Lightweight radio frequency SIP system based on frequency adjustable antenna

Abstract

The invention discloses a lightweight radio frequency SIP (Session Initiation Protocol) system based on a frequency-adjustable antenna, which relates to the technical field of safety management and comprises a frequency-adjustable antenna module, a feed module, a filter module, a power division module, a distribution adjustment module, an amplitude-phase control module, a transceiving component module, a power supply module and a radio frequency signal interconnection module. According to the invention, the beamforming precision can be improved, the directivity can be enhanced, the working frequency band can be dynamically adjusted, the system flexibility is improved, the anti-interference capability is improved, the communication reliability is enhanced, and the frequency adjustment function of the horn antenna within a certain frequency range is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of security management, and particularly relates to a lightweight radio frequency SIP system based on a frequency-tunable antenna. Background Art

[0002] With the development of radar detection technology, modern direction-finding systems, whether for military radar detection or police patrol detection, have put forward higher requirements for detection range, measurement accuracy, and portability characteristics, and can more accurately judge parameters such as the position, speed, object size, and material properties of targets, so as to achieve a full understanding of the targets and realize the "advanced" detection intelligence. As a converter for transmitting and receiving electromagnetic wave signals, an antenna is an important transceiver device and also a spatial filter for the amplitude, phase, frequency, and polarization characteristics of electromagnetic waves. The antenna module is gradually developing towards a smaller overall volume, lighter weight, thinner thickness, higher integration, and lower power consumption, promoting the development of the easier integration of antennas and radio frequencies. Therefore, the SIP concept has been proposed, and SIP technology has been widely used in various communication fields such as satellite communication, shipborne fire control, missile guidance, and ground warning. Therefore, in addition to meeting the relevant antenna performance, high transceiver gain, low power consumption, low noise, and other basic electrical characteristics required for different application scenarios, it is also necessary to be able to have a smaller size, lower profile, be compatible with different device requirements, and be able to achieve better integration with the system at the same time. Therefore, the future development components of the radio frequency module SIP with an antenna are developing towards miniaturization, high performance, high integration, and multi-scenario use. For this reason, we propose a lightweight radio frequency SIP system based on a frequency-tunable antenna. Summary of the Invention

[0003] The object of the present invention is to solve the above-mentioned problems and provide a lightweight radio frequency SIP system based on a frequency-tunable antenna.

[0004] The present invention proposes a lightweight radio frequency SIP system based on a frequency-tunable antenna, and the system includes:

[0005] A frequency-adjustable antenna module, a feeding module, a filtering module, a power splitting module, a distribution adjustment module, an amplitude-phase control module, a transceiver component module, a power supply module, and a radio frequency signal interconnection module;

[0006] The frequency-adjustable antenna module adopts an adjustable horn antenna structure, and is used to generate corresponding radio frequency signals by changing its own radiation frequency range to adapt to different usage scenarios;

[0007] The feeding module includes a feeding unit and a dielectric unit;

[0008] Among them, the feeding unit is used to transmit radio frequency signals from the feeding network to the frequency adjustable antenna module; the dielectric unit is used to perform impedance matching on the propagation of radio frequency signals;

[0009] The filtering module is used to perform frequency selection on radio frequency signals and transmit them to the power splitting module;

[0010] The power splitting module receives the filtered radio frequency signals and distributes the radio frequency signal power inside the radio frequency SIP system;

[0011] The distribution adjustment module is used to optimize the radio frequency signal power distribution of the power splitting module;

[0012] The amplitude and phase control module is used to control the amplitude and phase of the frequency adjustable antenna module according to the radio frequency signal power distribution and optimize the beamforming of the antenna array;

[0013] The transceiver component module is used to process received and transmitted radio frequency signals;

[0014] The power supply module is used to provide a stable working power supply for the entire radio frequency SIP system;

[0015] The radio frequency signal interconnection module uses a vertical interconnection method with a coaxial-like structure to connect the radio frequency signal paths within the system.

