process.

2) Using a set of Gaussian random processes u _i,0 (t)～N(0,1) generated by the method in step 1), perform nonlinear transformation to obtain a random process representing shadow fading, that is,

where σ _x , μ _x are the standard deviation and regional mean of shadow fading, respectively.

3) Using the method of step 1) to generate multiple groups of Gaussian random processes, perform nonlinear transformation, and generate random processes representing multipath fading, that is,

Among them, u _i (t)～N(0,σ ² ) and σ ² =Ω/2m, Ω=E[x ² ] represents the average power of multipath fading; m represents the fading factor, which is used to describe the signal caused by different scattering environments degree of decline.

4) Repeat the generation process of steps 2) and 3), and use the following formula to obtain the equivalent fading that obeys the distribution of formula (25),

与

分别表示地空与空地链路的阴影衰落和多径衰落的随机过程。本案利用GPU并行计算的特点，具体产生该等效信道衰落的过程如图4所示，图中先产生2m+1路高斯随机过程，分别经过非线性变化后得到代表地空链路的阴影衰落和多径衰落的随机过程，再通过二者相乘获得地空链路的信道衰落随机过程，然后按照同样方法获得空地链路的信道衰落随机过程，并与地空链路的信道衰落随机过程相乘得到级联衰落随机过程。in,

and

represent the random processes of shadow fading and multipath fading of ground-air and air-ground links, respectively. This case uses the characteristics of GPU parallel computing, and the specific process of generating the equivalent channel fading is shown in Figure 4. In the figure, 2m+1 Gaussian random processes are generated first, and after nonlinear changes, the shadow fading representing the ground-air link is obtained. and the random process of multipath fading, and then multiply the two to obtain the random process of channel fading of the ground-air link, and then obtain the random process of channel fading of the air-ground link according to the same method. Multiplying to get the cascade fading random process.

利用如下方法计算获得第i时刻干扰源的瞬时数目，5) According to the scenario input by the user, obtain the generation probability λ _G and the average number of interference sources

Use the following method to calculate and obtain the instantaneous number of interference sources at the i-th moment,

Among them, B _r ~ B(1, P _r ) is a random variable obeying Bernoulli distribution, P _r represents the probability of each interference source surviving within Δt,

表示地面接收节点N₂的速度；

P_F,

分别表示所有干扰源移动的平均概率与平均速度；D_c表示相干距离。in,

Represents the speed of the ground receiving node N ₂ ;

P _F ,

Represents the average probability and average speed of all interference sources moving; D _c represents the coherence distance.

6) Calculate each parameter separately and use formula (27) to obtain l(d _k ), k=1,...,M, and then generate 4M Gaussian random process u _k (t by formula (28) and the method of Fig. 3 )～N(0,0.5), k=1,2,...,4M, and then M interference source signals are obtained.

7) Generate a two-way Gaussian random process u _i (t)～N(0,σ ² ) by formula (28) and the method of Fig. 3, i=1,2, where σ ² =P _N /2, P _N represents the complex Noise power, the two paths correspond to the real and imaginary parts of the complex noise, respectively.

In the fifth step, the network channel combination and superposition unit 1-4 simulates the UAV relay network channel superposition process, and the network channel modeling and generation unit 1-3 generates the UAV relay network channel and superimposes it on the network node transmission signal input unit. The baseband signal input from 1-5 is sent to the network node in the FPGA to receive the signal output unit 1-6, and the network node receives the signal output unit 1-6 to convert the baseband signal after passing through the channel into an intermediate frequency or radio frequency signal through up-conversion. output.

The above are only the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, several improvements can be made without departing from the principles of the present invention, and these improvements should also be regarded as the invention. protected range.

