A system for reducing communication congestion in wireless communication networks

Abstract

The present invention belongs to the field of wireless communication technology and relates to a system for reducing communication congestion in wireless communication networks. The physical layer adopts an updated frequency hopping sequence to resist interference, and is linked with the upper-layer transmission strategy and topology structure to achieve global optimization; the link layer resource allocation adopts a dynamic adjustment channel coding method to adapt to the channel state to avoid service packet loss due to congestion; the network layer topology optimization provides real-time spectrum information support, and path planning avoids interference areas. The present invention breaks the limitations of traditional layered architecture by dynamically linking the physical layer, link layer, and network layer to perform cross-layer collaborative anti-blocking, forming a "detection-decision-execution" closed loop, achieving dual avoidance of spectrum resources and topological paths, and improving network anti-blocking capabilities and transmission efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communications, and in particular relates to a system for reducing communication congestion in a wireless communication network. Background Art

[0002] In wireless communication networks, factors such as scarce spectrum resources, frequent interference (such as narrowband interference and pulsed interference), and dynamic changes in network topology can easily lead to communication congestion, manifesting as increased bit error rates, increased latency, and decreased throughput. Traditional anti-congestion methods often rely on single-dimensional optimization (such as frequency hopping or routing adjustments) and lack cross-layer coordination, resulting in low anti-interference efficiency and insufficient resource utilization. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention is implemented through the following technical solutions:

[0004] A system for reducing communication congestion in a wireless communication network is proposed, comprising:

[0005] The physical layer is used to detect interference signals in the communication environment in real time. When a new interference signal is detected, a notification message is received from the link layer, or a notification message is received from the network layer, the physical layer updates the frequency hopping sequence and sends corresponding notification messages to the link layer and the network layer respectively.

[0006] The link layer is configured to determine the current channel coding mode based on the notification message sent by the physical layer or the notification message sent by the network layer, and detect the packet loss rate of each data transmission path under the current channel coding mode. If the packet loss rate of one or more data transmission paths exceeds a first threshold, the link layer sends corresponding notification messages to the physical layer and the network layer respectively.

[0007] The network layer is used to determine the current data transmission path based on the notification message sent by the physical layer or the notification message sent by the link layer, and detect the impedance of each node on the current data transmission path. If the impedance of one or more nodes is greater than the second threshold, the corresponding notification message is sent to the physical layer and the link layer respectively.

[0008] Compared with existing technologies, this invention offers the following advantages and benefits: It achieves cross-layer collaborative anti-blocking through dynamic linkage between the physical, link, and network layers, breaking the limitations of traditional layered architectures and forming a closed "detection-decision-execution" loop. This achieves dual avoidance of spectrum resources and topological paths, improving network anti-blocking capabilities and transmission efficiency. Specifically, the physical layer uses updated frequency hopping sequences to combat interference and collaborates with upper-layer transmission strategies and topological structures to achieve global optimization. Link-layer resource allocation dynamically adjusts channel coding to adapt to channel conditions, avoiding packet loss due to congestion. Network-layer topology optimization provides real-time spectrum information support, allowing path planning to avoid interference areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0010] Figure 1 A schematic diagram of a system organizational architecture for reducing communication congestion in a wireless communication network provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with the examples. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention. The embodiments described below are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0012] In the following description, numerous specific details are set forth to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that these specific details are not necessarily required to practice the present invention. In other examples, well-known structures, materials, or methods are not specifically described to avoid obscuring the present invention. The materials, instruments, and reagents used in the following examples, unless otherwise specified, are commercially available. The techniques used in the examples, unless otherwise specified, are conventional techniques well known to those skilled in the art.

[0013] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. Throughout the description of this application, "plurality" means two or more, unless otherwise specifically defined.

[0014] Embodiment: Provide a Figure 1 The system for reducing wireless communication network congestion reduces communication congestion through physical layer spectrum sensing, link layer transmission strategy adaptation, network layer topology optimization, and cross-layer coordination mechanisms. The following explains the relationships between the various layers, functional modules, and functional units in this system.

[0015] (1) Physical layer

[0016] The physical layer consists of an interference signal detection module, a frequency hopping sequence update module, a first message sending module and a backup frequency band scanning module. It is used to detect interference signals in the communication environment in real time. When a new interference signal is detected, the frequency hopping sequence is updated and corresponding notification messages are sent to the link layer and network layer respectively, triggering a cross-layer collaborative response to achieve global linkage of interference avoidance.

[0017] 1. Interference signal detection module

[0018] The interference signal detection module is used to detect interference signals in the communication environment in real time. The interference signal detection module consists of a signal acquisition unit, a signal simulation unit, a signal fusion unit, a signal processing unit, a signal transformation unit, a feature extraction unit, a feature fusion unit, a model training unit, and a signal detection unit.

[0019] (1) Signal acquisition unit

[0020] To obtain comprehensive and representative data, it is necessary to extensively collect a large amount of time-frequency domain data samples, which should cover signals under normal communication conditions and various types of malicious interference signals. The signal acquisition unit described in this embodiment is used to collect network signal data in a communication environment to obtain a first signal database. It should be noted that these data sources are diversified, and signal monitoring equipment can be deployed in actual communication environments to continuously collect signal data for a long time to truly reflect the signal characteristics in complex communication scenarios.

[0021] (2) Signal simulation unit and signal fusion unit

[0022] The number of interference signals contained in the first signal database established by the signal acquisition unit is limited. In order to further enrich the data samples, this embodiment simulates various interference signal data in the communication environment through the signal simulation unit to obtain a second signal database, and then uses the signal fusion unit to fuse the first signal database with the second signal database to obtain a third signal database.

[0023] The signal simulation unit integrates simulation software, which can be used to construct various interference scenarios and generate corresponding interference signal data. The specific implementation method of constructing various interference scenarios using simulation software is as follows:

[0024] Step 1: Determine the simulation software and set parameters.

[0025] The simulation software described in this embodiment can be MATLAB, GNU Radio, etc., which has powerful signal processing capabilities and a rich signal generation library, and can be directly used to generate interference signals for various interference scenarios. The type of interference signal and related parameters are determined according to actual needs. Among them, the types of interference signals include narrowband interference, broadband interference, and pulse interference, and the related parameters include frequency, bandwidth, power, and pulse width.

[0026] Step 2: Generate interference signal.

[0027] Generate a narrowband interference signal. First, determine the center frequency and bandwidth of the narrowband interference (e.g., a center frequency of 1 GHz and a bandwidth of 1 MHz). Then, use a signal generation function in the software, such as the sin function in MATLAB or the "Signal Source" module in GNU Radio, to generate a sine wave signal as the narrowband interference signal. Set its frequency to the center frequency of the narrowband interference and adjust the amplitude based on the desired interference power. Furthermore, if a more complex narrowband interference signal is required, you can add some modulation methods, such as frequency modulation (FM) or amplitude modulation (AM). For example, use the fmmod function in MATLAB to frequency-modulate the sine wave signal to generate an FM narrowband interference signal.