[0016] As a further solution of the present invention, the horn antenna structure of the frequency adjustable antenna module includes various deformed horn forms such as rectangular horns, circular horns, ridged horns, diaphragm horns, and box-shaped horns, and its specific horn antenna structure is adjusted accordingly according to different electrical performances required by specific usage scenarios;

[0017] The frequency adjustable antenna module adjusts the expansion and contraction of different horn main parts through locking screws;

[0018] The feeding module adopts a single-layer microstrip patch form;

[0019] The filtering module adopts a stripline form;

[0020] The power splitting module adopts the form of Wilkinson one-to-two power splitting, and a 100Ω resistor is used in the power splitting module. Its form is a thin film resistor, which is processed by an embedded resistor process.

[0021] As a further solution of the present invention, the specific steps for the filtering module to perform frequency selection on radio frequency signals are as follows:

[0022] S1.1: The filtering module converts the received groups of RF signals into the form of Fourier series expansion to obtain the power spectral density of each RF signal, and then takes it as the initial state of signal processing. Taking the current initial state as the root node, and according to the signal processing paths selected by different filters, new signal processing states are generated and used as child nodes, and a search tree is constructed together with the root node. At the same time, the reward values and visit counts of each node in the search tree are initialized;

[0023] S1.2: Calculate the upper confidence bound UCB of each node in the current search tree, and starting from the root node, select the child node with the highest UCB value layer by layer according to the upper confidence bound strategy. If the selected child node is not fully expanded, stop node selection, and select a new filter scheme based on the current child node for evaluation, and add the generated new processing state as a child node to the search tree;

[0024] S1.3: After simulating the filtering of the RF signal by the newly selected filter scheme, calculate the quality index of the filtered RF signal, and use the MSE function to calculate the difference value between the filtered RF signal and the ideal target spectrum. Then, trace back the simulated generated groups of data to the root node, and update the visit counts and reward values of each child node in the path;

[0025] S1.4: Repeatedly perform selection, expansion, simulation, and backtracking until the change value of the difference value converges to the preset range after multiple iterations, then stop the construction of the search tree. Then, traverse the final search tree and select the filter scheme corresponding to the child node with the most visit counts as the final scheme;

[0026] S1.5: The filtering module filters the currently received RF signal according to the filter scheme. After filtering, an output signal containing the components of the required frequency band is generated, and then through the inverse Fourier transform, the output signal is converted from the frequency domain back to the time domain to obtain the final filtered RF signal.

[0027] As a further solution of the present invention, the specific calculation formula of the power spectral density described in S1.1 is as follows:

[0028]

[0029] S in (f psd )=|F(s in (t))| 2

[0030] In the formula, F(s in (t)) represents the Fourier series expansion of the RF signal at the t-th second; A k represents the amplitude of the k-th frequency component; f k represents the frequency of the k-th frequency component; φk represents the initial phase of the k-th frequency component; S in (f psd ) represents the power spectral density of the input RF signal.

[0031] As a further solution of the present invention, the specific steps for the allocation adjustment module to optimize the RF signal power allocation of the power splitting module are as follows:

[0032] S2.1: The power splitting module receives the filtered RF signal, extracts the signal power spectral density, signal-to-noise ratio, and channel gain of each RF signal, and then the allocation adjustment module establishes a power allocation model based on the DNN network, and divides the extracted signal features into a training set and a test set;

[0033] S2.2: Input the training set into the power allocation model. The power allocation model performs forward propagation on the training set. The input layer receives the training set data. The hidden layer in the power allocation model uses multiple layers of neurons to perform non-linear transformation on each training set. Then the output layer uses the softmax function to normalize the power allocation ratio of each channel, and calculates the loss value between the model prediction value and the true label through the cross-entropy loss function;

[0034] S2.3: Starting from the output layer of the power allocation model, the loss value is propagated backward layer by layer, and the gradient of the loss value with respect to the parameters of each layer is calculated. Then the Adam optimizer is used to update the parameters of each layer. After that, the validation set is input into the power allocation model to evaluate the power allocation ability of the model. If the power allocation ability reaches the preset standard, stop training. Otherwise, repeat the training and verification of the power allocation model until the model loss value converges within the preset threshold, and then stop training;

[0035] S2.4: Input the signal features collected in real time into the trained power allocation model for forward propagation, and output the power allocation ratio of each channel through the output layer of the model. Based on the output power allocation ratio, the power splitting module allocates the RF signal power inside the RF SIP system.