Claims (4)

1. a large-scale unmanned aerial vehicle relay network channel simulation device, is characterized in that: comprise network node dynamic topology parameter input unit (1-1), network channel parameter estimation unit (1-2), network channel modeling and a generating unit (1-3), a network channel combining and superimposing unit (1-4), a network node transmitting signal input unit (1-5) and a network node receiving signal output unit (1-6); The network node dynamic topology parameter input unit (1-1) is connected to the network channel parameter estimation unit (1-2), and is used for the user to input communication scene parameters of each node of the network; The network channel parameter estimation unit (1-2) is used to convert the communication scene parameters of each node in the network node dynamic topology parameter input unit (1-1) into model parameters of each node channel of the UAV relay network, and then The calculated results are framed according to the discrete time sequence, and are sequentially transmitted to the network channel modeling and generation unit (1-3) based on GPU parallel computing according to the network channel state update interval; The network channel modeling and generating unit (1-3) includes a ground transmitting node signal model, a UAV relay and forwarding node receiving signal model, a UAV relay receiving node receiving signal model, a ground receiving node signal model, and a ground receiving node signal model. The node interference signal model, the ground node noise model, and the network sub-channel model parameters of each frame calculated by the network channel parameter estimation unit (1-2) are used to generate the UAV relay network channel in turn through the above models, and output data. The network channel combination and superposition unit (1-4) in the FPGA are sequentially transmitted through the PCIE bus; The network channel combination and superposition unit (1-4) uses the network channel modeling and generation unit (1-3) to generate the UAV relay network channel fading and superimpose the network node transmission signal input unit (1-5) in the FPGA The input baseband signal is sent to the network node in the FPGA to receive the signal output unit (1-6); The network node transmission signal input unit (1-5) converts the input intermediate frequency IF or radio frequency RF signal into a complex baseband signal through down-conversion, and transmits it to the network channel combination and superposition unit (1-4); The network node receiving signal output unit (1-6) converts the complex baseband signal input by the network channel combination and superposition unit (1-4) after passing through the UAV relay network channel into an intermediate frequency or radio frequency signal through up-conversion. 2. a GPU real-time simulation method of a large-scale unmanned aerial vehicle relay network channel simulation device, is characterized in that: comprise the steps: In the first step, the user inputs communication scene parameters through the network node dynamic topology parameter input unit (1-1), and the communication scene parameters are sent to the network channel parameter estimation unit (1-2); In the second step, the network channel parameter estimation unit (1-2) calculates the model parameters of each node channel of the UAV relay network according to the user input parameters, and then groups the calculated results in discrete time sequence, and according to the network channel state The update interval is sequentially transmitted to the network channel modeling and generation unit (1-3); In the third step, the network channel modeling and generating unit (1-3) establishes the model parameters of each sub-channel network in each frame of the network input by the network channel parameter estimation unit (1-2). UAV relay receiving node receiving signal model, ground node receiving signal model, ground node interference signal model and ground node noise model; In the fourth step, using the above model, the data of each UAV relay network propagation channel is generated by GPU simulation, and the data is sequentially transmitted to the network channel combination and superposition unit (1-4) in the FPGA through the PCIE bus. At the same time, the intermediate frequency or radio frequency signal is input to the network node transmission signal input unit (1-5) in the FPGA, converted into a baseband signal through down-conversion, and transmitted to the network channel combination and superposition unit (1-4); In the fifth step, the network channel combination and superposition unit (1-4) simulates the UAV relay network channel superposition process, and the network channel modeling and generation unit (1-3) generates the UAV relay network channel and superimposes it on the network node. The baseband signal input by the signal input unit (1-5) is transmitted and sent to the network node in the FPGA to receive the signal output unit (1-6). The signal is converted into IF or RF signal output by up-conversion. 3. the GPU real-time simulation method of large-scale unmanned aerial vehicle relay network channel simulation device as claimed in claim 2, it is characterized in that: in step 3: ground launch node N ₁ , via N unmanned aerial vehicle relay nodes R ₁ to R _N , the received signal of a relay communication link that finally reaches the ground receiving node N ₂ is modeled as the following equivalent model

In the formula,

Represents the forwarding gain of the _UAV relay node Ri;

Represents the propagation signal loss factor between the ground-air, air-ground and UAV relay nodes R _i-1 and R _i respectively, and takes the value of α=32.44+20lg(f _MHz )+20lg(d _km ) (2) Among them, f _MHz represents the communication frequency, the unit is MHz; d _km represents the communication distance, the unit is km,