[0028] Generate a wideband interference signal. First, determine the frequency range and power spectral density of the wideband interference (e.g., a frequency range of 1 GHz to 2 GHz and a power spectral density of -100 dBm / Hz). Then, use a noise generation function in the software, such as the randn function in MATLAB or the "Noise Source" module in GNU Radio, to generate a Gaussian white noise signal as the wideband interference signal. Finally, set its power spectral density to the desired value and use a filter to limit it to the desired frequency range. For example, use the fir1 function in MATLAB to design a bandpass filter and pass the Gaussian white noise signal through the filter to obtain the desired wideband interference signal.

[0029] Generate a pulse interference signal. First, determine the pulse width, pulse period, and power of the pulse interference (for example, a pulse width of 1 μs, a pulse period of 100 μs, and a power of 1 W). Then, use a pulse generation function in the software, such as the rectpuls function in MATLAB or the "Pulse Generator" module in GNU Radio, to generate a pulse signal as the pulse interference signal. Finally, set its pulse width and pulse period to the desired values ​​and adjust its amplitude based on the desired power.

[0030] Step 3: Store and output data.

[0031] First, the generated interference signal data is stored as a file, such as a .mat file in MATLAB or a .dat file in GNU Radio, for subsequent analysis and processing. Then, the data is output through the software interface function to obtain the first signal database described in this embodiment. For example, in MATLAB, the fwrite function is used to write data to a file, and the file is transferred to the target device via a serial port or other communication interface. In GNU Radio, the "File Sink" module can be used to write data to a file, or the "TCP Sink" module can be used to transfer data to other devices via a network.

[0032] (3) Signal processing unit

[0033] The signal processing unit is used to perform normalization and filtering on each signal data in the third signal database. After collecting rich signal data samples, the data samples need to be preprocessed. Normalization is one of the key steps, which regularizes the numerical range of the data to the interval [0,1] or [-1,1], thereby effectively avoiding the problem of gradient disappearance or gradient explosion during neural network training due to excessive differences in numerical magnitude of the data, and accelerating the convergence process. At the same time, filtering algorithms such as Kalman filtering and Wiener filtering are used to accurately filter out interference such as white noise and Gaussian noise in the signal, greatly improving the purity and quality of the data, and facilitating the effective training of subsequent neural networks.

[0034] (4) Signal conversion unit

[0035] The signal conversion unit is used to perform time-frequency conversion on each signal data in the processed third signal database to obtain a time-frequency conversion result for each signal data.

[0036] Common methods for performing time-frequency transformation on signals include short-time Fourier transform (SFT) and wavelet transform (WT). The SFT divides the signal into multiple short time periods, each of which can be adjusted based on the signal characteristics and analysis accuracy requirements. For example, the length of the short time period can be adjusted between a few milliseconds and tens of milliseconds. By performing a Fourier transform on each short time period, the energy distribution of the signal at different times and frequencies is obtained, intuitively displaying the signal's time-frequency characteristics, such as the signal's main frequency components at a given moment and how the frequency changes over time. Furthermore, the wavelet transform possesses time-frequency localization characteristics, making it suitable for analyzing non-stationary signals. By selecting wavelet basis functions (such as Daubechies wavelets and Haar wavelets), the signal's mutation characteristics, such as the start and end moments of pulse interference, can be extracted.

[0037] (5) Feature extraction unit

[0038] The feature extraction unit is used to extract the time domain features and frequency domain features of each signal data according to the time-frequency transformation result of each signal data.

[0039] Based on the results of the time-frequency transform, features are extracted from multiple dimensions in the time-frequency domain. In the time domain, the signal's mean, variance, and peak are extracted. The mean reflects the average strength of the signal, the variance reflects the fluctuation in signal strength, and the peak highlights transient strong pulses in the signal. In the frequency domain, the signal's center frequency, bandwidth, and power spectral density are extracted. The center frequency determines the signal's primary frequency location, the bandwidth describes the frequency range occupied by the signal, and the power spectral density characterizes the signal's power distribution at different frequencies.

[0040] (6) Feature fusion unit

[0041] The feature fusion unit is used to splice the time domain features and frequency domain features of each signal data to obtain the feature vector of each signal data and establish a feature vector set.

[0042] A comprehensive feature vector is formed by using methods such as feature concatenation and weighted summation. For example, the mean, variance, and peak value in the time domain and the center frequency, bandwidth, and power spectrum density in the frequency domain are sequentially concatenated into a one-dimensional vector.

[0043] (7) Model training unit and signal detection unit

[0044] The model training unit is used to train the CNN-LSTM hybrid neural network model using a set of feature vectors. The signal detection unit is used to input real-time collected network signal data into the trained CNN-LSTM hybrid neural network model and output the type of interference signal in the network signal data.

[0045] The CNN-LSTM hybrid neural network model has the ability to extract signal features at multiple levels, model temporal dependencies, and adapt to complex scenarios. The CNN convolutional neural network automatically extracts local spatial and frequency domain features of the signal through convolutional and pooling layers. The LSTM long-short-term memory network solves the vanishing gradient problem of traditional recurrent neural networks (RNNs) through gating mechanisms (input gate, forget gate, and output gate), thereby capturing long-term dependencies in signal sequences. Therefore, this embodiment uses the CNN-LSTM hybrid neural network model to detect the types of interference signals in network signal data.

[0046] Before this, a CNN-LSTM hybrid neural network model needs to be established.

[0047] First, build multiple convolutional layers, typically 3-5. The size of the convolution kernel within each convolutional layer must closely match the signal characteristics. For example, for signals with fine time-frequency texture, a 3×3 convolution kernel can be used to more accurately capture subtle local features. For signals with large frequency spans, a 5×5 convolution kernel is recommended. Convolution operations can automatically reveal the signal's texture features in the time-frequency domain, such as the time-frequency distribution pattern of the signal under a specific modulation scheme and the distribution characteristics of specific frequency components along the time axis.

[0048] Next, add a pooling layer. Depending on your needs, you can use either maximum pooling or average pooling. Maximum pooling emphasizes peaks in the signal, while average pooling smoothes the data and preserves overall trends. The pooling kernel size is set to 2×2. Downsampling reduces the data dimension, easing subsequent computational burden while preserving key features.

[0049] Next, set up 1-3 LSTM layers. The number of neurons in each layer can be flexibly adjusted based on the actual data complexity and task requirements, such as 128 or 256 neurons. The LSTM layer uses a gating mechanism—the input gate, the forget gate, and the output gate—to handle long-term dependencies. The input gate determines the input level of new information, the forget gate controls whether old information is retained or forgotten, and the output gate determines the content of the output information. These three gates work together to deeply learn the long-term characteristics of the signal and accurately identify malicious interference signals.