[0036] As a further solution of the present invention, the specific steps for the amplitude-phase control module to optimize the beamforming of the antenna array are as follows:

[0037] S3.1: Collect the amplitude and phase combination schemes of different antenna elements, generate the antenna pattern corresponding to beamforming based on different combination schemes, and then establish an objective function according to the maximum main lobe gain, side lobe level, and null direction constraints. Calculate the goodness of each combination scheme's antenna pattern through the objective function, and use each combination scheme as a path and each variable in the scheme as a node;

[0038] S3.2: Set the number of exploration bodies, randomly initialize the concentration of pheromones for each amplitude and phase to represent the priority of each antenna element to select the amplitude or phase value. Take the performance of the antenna pattern under the current amplitude-phase combination scheme as heuristic information. Then initialize the starting nodes of each exploration body, calculate the selection probability of the next node based on the pheromone concentration and heuristic information, and select the node through the roulette wheel selection method. Stop the selection when a complete combination scheme is constructed;

[0039] S3.3: Calculate the goodness of each constructed combination scheme, update the pheromone concentration of each node based on the goodness of each combination scheme, repeat node selection, goodness calculation, and pheromone update until the preset number of iterations is reached, compare the goodness of the final combination schemes, and select the amplitude-phase combination scheme with the highest goodness to adjust the antenna pattern to form the optimal beamforming.

[0040] As a further solution of the present invention, the specific expression formula of the antenna pattern described in S3.1 is as follows:

[0041]

[0042] In the formula, F Rp (θ) represents the combined antenna pattern in the direction of angle θ; represents the amplitude excitation of the i Rp th antenna element; j Rp represents the imaginary unit; represents the phase excitation of the i Rp th antenna element; β Rp represents the wave number; represents the position of the i Rp th antenna element.

[0043] Advantages of the present invention:

[0044] In the present invention, the adjustable horn antenna body module is surface-mounted on the upper surface of the RF SIP module through a reflow soldering process. Then, the amplitude and phase combination schemes of different antenna units are collected, and the antenna pattern corresponding to beamforming is generated based on different combination schemes. Next, an objective function is established based on the constraints of the maximum main lobe gain, sidelobe level, and null direction. The goodness of each combination scheme's antenna pattern is calculated through the objective function. Each combination scheme is regarded as a path, and each variable in the scheme is regarded as a node. The number of exploration bodies is set, and the pheromone concentrations of each amplitude and phase are randomly initialized to represent the priority of each antenna unit to select the amplitude or phase value. The performance of the antenna pattern under the current amplitude-phase combination scheme is used as heuristic information. Then, the starting nodes of each exploration body are initialized. The selection probability of the next node is calculated based on the pheromone concentration and heuristic information, and the node is selected through the roulette selection method. When a complete combination scheme is constructed, the selection stops. The goodness of each constructed combination scheme is calculated, and the pheromone concentrations of each node are updated based on the goodness of each combination scheme. The node selection, goodness calculation, and pheromone update are repeated until the preset number of iterations is reached. The goodness of the final combination schemes is compared, and the amplitude-phase combination scheme with the highest goodness is selected to adjust the antenna pattern, forming an optimal beamforming, which can improve the beamforming accuracy, enhance the directivity, dynamically adjust the working frequency band, improve the system flexibility, improve the anti-interference ability, enhance the communication reliability, and realize the frequency adjustment function of the horn antenna within a certain frequency range. Description of the Drawings

[0045] The present invention will be further described below with reference to the accompanying drawings.