Represents the cascading fading of the ground-air and air-ground links, which is modeled as a random variable, and the corresponding probability density distribution is

In the formula, m _i , m _si , i = 1, 2 respectively reflect the severity of multipath fading and shadow fading of ground-air and air-ground links;

Represent the average power of channel fading of ground-air and air-ground links, respectively; The equivalent models of interference and noise are denoted as

where the noise is modeled as additive Gaussian noise,

Modeled as the following equivalent model,

In the formula, M represents the number of interference sources; d _k , k=1,...,M represents the distance between the k-th interference source and N ₂ ; P _k ,k=1,...,M represents the k-th interference source signal power;

respectively represent the k-th interference channel and the interference source signal, which obey the complex Gaussian distribution CN(0,1) of the independent and identical distribution; l(d) represents the large-scale fading function, which can be expressed as

In the formula, B _l ~B(1,p _l ) represents a random variable obeying Bernoulli distribution, where p _l represents the probability that the interference source and N ₂ have a line-of-sight path; L _l ~log(0,σ _l ) and L _n ～log(0,σ _n ) represent random variables obeying log-normal distribution, where σ _l , σ _n represent the shadow fading degree under line-of-sight paths and non-line-of-sight paths, respectively; α _l , α _n and β _l , β _n represent the path loss index and intercept under the line-of-sight path and the non-line-of-sight path, respectively. 4. the GPU real-time simulation method of large-scale unmanned aerial vehicle relay network channel simulation device as claimed in claim 3, is characterized in that: in step 4: The specific generation method of using GPU to generate channel fading, interference and noise is as follows: 1) First generate a Gaussian random process by the following method

where, N represents the number of indistinguishable scattering branches; σ ² represents the variance; ω _i,n =2πf _i,d cosα _i,n , where f _i,d =f ₀ v/c represents the maximum Doppler frequency, f ₀ , v, c correspond to the carrier frequency, the relative moving speed of the transceiver and the speed of light respectively; α _i,n , φ _i,n refer to the incident angle and initial phase of each scattering branch respectively, set the incident angle α _i,n to Values are taken at equal intervals in [0,2π), and the initial phase φ _i,n is set to be randomly and uniformly distributed in [0,2π), and a Gaussian random process is generated by utilizing the characteristics of GPU parallel computing; 2) Using a set of Gaussian random processes u _i,0 (t)～N(0,1) generated by the method in step 1), perform nonlinear transformation to obtain a random process representing shadow fading, that is,

where σ _x , μ _x are the standard deviation and regional mean of shadow fading, respectively; 3) Using the method of step 1) to generate multiple groups of Gaussian random processes, perform nonlinear transformation, and generate random processes representing multipath fading, that is,

Among them, u _i (t)～N(0,σ ² ) and σ ² =Ω/2m, Ω=E[x ² ] represents the average power of multipath fading; m represents the fading factor, which is used to describe the signal caused by different scattering environments degree of decline; 4) Repeat the generation process of steps 2) and 3), and use the following formula to obtain the equivalent fading that obeys the distribution of formula (3),

in,

and

represent the random processes of shadow fading and multipath fading of ground-air and air-ground links, respectively; 5) According to the scenario input by the user, obtain the generation probability λ _G and the average number of interference sources

Use the following method to calculate and obtain the instantaneous number of interference sources at the i-th moment,

Among them, B _r ~ B(1, P _r ) is a random variable obeying Bernoulli distribution, P _r represents the probability of each interference source surviving within Δt,

in,

Represents the speed of the ground receiving node N ₂ ;

Represents the average probability and average speed of all interference sources moving; D _c represents the coherence distance; 6) Calculate each parameter separately and use formula (5) to obtain l(d _k ), k=1,...,M, and then generate 4M Gaussian random process u _k (t)～N( 0,0.5), k=1,2,...,4M, and then M interference source signals are obtained; 7) Generate two-way Gaussian random process u _i (t)～N(0,σ ² ) by formula (6), i=1,2, where σ ² =P _N /2, P _N represents the complex noise power, and the two The paths correspond to the real and imaginary parts of the complex noise, respectively.

Images