[0050] Next, add a fully connected layer after the LSTM layer. The function of the fully connected layer is to map the feature vector output by the LSTM layer to a fixed-dimensional space, further deeply fuse the features, eliminate the spatial correlation between features, and enable the model to comprehensively consider signal characteristics from a global perspective.

[0051] Finally, add the output layer. This layer uses a softmax activation function, which converts the model's output into the probability of each signal belonging to a different category. For example, in the malicious interference signal identification task, the output results may include probabilities for normal signals, narrowband interference signals, broadband interference signals, and other categories. The maximum probability value can be used to accurately classify and identify malicious interference signals.

[0052] After the CNN-LSTM hybrid neural network model is established, it needs to be trained.

[0053] First, choose a loss function. You can choose mean squared error (MSE) or cross entropy loss, which measures the difference between the model's predicted values ​​and the true values.

[0054] Next, select an optimizer. You can choose Adam or RMSprop to update the model parameters to minimize the loss function.

[0055] Finally, the model is trained using the feature vector set. During each training cycle, the feature vectors are fed into the model, the loss function is calculated, and the model parameters are updated through backpropagation. During training, the model can be evaluated using a validation set to avoid overfitting.

[0056] 2. Frequency hopping sequence update module and first message sending module

[0057] The frequency hopping sequence updating module is used to update the frequency hopping sequence and trigger the first message sending module to operate when the interference signal detection module detects a new interference signal. The frequency hopping sequence updating module is composed of an initial sequence generating unit, an initial sequence adjusting unit and a frequency hopping sequence encrypting unit.

[0058] (1) Initial sequence generation unit

[0059] The initial sequence generating unit is used to generate an initial frequency hopping sequence by adopting a Logistic chaotic map.

[0060] Taking advantage of the unique randomness of chaotic systems and their extreme sensitivity to initial conditions, the Logistic chaotic map is used to generate the initial frequency hopping sequence. The iterative formula of the Logistic map is: ,in, x n is the sequence value before iteration (initial value x 0 needs to avoid 0 and 1). x n ∈(0,1), x n+1 is the sequence value after iteration, is the control parameter that determines the dynamic behavior of the system. ∈(0,3), the sequence converges to a stable value. =3.5633456…, the system enters a chaotic state (completely random and unpredictable). =4 (at this time, the chaotic characteristics are most significant and the sequence ergodicity is good).

[0061] The steps to generate the initial frequency hopping sequence are:

[0062] First, initialize the parameters. Set the control parameters =4 (ensure chaotic state), select initial value x 0 (e.g. x 0=0.3, needs to be confidential and unique), determine the number of available channels M and the hopping sequence length LThe length of the initial sequence is precisely determined based on communication requirements and system bandwidth. For example, in scenarios with narrow bandwidth and low communication rates, an initial sequence of 512 bytes can be generated. In broadband, high-speed communication scenarios, to meet the needs of more frequency hopping points, an initial sequence of 1024 bytes or even longer can be generated.

[0063] Then, the chaotic sequence is generated iteratively. x Start at 0 and iterate L Logistic mapping, we get the sequence { x 0, x 2,…, x L}. For example, when =4, x When 0=0.3, x 1=4×0.3×(1-0.3)=0.84, x 2=4×0.84×(1-0.84)=0.5376, x 3=4×0.5376×(1-0.5376)=0.9942…….

[0064] Next, the chaotic sequence is quantized into channel indices. x n ∈(0,1) is mapped to channel number 0~ M -1, commonly used methods are: ,in, k n is the channel number, is the floor function, M is the number of channels used, x n is the iterative sequence value. For example, when M =8, (corresponding to channel 6), (corresponding to channel 4), (Corresponding to channel 7)……

[0065] Finally, generate the initial frequency hopping sequence. Arrange the quantized channel indexes in order to obtain the initial frequency hopping sequence of length L. Initial frequency hopping sequence = [ k n1 , k n2 , k 3, … , k L ]=[6,4,7…] (corresponding to the hopping order of frequency channels).

[0066] (2) Initial sequence adjustment unit

[0067] The initial sequence adjustment unit is used to adjust the order of each element in the initial frequency hopping sequence using the cyclic shift method, and to change the value of each element in the initial frequency hopping sequence using the exclusive OR operation.

[0068] On the one hand, multiple rounds of optimization operations are performed on the initial frequency hopping sequence, and cyclic shift is used to cyclically shift the elements in the sequence according to certain rules to change the order of the sequence. Circular shift means that the elements in the sequence are shifted by a certain number of positions in a fixed direction (left or right), and the first and last elements are connected cyclically. Left cyclic shift: The sequence is shifted to the left. t Position, front t Elements are shifted to the end in sequence, for example: the sequence [ a , b , c , d ] shifted left by 2 bits, we get the sequence [ c , d , a , b ]; Right circular shift: the sequence moves to the right t Position, front t Elements are shifted to the beginning in sequence, for example: the sequence [ a , b , c , d ] shifted right by 2 bits, we get the sequence [ c , d , a , b ] (The result is the same as the left shift by 2 bits, only the direction is different). The circular shift optimization method is as follows:

[0069] Assume that the initial frequency hopping sequence is S =[ s 0, s 1, s 2,…, s L-1 ], the length is L, the number of optimization rounds is N, and the shift parameter of each round is t n ( n =1,2,…, N ).

[0070] A single round of cyclic shift operation includes the following steps:

[0071] Step 1: Determine the shift direction and number of bits.

[0072] First, determine the displacement direction, which is usually fixed to the left to simplify the implementation.

[0073] Then, determine the number of shifts t . Can be set t is a fixed value, such as t =1 (bit-by-bit shift),t = L / 2 (half shift), can also be set t is a dynamic value, that is, it is calculated in real time based on the key, timestamp or spectrum status, such as t = hash ( k , t ) mod L ,in, k is the key.

[0074] Step 2: Perform the shift operation. The left circular shift formula is: S’ =[ s t , s t+1 , s t+2 ,…, s L-1 , s 0, s 1,…, s t-1 ], the right circular shift formula is: S’ =[ s L-t , s L-t+1 , s L-t+2 ,…, s L-1 , s 0, s 1,…, s L-t-1 ], S’ is the shifted sequence. For example, for the initial sequence [0,1,2,3,4,5,6,7], the left shift t =3, sequence length L =8, the shifted sequence is [3,4,5,6,7,0,1,2].

[0075] Multiple rounds of cascade optimization include the following steps:

[0076] Through multiple rounds of cyclic shifts with different parameters, the sequence order is further disrupted to improve the optimization effect. Including: the number of shift bits increases, that is, the number of shift bits in each round t n = t n -1+△ t (such as △ t =1 or a random number); alternating direction, that is, left shift and right shift are performed alternately (such as left shift in the first round, right shift in the second round); key driven, that is, each round of shift parameter t n Generated by key stream or chaotic sequence (such ast n =logistic( n ) mod L For example, for the initial sequence [6,4,7,2,5,3,1,0]) (assuming it is generated by Logistic mapping), 2 rounds of left shift, the sequence length L =8, the sequence obtained after the first round of left shift is t 1=2: [7,2,5,3,1,0,6,4], t 2=3: [1,0,6,4,7,2,5,3].