[0046] Figure 1 It is a framework diagram of a lightweight RF SIP system based on a frequency-tunable antenna. Detailed Embodiments

[0047] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0048] Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0049] The embodiments of the present invention provide a lightweight RF SIP system based on a frequency-tunable antenna. Refer to Figure 1 , Figure 1A framework diagram of a lightweight RF SIP system based on a frequency-tunable antenna provided by an embodiment of the present invention. The system includes:

[0050] A frequency-tunable antenna module, a feeding module, a filtering module, a power splitting module, a distribution adjustment module, an amplitude-phase control module, a transceiver component module, a power supply module, and a RF signal interconnection module;

[0051] The frequency-tunable antenna module adopts an adjustable horn antenna structure, which is used to adapt to different usage scenarios by changing its own radiation frequency range.

[0052] It should be noted that the horn antenna structure of the frequency-tunable antenna module includes various deformed horn forms such as rectangular horns, circular horns, ridged horns, diaphragm horns, and box-shaped horns. Its specific horn antenna structure is adjusted according to different electrical performances required by specific usage scenarios. And the frequency-tunable antenna module adjusts the expansion and contraction of different horn main parts through locking screws. The frequency-tunable antenna module and the chips included in the lower feeding module, filtering module, power splitting module, amplitude-phase control module, and transceiver component module are processed by surface mounting technology, and finally a lightweight RF SIP system is formed.

[0053] In addition, it should be noted that by disassembling the horn antenna main part into X segments, where X is a natural number greater than 1, the corresponding adjustable frequency type is X + 1; the number of radiation unit main bodies is M * N, where M and N are natural numbers greater than or equal to 1, and the number comes from the usage scenario, far and near distance coverage, corresponding different gain size requirements, and the 3dB beamwidth requirements corresponding to the required direction finding range;

[0054] When the number of radiation units M and N > 2, the element spacings d1 (in the M direction) and d2 (in the N direction) in the vertical and horizontal directions do not need to be kept consistent, and can be appropriately adjusted between 0.5λ - 0.85λ (λ is the wavelength corresponding to the center frequency) according to the electrical performance indicators required by the actual application scenario.

[0055] The microstrip feeding module includes a feeding unit and a dielectric unit; among them, the feeding unit is used to transmit the RF signal from the feeding network to the frequency-tunable antenna module; the dielectric unit is used to perform impedance matching on the propagation of the RF signal.

[0056] It should be noted that the microstrip feeding module adopts a single-layer microstrip patch form.

[0057] The filtering module is used to perform frequency selection on the RF signal and transmit it to the power splitting module.

[0058] Specifically, the wave module converts each received radio frequency signal into the form of Fourier series expansion to obtain the power spectral density of each radio frequency signal. Then, it takes this as the initial state of signal processing, uses the current initial state as the root node, and generates new signal processing states according to the signal processing paths selected by different filters, taking these new states as child nodes. Together with the root node, a search tree is constructed. At the same time, the reward values and visit counts of each node in the search tree are initialized. The upper confidence bound (UCB) value of each node in the current search tree is calculated, and starting from the root node, the child node with the highest UCB value is selected layer by layer according to the upper confidence bound strategy. If the selected child node is not fully expanded, stop node selection, select a new filter scheme based on the current child node for evaluation, add the generated new processing state as a child node to the search tree. After simulating the filtering of the radio frequency signal by the newly selected filter scheme, calculate the quality index of the filtered radio frequency signal, and use the MSE function to calculate the gap value between the filtered radio frequency signal and the ideal target spectrum. Then, trace back the simulated generated data groups to the root node, and update the visit counts and reward values of each child node in the path. Repeat the processes of selection, expansion, simulation, and backtracking until the change value of the gap value converges to the preset range after multiple iterations, then stop the construction of the search tree. After that, traverse the final search tree and select the filter scheme corresponding to the child node with the most visit counts as the final scheme. The filtering module filters the currently received radio frequency signal according to the filter scheme. After filtering, an output signal containing the components of the required frequency band is generated, and then through inverse Fourier transform, the output signal is converted from the frequency domain back to the time domain to obtain the final filtered radio frequency signal.

[0059] In addition, it should be noted that the specific calculation formula for the power spectral density is as follows:

[0060]

[0061] S in (f psd )=|F(s in (t))| 2

[0062] In the formula, F(s in (t)) represents the Fourier series expansion of the radio frequency signal at the t-th second; A k represents the amplitude of the k-th frequency component; f k represents the frequency of the k-th frequency component; φ k represents the initial phase of the k-th frequency component; S in (f psd ) represents the power spectral density of the input radio frequency signal.

[0063] It should be noted that the filtering module is in the form of a stripline, and adopts low-pass, band-pass, and high-pass forms, which are determined according to different frequency bands required by the frequency scenarios used.

[0064] The power splitting module receives the filtered radio frequency signal and distributes the power of the radio frequency signal inside the radio frequency SIP system.

[0065] Specifically, the power splitting module receives the filtered radio frequency signal, extracts the signal power spectral density, signal-to-noise ratio, and channel gain of each radio frequency signal, and then the allocation adjustment module establishes a power allocation model based on the DNN network, and divides the extracted signal features into a training set and a test set. The training set is input into the power allocation model, and the power allocation model performs forward propagation on the training set. The input layer receives the training set data. The hidden layer in the power allocation model uses multiple layers of neurons to perform non-linear transformation on each training set. Then the output layer uses the softmax function to normalize the power allocation ratio of each channel, and calculates the loss value between the model prediction value and the true label through the cross-entropy loss function. The loss value starts from the output layer of the power allocation model and is propagated backward layer by layer, and calculates the gradient of the loss value with respect to the parameters of each layer. Then the Adam optimizer is used to update the parameters of each layer. Then the validation set is input into the power allocation model to evaluate the power allocation ability of the model. If the power allocation ability reaches the preset standard, the training is stopped. Otherwise, the power allocation model is repeatedly trained and verified until the model loss value converges within the preset threshold, and then the training is stopped. The signal features collected in real time are input into the trained power allocation model for forward propagation, and the power allocation ratio of each channel is output through the output layer of the model. Based on the output power allocation ratio, the power of the radio frequency signal inside the radio frequency SIP system is distributed through the power splitting module.

[0066] In addition, it should be noted that the power splitting module adopts the form of Wilkinson one-to-two power splitting, and a 100Ω resistor is used in the power splitting module. Its form is a thin film resistor, which is processed by the buried resistor process.

[0067] The amplitude-phase control module is used to control the amplitude and phase of the frequency-adjustable antenna module according to the radio frequency signal power distribution, and optimize the beamforming of the antenna array.

[0068] Specifically, collect the amplitude and phase combination schemes of different antenna elements, generate the corresponding antenna patterns for beamforming based on different combination schemes, then establish an objective function according to the constraints of the maximum main lobe gain, sidelobe level, and null direction. Calculate the goodness of each combination scheme's antenna pattern through the objective function. Take each combination scheme as a path and each variable in the scheme as a node. Set the number of exploration bodies, randomly initialize the pheromone concentrations of each amplitude and phase to represent the priority of each antenna element to select amplitude or phase values. Use the performance of the antenna pattern under the current amplitude-phase combination scheme as heuristic information. Then initialize the starting nodes of each exploration body, calculate the selection probability of the next node according to the pheromone concentration and heuristic information, and select the node through the roulette wheel selection method. When a complete combination scheme is constructed, stop the selection, calculate the goodness of each constructed combination scheme, update the pheromone concentration of each node based on the goodness of each combination scheme, repeat the node selection, goodness calculation, and pheromone update until the preset number of iterations is reached, compare the goodness of the final combination schemes, and select the amplitude-phase combination scheme with the highest goodness to adjust the antenna pattern to form the optimal beamforming.

[0069] It should be noted that the specific expression formula of the antenna pattern is as follows:

[0070]

[0071] In the formula, F Rp (θ) represents the combined antenna pattern in the direction of angle θ; represents the amplitude excitation of the i Rp -th antenna element; j Rp represents the imaginary unit; represents the phase excitation of the i Rp -th antenna element; β Rp represents the wave number; represents the position of the i Rp -th antenna element.