[0077] On the other hand, an XOR operation is used to XOR the sequence elements with a specific binary sequence to change the value of the element and further enhance the randomness of the sequence. The following steps are included:

[0078] Step 1: Determine the mask sequence.

[0079] The available mask sequences include: fixed mask, that is, all elements use the same binary sequence (such as 0011); dynamic mask, that is, different masks are generated according to the element position, timestamp or other rules (such as the i-th element uses mask i%4); key stream mask, that is, a mask sequence of the same length as the original sequence is generated through a pseudo-random number generator (such as LFSR, AES-CTR mode).

[0080] Step 2: Perform element-wise XOR operation.

[0081] Original sequence S =[ s 1, s 2,..., s n ] for each element s i ,implement , where mask is the mask value of the corresponding position, is the exclusive OR operator.

[0082] (3) Frequency hopping sequence encryption unit

[0083] The frequency hopping sequence encryption unit is used to encrypt the adjusted frequency hopping sequence.

[0084] The optimized sequence is encrypted using the AES encryption algorithm. During the encryption process, the sequence is encrypted using a specific key, which can be shared between the communicating parties through a secure key management protocol. This ensures the security of the frequency hopping sequence, effectively preventing it from being cracked and predicted by the enemy, and ensuring the confidentiality and anti-interference capabilities of the communication. The specific method is as follows:

[0085] Step 1: Format the optimized sequence.

[0086] This includes data type conversion and block padding. Data type conversion involves converting optimized sequences (such as integers and floating-point numbers) into byte streams (e.g., converting floating-point numbers to bytes using the IEEE 754 standard, or serializing integers to a fixed byte length). Block padding involves using a padding scheme (such as PKCS#7) to fill in the total length of a sequence when it is not an integer multiple of 128 bits (16 bytes). For example, if the sequence length is 20 bytes, 4 bytes of padding (value 0x04) are added, and the padding is removed after encryption.

[0087] Step 2: Configure AES encryption parameters.

[0088] For example, configure the key length to 256 bits (suitable for highly sensitive data), set the encryption mode to CTR mode (suitable for continuous sequence encryption), set the initialization vector (IV) to randomly generate a 128-bit IV, and set key management to dynamic key.

[0089] Step 3: Encrypt the optimized sequence using AES.

[0090] Take CTR mode as an example:

[0091] First, a random number is generated. Combined with the initialization vector (IV), it generates the initial value of the counter, ensuring that the counter is unique each time it is encrypted.

[0092] Then, a counter is generated. The counter increments from the initial value, generating a unique value for each packet, and a key stream is generated through AES encryption.

[0093] Finally, XOR encryption is performed. The key stream is XORed with the optimized sequence of byte streams bit by bit to generate ciphertext. For example, ciphertext = plaintext AES (key, counter value).

[0094] Step 4: Ciphertext output and synchronization.

[0095] This includes encapsulation format and synchronization mechanism. The encapsulation format encapsulates information such as the ciphertext, IV (if any), and padding length into a specific format (such as JSON or a binary header) to facilitate parsing by the decryption end. The synchronization mechanism synchronizes the counters between the encryption and decryption ends during streaming (e.g., through timestamp or sequence number calibration).

[0096] Furthermore, after the frequency hopping sequence is updated, the first message sending module is triggered. At this point, the first message sending module needs to send notification messages to both the link layer and the network layer to trigger a cross-layer coordinated response and achieve global linkage for interference avoidance. The notification message sent to the link layer contains the current available frequency band information, including spectrum hole update information and frequency hopping sequence update parameters. The spectrum hole update information includes the newly released spectrum hole range (such as the starting frequency and bandwidth of the available frequency band) after frequency hopping avoidance of the original interference frequency band, as well as the list of available frequency bands corresponding to the current valid frequency hopping sequence (such as the set of frequencies in the hopping pattern). The frequency hopping sequence update parameters include the sequence period (such as > 2³²), chaotic mapping parameters (such as Logistic mapping μ = 4), encryption key identifiers (such as AES key version numbers), and synchronization header information (such as sequence index values). The notification message sent to the network layer contains the current interference frequency band information, including interference characteristics and interference constraints. Interference characteristics include: interference classification results (such as narrowband interference, broadband interference, and pulse interference), as well as the time-frequency location of the interference (such as center frequency, bandwidth, and duration). Interference constraints include: interference band ban tickets, which mark the frequency bands of links affected by interference as "unavailable" and require the network layer to remove or increase the impedance weight of the corresponding edges in the graph model. For example, if the frequency band used by a link is occupied by interference, the channel quality (CQ) in its edge weight is set to 0, and the impedance factor Z = 0.5 × 0 + 0.3NL + 0.2ES increases significantly, forcing the topology optimization algorithm to bypass the link.

[0097] It should be noted that: (1) The physical layer is also used to update the frequency hopping sequence when receiving a notification message from the link layer or a notification message from the network layer, and trigger the first message sending module to work. (2) The first message sending module is also used to send a notification message to the link layer and the network layer when there is no clean frequency band in the backup frequency band, to inform that there is no clean frequency band. (3) The physical layer also includes a backup frequency band scanning module, which is used to scan the backup frequency band after receiving a notification message from the link layer. If there is a clean spectrum in the backup frequency band, the frequency hopping sequence update module is triggered to work. If there is no clean spectrum in the backup frequency band, the first message sending module is triggered to work. The above supplementary explanation will be expanded in the subsequent explanation of the link layer and the topology layer.

[0098] (2) Link Layer

[0099] The link layer consists of a channel status detection module, a coding mode screening module, a wireless channel coding module, a relay node screening module, a transmission efficiency detection module, a second message sending module, a coding mode adjustment module, a relay node switching module and a cache space allocation module. It is used to determine the current channel coding mode based on the notification message sent by the physical layer, and detect the packet loss rate of each data transmission path under the current channel coding mode. If the packet loss rate of one or more data transmission paths is greater than the first threshold, corresponding notification messages are sent to the physical layer and the network layer respectively, and the interference source is located through cross-layer collaboration and congestion is dynamically avoided.

[0100] (1) Channel status detection module

[0101] The channel status detection module is used to detect the current wireless channel status according to the latest available frequency band information.

[0102] As mentioned above, after the frequency hopping sequence is updated, the notification message sent by the physical layer to the link layer includes: spectrum hole update information and frequency hopping sequence update parameters. The spectrum hole update information includes: the range of spectrum holes newly released after the original interference frequency band is avoided due to frequency hopping (such as the starting frequency and bandwidth of the available frequency band), and the list of available frequency bands corresponding to the current valid frequency hopping sequence (such as the set of frequency points in the frequency hopping pattern); the frequency hopping sequence update parameters include: sequence period (such as >2³²), chaotic mapping parameters (such as Logistic mapping μ=4), and encryption key identifier (such as AES key version number) and synchronization header information (such as sequence index value). The channel state detection module can use the minimum mean square error (MMSE) algorithm based on the available frequency band information to estimate the channel fading coefficient through continuous iterative optimization, and then evaluate the channel quality. The following operations are required to estimate the channel fading coefficient:

[0103] 1) Extract available frequency band parameters.

[0104] On the one hand, get the starting frequency from the "newly released spectrum hole range" f start and bandwidth B , calculate the center frequency of the available frequency band f c = f start + B / 2. On the other hand, a discrete frequency point set is extracted from the "available frequency band list corresponding to the valid frequency hopping sequence" , as the frequency domain sampling point of the MMSE algorithm, K is the frequency point number.

[0105] 2) Construct the frequency domain measurement matrix. Convert the discrete frequency points into normalized frequencies w k =2πfk / f s , used for subsequent channel frequency response modeling, f s is the sampling frequency.

[0106] 3) Obtain the sequence period and chaos parameters. 32 " and "Logistic Mapping Parameters μ =4" Generate chaotic frequency hopping sequence { c ( n )}, as the spreading sequence at the transmitting end, used to suppress multipath interference.

[0107] 4) Synchronization header and key identifier: On the one hand, the "sequence index value" is used to achieve time synchronization between the receiver and the transmitter, ensuring the alignment of the frequency hopping pattern; on the other hand, the "AES key version number" is used to decrypt the frequency hopping sequence and restore the original spread spectrum signal.

[0108] 5) Establish a channel frequency domain model. Considering the wireless channel as a frequency selective fading channel, its frequency domain response can be expressed as: ,in, is the frequency response function (complex function), which is used to describe the response characteristics of the system to input signals of different frequencies. l is the path number, L is the number of multipaths, h l For the l The fading coefficient of the path (a complex random variable), For the corresponding delay, j is the imaginary unit, is the angular frequency.

[0109] 6) Establish a received signal model. In frequency hopping communication, the received signal can be expressed as: ,in, Indicates that the received signal is n The value at the sampling moment, is the center frequency f k The corresponding transmitted signal (modulated frequency hopping sequence), h ( n ) is the channel impulse response, and its frequency domain response is H ( w k ), w(n) is additive white Gaussian noise (AWGN), and the power spectral density is N 0.

[0110] 7) Define the mean square error (MSE) cost function. The mean square error (MSE) cost function is: ,in, is the cost function, h is the fading coefficient, is the fading coefficient h The estimated value of E is the expected value operator.

[0111] 8) Initialize frequency domain sampling and observation vector.

[0112] First, the frequency domain sampling points are generated using the available frequency band list. Collect frequency domain samples of the received signal R ( w k )= S ( w k ) H ( w k )+ W ( w k ),in, R ( w k ) is the frequency domain of the received signal, S ( w k ) is the frequency domain of the transmitted signal, H ( w k ) is the frequency response of the channel, W ( w k ) is the frequency domain of additive noise, w k Indicates the k frequency points.

[0113] Then, the observation matrix is ​​constructed. The frequency domain samples are expressed as a vector: R = SH + W, where S is the frequency domain matrix of the transmitted signal, H is the channel frequency domain response vector, and W is the noise vector.

[0114] 9) Use the recursive least mean square error (RLS) algorithm to achieve iterative optimization (suitable for time-varying channels).

[0115] First, calculate the gain matrix, the formula is ,in, K ( n ) indicates that n The gain matrix at each moment is used to adjust the filter coefficients to minimize the error; is the error covariance matrix; S ( n ) indicates that n The reference signal at a certain moment; for S( n )'s conjugate transpose; is the noise power.

[0116] Then, update the fading coefficient, the formula is: ,in, Indicates in n The channel estimation value at time instant is Indicates in n- The channel estimation value at one moment, Indicates in n The received signal at a moment.

[0117] Finally, update the covariance matrix as follows: ,in, Indicates in n The covariance matrix at each moment, I is the identity matrix, Indicates in n- The covariance matrix at 1 moment.

[0118] 10) Use the available frequency band information to constrain the iterative process.

[0119] Spectrum hole range constraint: During iteration, the fading coefficients are estimated only for the frequencies within the newly released spectrum holes, ignoring the frequencies in the interference band, thus reducing computational complexity.

[0120] Frequency hopping sequence period constraint: Since the frequency hopping sequence period T>2 32 ,It can be considered that the channel time variation is weak in a short time ,period, so a fixed step size iteration is adopted within a single ,frequency hopping cycle, and the initial estimation value is reset only ,after the cycle ends based on the new synchronization header information.

[0121] After detecting the current channel fading coefficient, the channel fading coefficient is constructed and input into the support vector machine, which outputs the wireless channel status corresponding to the current channel fading coefficient. It should be noted that the wireless channel status can be divided into multiple levels such as excellent, good, medium, and poor according to actual needs.

[0122] (2) Encoding method screening module

[0123] The coding mode screening module is used to screen out the current channel coding mode from the mapping relationship table according to the current wireless channel status and wireless communication service type.

[0124] Before selecting channel coding methods, it's necessary to categorize wireless communication services based on actual service scenarios and analyze the corresponding Quality of Service (QoS) requirements for each service. For example, wireless communication services can be divided into voice, video, and data services. Voice services, as they involve real-time communication, have high real-time requirements and low latency tolerance (end-to-end latency must be within tens of milliseconds; otherwise, call quality and user experience will be severely impacted). Video services have high bandwidth and image quality requirements. High-definition video streams require large bandwidth to ensure smooth playback, while also imposing certain bit error rate limits to avoid image artifacts such as pixelation and stuttering. Data services have specific requirements for data accuracy and throughput. For example, file transfers must ensure complete and error-free data transmission while maximizing transmission rates and minimizing transmission times.

[0125] On this basis, corresponding channel coding methods are developed based on different combinations of service types and channel conditions, and a mapping table is established. This mapping table contains the channel coding method corresponding to each combination. These combinations refer to the combinations between wireless communication services and wireless communication conditions. For example, when channel conditions are excellent, high-speed and low-redundancy coding, such as low-density parity-check (LDPC) codes with a bit rate of 7 / 8, can be used for voice services to ensure real-time performance and make voice calls smoother and more natural. For video services, more bandwidth resources are allocated, and efficient video coding standards, such as H.265, are used to improve image quality and present clearer images. When channel conditions are poor, error-correcting coding, such as Turbo codes with a bit rate of 1 / 3, can be used for data services. This sacrifices transmission rate to ensure data accuracy, reduce bit error rates, and guarantee file transfer integrity.