[0072] The transceiver component module is used to process the received and transmitted radio frequency signals; the power supply module is used to provide a stable working power supply for the entire radio frequency SIP system; the radio frequency signal interconnection module uses a vertical interconnection method with a coaxial-like structure to connect the radio frequency signal paths in the system.

[0073] The above has described a specific embodiment of the present invention in detail, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of the present invention application should still fall within the scope covered by the patent of the present invention.

Claims

1. A lightweight radio frequency SIP system based on a frequency-adjustable antenna, characterized in that: include: Frequency adjustable antenna module, feeding module, filtering module, power division module, distribution adjustment module, amplitude and phase control module, transceiver component module, power supply module and radio frequency signal interconnection module; The frequency-adjustable antenna module adopts an adjustable horn antenna structure, which is used to generate corresponding radio frequency signals by changing its own radiation frequency range to adapt to different usage scenarios; The feeding module comprises a feeding unit and a dielectric unit; The feeding unit is used to transmit the radio frequency signal from the feeding network to the frequency adjustable antenna module; the dielectric unit is used to perform impedance matching on the propagation of the radio frequency signal; The filtering module is used to select the frequency of the radio frequency signal and transmit it to the power division module; The power division module receives the filtered RF signal and distributes the RF signal power inside the RF SIP system; The distribution adjustment module is used to optimize the radio frequency signal power distribution of the power division module; The amplitude and phase control module is used to control the frequency according to the radio frequency signal power distribution to adjust the amplitude and phase of the antenna module and optimize the beamforming of the antenna array; The transceiver component module is used to process receiving and transmitting radio frequency signals; The power supply module is used to provide a stable working power supply for the entire radio frequency SIP system; The radio frequency signal interconnection module adopts a vertical interconnection method of a coaxial structure to connect the radio frequency signal paths in the system.

2. A lightweight radio frequency SIP system based on a frequency-adjustable antenna according to claim 1, characterized in that: The horn antenna structure of the frequency adjustable antenna module includes various deformed horn forms such as rectangular horn, circular horn, ridged horn, diaphragm horn and box-shaped horn, and its specific horn antenna structure is adjusted accordingly according to the different electrical performances required by the specific usage scenario; The frequency adjustable antenna module is used to extend and adjust different speaker main body parts by locking screws; The feeding module adopts a single-layer microstrip patch form; The filtering module adopts a stripline form; The power division module adopts the form of Wilkinson one-to-two power division, and a 100Ω resistor is used in the power division module, which is in the form of a thin film resistor and is processed by an embedded resistor process.

3. A lightweight radio frequency SIP system based on a frequency-adjustable antenna according to claim 1, characterized in that: The specific steps of the filtering module for frequency selection of the radio frequency signal are as follows: S1.1: The filtering module converts each group of received RF signals into the form of Fourier series expansion to obtain the power spectrum density of each RF signal, and then uses it as the initial state of signal processing. The current initial state is used as the root node, and a new signal processing state is generated according to the signal processing path after different filter selections. It is used as a child node to build a set of search trees together with the root node, and the reward value and visit count of each node in the search tree are initialized at the same time; S1.2: Calculate the upper confidence limit value UCB of each node in the current search tree, and start from the root node, select the child node with the highest UCB value layer by layer according to the upper confidence limit strategy. If the selected child node is not fully expanded, stop node selection, and select a new filter scheme based on the current child node for evaluation, and add the generated new processing state as a child node to the search tree; S1.3: After simulating the newly selected filter scheme to filter the RF signal, the quality index of the filtered RF signal is calculated, and the MSE function is used to calculate the difference between the filtered RF signal and the ideal target spectrum. After that, each set of data generated by the simulation is traced back to the root node, and the number of visits and reward values ​​of each child node in the path are updated; S1.4: Repeatedly select, expand, simulate, and backtrack until the gap value converges to the preset range after multiple rounds of iterations, then stop building the search tree, traverse the final search tree, and select the filter solution corresponding to the child node with the most visits as the final solution; S1.5: The filtering module filters the currently received RF signal according to the filter scheme. After filtering, an output signal containing components of the required frequency band is generated. Then, the output signal is converted from the frequency domain back to the time domain through inverse Fourier transform to obtain the final filtered RF signal.