[0126] In addition, it should be noted that when selecting the channel coding method, a fusion of adaptive coded modulation and network coding is employed. Specifically: 1) For adaptive coded modulation, when the channel conditions are good, signal transmission is stable, and interference is low, a high-order coded modulation method is selected, such as 64-QAM (quadrature amplitude modulation) combined with a high-rate channel code, such as a Turbo code with a code rate of 3 / 4. 64-QAM can transmit more data bits per unit bandwidth. Combined with the powerful error correction capabilities of the high-rate Turbo code, it can significantly increase the data transmission rate to meet the needs of high-speed data services. When the channel conditions deteriorate, such as fading or increased interference, a low-order coded modulation method is promptly switched to, such as QPSK (quaternary phase shift keying) combined with a low-rate channel code, such as a convolutional code with a code rate of 1 / 2. QPSK has stronger interference resistance, and the low-rate convolutional code further enhances the signal's error correction capabilities, effectively ensuring reliable signal transmission in harsh channel environments. 2) For network coding, network coding is performed on multiple data packets at the transmitter. For example, using random linear network coding, the transmitter randomly generates a set of coding coefficients, linearly combines multiple original data packets according to these coefficients, generates a coded data packet, and then transmits it. The receiver collects a sufficient number of coded data packets and uses linear algebraic operations, such as Gaussian elimination, to decrypt the original data packets. Network coding can significantly improve data transmission reliability and reduce retransmissions, especially in complex network scenarios with communication congestion and packet loss, effectively improving network performance and increasing data transmission efficiency. 3) Combining adaptive coding and modulation with network coding to form a joint channel coding strategy. Dynamically adjust the coding and modulation parameters and the degree of network coding based on different channel conditions and service requirements. For example, when the channel is lightly congested, adaptive coding and modulation is primarily used to address this, adjusting the coding and modulation scheme, such as switching from 64-QAM to 16-QAM, while maintaining low network coding redundancy. When the channel is severely congested, the network coding strength is increased, increasing the redundancy of the coded data packets, such as by expanding the range of coding coefficients. At the same time, an appropriate coding and modulation scheme, such as QPSK with a low-rate convolutional code, is combined to ensure reliable data transmission and enhance the network's resilience to congestion.

[0127] Based on the above explanations, the mapping relationship table may be in the form shown in Table 1.

[0128] Combination Channel coding method Voice service & wireless channel status is excellent Low-density parity-check code (LDPC) with a code rate of 7 / 8 Data service & wireless channel status is good Turbo code with a code rate of 1 / 3 Data service & wireless channel status is poor Network Coding …… ……

[0129] Table 1 Mapping relationship between channel status-service type combination mode and channel coding mode

[0130] (3) Wireless channel coding module

[0131] The wireless channel coding module is used to perform wireless channel coding according to the current channel coding mode.

[0132] (4) Relay node screening module

[0133] The relay node screening module is used to select one or more relay nodes from each user equipment using a distributed algorithm according to the current status of each user equipment when the target user equipment cannot receive signal data sent by the source user equipment.

[0134] In a communication network, nodes discover each other through broadcast messages. Each node periodically broadcasts its own status information to the surrounding area, including remaining power, storage capacity, device identification, etc. After receiving the broadcast message, other nodes store it in the local node information table and use the distributed algorithm to select suitable user devices as temporary forwarding nodes based on the remaining power, storage capacity, distance between the selected device and the source node, and distance between the selected device and the destination node of each node. For example, priority is given to user devices with sufficient remaining power and the ability to work continuously for a long time; devices with large storage capacity and the ability to cache large amounts of data; and devices that are close to the source node and destination node, with short signal data transmission paths and low losses. The formula for the distributed algorithm is: ,in, P i Representation node i The probability of being selected, w e represents the energy weight coefficient, w s Represents sensor data, w d represents the weight coefficient of distance, e i Representation node i The normalized residual energy of E i is a node i The remaining energy, E max is the maximum energy of a node in the network, s i Representation node i The normalized sensor data volume, S i is a node i The amount of data collected, S max is the maximum amount of data for a node in the network, Represents the distance from the source node to the node i The normalized distance, d src,i From source node to node i distance, D max is the maximum distance between nodes in the network, Represents a slave node i The normalized distance to the destination node, d dst,i It is a slave node i The distance to the destination node, ε Represents a positive number.

[0135] 6) Cache space allocation module

[0136] The cache space allocation module is used to allocate cache space to each relay node.

[0137] The selected relay nodes are equipped with a dedicated cache module to cache received data. The cache module can use a cache management algorithm (such as the Least Recently Used (LRU) algorithm) to rationally manage cache space and ensure efficient cache utilization. When the destination node cannot directly receive data due to channel congestion, signal interference, or other reasons, the forwarding node forwards the cached data to the destination node or other relay nodes based on the routing strategy. The routing strategy can use a location-based routing algorithm, such as the Greedy Peripheral Stateless Routing (GPSR) algorithm. This algorithm selects the next-hop forwarding node closest to the destination node and that is reachable based on the node's location information. Through hop-by-hop forwarding, it ensures that data is efficiently transmitted to the destination node.

[0138] 6) Transmission efficiency detection module and second message sending module

[0139] The transmission efficiency detection module is used to detect the packet loss rate of each data transmission path under the current channel coding mode. If the packet loss rate of one or more data transmission paths is greater than or equal to the first threshold, the second message sending module is triggered to work. If the packet loss rate of each data transmission path is less than the first threshold, the current channel coding mode is maintained.

[0140] The second message sending module is used to send notification messages to the physical layer and the network layer. When the packet loss rate of one or more data transmission paths is greater than or equal to a first threshold, the notification message sent to the physical layer is used to trigger the operation of the backup spectrum scanning module, and the notification message sent to the network layer is used to request the network layer to re-determine the data transmission path.

[0141] Reasons for increased packet loss rates in data transmission paths include: 1) the presence of undetected interference signals (such as burst interference or narrowband interference) in the current frequency band, or the original frequency hopping sequence has been "tracked" by the interference source, resulting in reduced physical layer transmission reliability. 2) The presence of bottleneck nodes in the network topology (such as nodes with low residual energy, high processing load, high link latency, or poor channel quality). When the transmission efficiency detection module detects that the packet loss rate of one or more data transmission paths is greater than the first threshold, it indicates that the transmission quality of the current data transmission path has significantly deteriorated and may face the risk of congestion or interference. At this time, the physical layer is notified to scan for alternative frequency bands and the network layer is requested to recalculate the path.

[0142] When the physical layer receives a notification message from the link layer's second message sending module, it triggers the backup frequency band scanning module. Using short-time Fourier transforms (STFTs) or wavelet transforms, it generates a time-frequency map to identify the interference type (e.g., narrowband, broadband, or pulsed) and distribution range in the current frequency band. This determines whether the frequency band used by the original path is blocked. It also identifies "spectrum holes" unoccupied by interference, providing available resource information for frequency hopping or topology-layer path planning. If clean spectrum exists in the backup frequency band, the frequency hopping sequence update module is triggered to generate a new frequency hopping sequence to avoid the interfering frequency band. If no clean spectrum exists in the backup frequency band, a notification message is sent to the link layer and network layer to inform them that there is no clean frequency band and that the link layer must adjust the channel coding scheme and the network layer must update the network topology to reduce communication congestion. For details on adjusting the channel coding scheme at the link layer, refer to the following explanation of the coding scheme adjustment module; for updating the network topology at the network layer, refer to the following explanation of the network layer.

[0143] 7) Encoding adjustment module

[0144] The coding mode adjustment module is used to adjust the current channel coding mode and trigger the wireless channel coding module to work when receiving a notification message that there is no clean frequency band.

[0145] Adjusting the current channel coding method includes: reducing the modulation order.

[0146] The modulation order is reduced by presetting the minimum available SNR threshold for different modulation orders, such as SNR ≥ 18dB for 16QAM, SNR ≥ 10dB for QPSK, and SNR ≥ 3dB for BPSK. When the real-time monitored SNR falls below the threshold for the current modulation order, a downgrade is triggered (e.g., from 16QAM to QPSK). For example, if the original configuration was 16QAM with rate 3 / 4 LDPC (high efficiency but weak interference immunity), the new configuration, after adjustment, is QPSK with rate 1 / 2 convolutional code (reduced throughput but significantly lower BER).

[0147] 8) Relay node switching module

[0148] The relay node switching module is used to trigger the relay node screening module to work when receiving a notification message sent by the network layer.

[0149] For the relay node switching module, please refer to the following explanation about the network layer.

[0150] (3) Network layer

[0151] The network layer consists of a network topology generation module, a transmission path generation module, a node impedance detection module, a third message sending module and a network topology reconstruction module. It is used to determine the current data transmission path based on the notification message sent by the physical layer or the notification message sent by the link layer, and detect the impedance of each node on the current data transmission path. If the impedance of one or more nodes is greater than the second threshold, corresponding notification messages are sent to the physical layer and link layer respectively.

[0152] (1) Network topology generation module

[0153] The network topology generation module is used to generate a topology structure corresponding to the wireless communication network and use the latest interference frequency band information to assign values ​​to the edges of the topology structure.

[0154] The wireless communication network is abstracted and constructed as a graph structure, where nodes represent communication devices and edges represent communication links between devices. Each node and edge is assigned corresponding attributes. Node attributes include the node's geographic location, remaining energy, and data processing capacity; edge attributes include information about the interference frequency band of the link.

[0155] (2) Transmission path generation module

[0156] The transmission path generation module is used to input the topology structure into the graph convolutional neural network and output the current data transmission path.

[0157] Before doing this, do the following:

[0158] 1) Preprocess the topology structure. Including:

[0159] Establish node feature matrix X :Each row corresponds to a node, and the columns contain node attributes (such as remaining power, storage capacity, and distance from the source / destination node). The dimension is N × F . N is the number of nodes, F is the number of features.

[0160] Build an adjacency matrix A : describes the node connection relationship in an undirected graph A i,j Representation nodei With node j The connection relationship, A i,j =1 indicates a node i With node j Directly connected, otherwise A i,j =0.

[0161] Build the normalized Laplace matrix L = D -1 / 2 ( D - A ) D -1 / 2 ,in, D is the degree matrix (diagonal matrix), D i,j For nodes i With node j The degree of connection, .

[0162] 2) Build a graph convolutional neural network. This includes:

[0163] Define the propagation rules of the GCN layer: ,in, H (l+1) Indicates in l +1 layer output feature matrix, is the normalized adjacency matrix with self-loops, I is the identity matrix, H (l) For the l The node feature matrix of the layer is initially X , W (l) is the learnable weight matrix, is the activation function (such as ReLU, Sigmoid).

[0164] Build a graph convolutional neural network structure. The input layer is used to receive the node feature matrix X and the adjacency matrix A; the hidden layer includes multiple GCN layers, which extract topological features layer by layer. The low-level GCN layer is used to capture the node's direct neighbor information (such as the node status within a hop range), and the high-level GCN layer is used to capture the global topological features (such as the connectivity of multi-hop paths); the output layer is built according to the actual task. If the output path probability distribution is used, the Softmax activation function can be used, and the dimension is N (the probability of each node being the next hop), if the edge selection of the output path can be designed as a binary classification (whether the edge is in the path), the dimension is N × N .

[0165] On this basis, the above-mentioned graph convolutional neural network is trained. The preprocessed topology structure is input into the trained graph convolutional neural network, and the current data transmission path is output.

[0166] (3) Node impedance detection module

[0167] The node impedance detection module is used to detect the impedance of each node on the current data transmission path.

[0168] (4) The third message sending module

[0169] The third message sending module is configured to send corresponding notification messages to the physical layer and link layer when the impedance of one or more nodes exceeds a second threshold. When the impedance of each node is less than the second threshold, the current data transmission path is maintained. The notification message sent to the physical layer triggers the operation of the frequency hopping sequence update module, and the notification message sent to the link layer requests the link layer to switch relay nodes.

[0170] Node impedance (Z value) is a comprehensive indicator that measures a node's data processing capabilities, link quality, and load status. When the topology layer detects that a node's impedance exceeds a threshold, it indicates that the node has become a bottleneck or "blocking point" for data transmission in the network. This can lead to increased link latency, increased packet loss, or even disconnection. The node is no longer able to efficiently process or forward data, and the blockage must be removed through the physical and link layers.

[0171] When the physical layer receives a notification message from the third message sending module, it triggers the frequency hopping sequence update module to operate. Refer to the explanation of the frequency hopping sequence update module above. When the link layer receives a notification message from the third message sending module, it triggers the relay node switching module to operate. Relay node switching uses a re-screening method for relay nodes. Therefore, when the link layer receives a notification message from the network layer, the relay node switching module triggers the relay node screening module to reselect a relay node.

[0172] (5) Network topology reconstruction module

[0173] The network topology reconstruction module is used to delete the communication link with the lowest utilization rate in the topology structure and trigger the transmission path generation module to work when receiving a notification message that there is no clean frequency band or receiving a notification message sent by the link layer.

[0174] The communication link with the lowest utilization in this embodiment is the link with the lowest utilization among the redundant links (e.g., the link with the lowest utilization among multiple parallel links). Each link in the topology is screened, and once the link with the lowest utilization is found, it is deleted by deleting the router interface configuration.

[0175] It should be noted that the mapping relationship table includes a channel coding mode corresponding to each combination mode, and the combination mode refers to a combination mode between a wireless communication service and a wireless communication state.

[0176] It should be understood that the terms "system," "device," "unit," and / or "module" used in this specification are a method for distinguishing different components, elements, parts, portions, or assemblies at different levels. However, if other terms can achieve the same purpose, the terms may be replaced by other expressions.