4. A lightweight radio frequency SIP system based on a frequency-adjustable antenna according to claim 3, characterized in that: The specific calculation formula for the power spectrum density described in S1.1 is as follows: S in (f psd )=|F(s in (t))| 2 In the formula, F(s in (t)) represents the Fourier series expansion of the RF signal at the tth second; A k represents the amplitude of the kth frequency component; f k represents the frequency of the kth frequency component; φ k represents the initial phase of the kth frequency component; S in (f psd ) represents the power spectral density of the input RF signal.

5. A lightweight radio frequency SIP system based on a frequency-adjustable antenna according to claim 3, characterized in that: The specific steps of optimizing the radio frequency signal power distribution of the power division module by the distribution adjustment module are as follows: S2.1: The power division module receives the filtered RF signal and extracts the signal power spectrum density, signal-to-noise ratio and channel gain of each RF signal. Then the allocation adjustment module establishes a power allocation model based on the DNN network and divides the extracted signal features into a training set and a test set. S2.2: The training set is input into the power allocation model. The power allocation model performs forward propagation on the training set. The input layer receives the training group data. The hidden layer in the power allocation model uses multi-layer neurons to perform nonlinear transformation on each training group. After that, the output layer uses the softmax function to normalize the power allocation ratio of each channel, and the cross entropy loss function is used to calculate the loss value between the model prediction value and the true label. S2.3: Starting from the output layer of the power allocation model, the loss value is back-propagated layer by layer, and the gradient of the loss value for the parameters of each layer is calculated. Then, the parameters of each layer are updated using the Adam optimizer. After that, the verification set is input into the power allocation model to evaluate the power allocation capability of the model. If the power allocation capability reaches the preset standard, the training is stopped. Otherwise, the power allocation model is repeatedly trained and verified until the model loss value converges to the preset threshold, and then the training is stopped. S2.4: Input the signal features collected in real time into the trained power allocation model for forward propagation, and output the power allocation ratio of each channel through the model output layer. Based on the output power allocation ratio, the RF signal power inside the RF SIP system is allocated through the power division module.

6. A lightweight radio frequency SIP system based on a frequency-adjustable antenna according to claim 1, characterized in that: The specific steps of the amplitude and phase control module optimizing the beamforming of the antenna array are as follows: S3.1: Collect the amplitude and phase combination schemes of different antenna units, generate the antenna radiation pattern of the corresponding beamforming based on the different combination schemes, and then establish the objective function according to the maximum main lobe gain, side lobe level and zero point direction constraint. The quality of the antenna radiation pattern of each combination scheme is calculated through the objective function, and each combination scheme is used as a path, and each variable in the scheme is used as a node; S3.2: Set the number of exploration bodies, randomly initialize the concentration of each amplitude and phase pheromone to indicate the priority of each antenna unit in selecting the amplitude or phase value, use the antenna pattern performance under the current amplitude-phase combination scheme as heuristic information, and then initialize the starting node of each exploration body. Calculate the selection probability of the next node according to the pheromone concentration and heuristic information, and select the node through the roulette selection method. When a complete combination scheme is constructed, stop the selection; S3.3: Calculate the merits of each constructed combination scheme, update the pheromone concentration of each node based on the merits of each combination scheme, repeat node selection, merits calculation and pheromone update until the preset number of iterations is reached, compare the merits of the final combination schemes, and select the amplitude and phase combination scheme with the highest merit to adjust the antenna radiation pattern and form the optimal beamforming.

7. A lightweight radio frequency SIP system based on a frequency-adjustable antenna according to claim 5, characterized in that: The specific expression formula of the antenna pattern described in S3.1 is as follows: In the formula, F Rp (θ) represents the composite antenna pattern in the direction of angle θ; Represents the i Rp The amplitude excitation of each antenna element; Rp represents the imaginary unit; Represents the i Rp The phase excitation of each antenna element; β Rp represents the wave number; Represents the i Rp The location of the antenna unit.

Images