[0177] As used in this specification and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not refer to the singular but also include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0178] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0179] It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification for understanding and reading by those familiar with this technology, and are not used to limit the conditions for implementation of the present invention. Therefore, they have no substantial technical significance. Any modification of the structure, change in the proportional relationship, or adjustment of the size should still fall within the scope of the technical content disclosed by the present invention without affecting the efficacy and purpose of the present invention. At the same time, the terms such as "upper", "lower", "left", "right", "middle", etc. quoted in this specification are only for the convenience of description and are not used to limit the scope of implementation of the present invention. Changes or adjustments in their relative relationships should also be regarded as the scope of implementation of the present invention without substantially changing the technical content.

Claims

1. A system for reducing communication congestion in a wireless communication network, characterized in that: include: The physical layer is used to detect interference signals in the communication environment in real time. When a new interference signal is detected, a notification message is received from the link layer, or a notification message is received from the network layer, the physical layer updates the frequency hopping sequence and sends corresponding notification messages to the link layer and the network layer respectively. The link layer is configured to determine the current channel coding mode based on the notification message sent by the physical layer or the notification message sent by the network layer, and detect the packet loss rate of each data transmission path under the current channel coding mode. If the packet loss rate of one or more data transmission paths exceeds a first threshold, the link layer sends corresponding notification messages to the physical layer and the network layer respectively. The network layer is configured to determine the current data transmission path based on the notification message sent by the physical layer or the notification message sent by the link layer, and detect the impedance of each node on the current data transmission path. If the impedance of one or more nodes exceeds a second threshold, the network layer sends corresponding notification messages to the physical layer and the link layer respectively. The physical layer includes: Interference signal detection module, used for real-time detection of interference signals in the communication environment; A frequency hopping sequence updating module, configured to update the frequency hopping sequence and trigger the operation of the first message sending module when a new interference signal is detected or a notification message is received from the network layer; A first message sending module is used to send corresponding notification messages to the link layer and the network layer respectively; A spare frequency band scanning module is configured to scan the spare frequency band upon receiving a notification message from the link layer. If a clean spectrum exists in the spare frequency band, the frequency hopping sequence updating module is triggered to operate. If a clean spectrum does not exist in the spare frequency band, the first message sending module is triggered to operate. After the frequency hopping sequence is updated, the notification message sent to the link layer includes: the current available frequency band information; After the frequency hopping sequence is updated, the notification message sent to the network layer includes: current interference frequency band information; When there is no clean frequency band in the backup frequency band, a notification message is sent to the link layer and the network layer to inform them that there is no clean frequency band. The link layer includes: a transmission efficiency detection module, configured to detect the packet loss rate of each data transmission path under the current channel coding mode, and trigger the operation of the second message sending module if the packet loss rate of one or more data transmission paths is greater than or equal to a first threshold; and maintain the current channel coding mode if the packet loss rate of each data transmission path is less than the first threshold; A second message sending module, configured to send notification messages to the physical layer and the network layer; The coding mode adjustment module is used to adjust the current channel coding mode and trigger the operation of the wireless channel coding module when receiving a notification message that there is no clean frequency band. Adjusting the current channel coding mode includes: reducing the modulation order; The relay node switching module is used to trigger the relay node screening module to work when receiving the notification message sent by the network layer; When the packet loss rate of one or more data transmission paths is greater than or equal to a first threshold, a notification message is sent to the physical layer to trigger the operation of the backup spectrum scanning module, and a notification message is sent to the network layer to request the network layer to redetermine the data transmission path; The network layer includes: A third message sending module is configured to send corresponding notification messages to the physical layer and the link layer respectively when the impedance of one or more nodes is greater than or equal to the second threshold, and to maintain the current data transmission path when the impedance of each node is less than the second threshold; The network topology reconstruction module is used to delete the communication link with the lowest utilization rate in the topology structure and trigger the transmission path generation module to work when receiving a notification message that there is no clean frequency band or a notification message sent by the link layer; When the impedance of one or more nodes is greater than or equal to the second threshold, the notification message sent to the physical layer is used to trigger the frequency hopping sequence update module to operate, and the notification message sent to the link layer is used to request the link layer to switch the relay node.

2. The system for reducing communication congestion in a wireless communication network according to claim 1, wherein: The interference signal detection module includes: A signal acquisition unit, configured to acquire network signal data in a communication environment to obtain a first signal database; A signal simulation unit, configured to simulate a variety of interference signal data in a communication environment to obtain a second signal database; a signal fusion unit, configured to fuse the first signal database with the second signal database to obtain a third signal database; a signal processing unit, configured to perform normalization and filtering on each signal data in the third signal database; A signal conversion unit, configured to perform time-frequency conversion on each signal data in the processed third signal database to obtain a time-frequency conversion result for each signal data; A feature extraction unit, configured to extract time domain features and frequency domain features of each signal data according to a time-frequency transformation result of each signal data; A feature fusion unit is used to combine the time domain features and frequency domain features of each signal data to obtain the feature vector of each signal data and establish a feature vector set; A model training unit is used to train the CNN-LSTM hybrid neural network model using a set of feature vectors; The signal detection unit is used to input the real-time collected network signal data into the trained CNN-LSTM hybrid neural network model and output the type of interference signal in the network signal data.

3. The system for reducing communication congestion in a wireless communication network according to claim 1, wherein: The frequency hopping sequence update module includes: An initial sequence generating unit, used for generating an initial frequency hopping sequence by using a Logistic chaotic map; An initial sequence adjustment unit, configured to adjust the order of elements in the initial frequency hopping sequence using a cyclic shift method, and to change the value of each element in the initial frequency hopping sequence using an exclusive-OR operation; The frequency hopping sequence encryption unit is used to encrypt the adjusted frequency hopping sequence.

4. The system for reducing communication congestion in a wireless communication network according to claim 1, wherein: The link layer also includes: The channel status detection module is used to detect the current wireless channel status based on the latest available frequency band information; The coding mode screening module is used to screen the current channel coding mode from the mapping relationship table according to the current wireless channel status and wireless communication service type; the mapping relationship table contains the channel coding mode corresponding to each combination mode; the combination mode refers to the combination mode between the wireless communication service and the wireless communication status; A wireless channel coding module, configured to perform wireless channel coding according to a current channel coding method; The relay node screening module is used to select one or more relay nodes from each user device using a distributed algorithm according to the current status of each user device when the target user device cannot receive the signal data sent by the source user device.

5. The system for reducing communication congestion in a wireless communication network according to claim 4, wherein: The link layer also includes: a cache space allocation module, which is used to allocate cache space to each relay node.

6. The system for reducing communication congestion in a wireless communication network according to claim 1, wherein: The network layer also includes: The network topology generation module is used to generate a topology structure corresponding to the wireless communication network and assign values ​​to the edges of the topology structure using the latest interference frequency band information; The transmission path generation module is used to input the topology structure into the graph convolutional neural network and output the current data transmission path; The node impedance detection module is used to detect the impedance of each node on the current data transmission path.

Images