RNTI adaptive detection method and system in NR network environment

Abstract

The invention relates to the technical field of communication, and discloses an RNTI adaptive detection method and system in an NR network environment, and the system comprises a processor, a data collection module, a preprocessing module, a feature extraction module, an adaptive adjustment module, a detection module and a verification module. The data acquisition module acquires network data, and the preprocessing module processes the data to obtain a preprocessed data set. The feature extraction module extracts multi-dimensional feature vectors, and the adaptive adjustment module dynamically adjusts RNTI detection parameters according to the multi-dimensional feature vectors. The detection module identifies the RNTIs in combination with a blind detection algorithm and a neural network model and outputs a candidate RNTI set, and the verification module screens out an effective RNTI list and sends the effective RNTI list to a network scheduling end. According to the invention, by collecting the network data in real time and adaptively adjusting the detection parameters, the accuracy of RNTI detection and the system adaptability are improved, the network resource scheduling is optimized, the detection overhead is reduced, and the NR network performance and the user experience are improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and particularly to a method and system for adaptively detecting RNTI in an NR network environment. Background Art

[0002] With the wide application of 5G technologies, the NR network, with its characteristics of high speed, low latency, and large capacity, has brought a brand-new communication experience to users. However, in the actual NR network environment, there are many complex factors, which pose great challenges to the detection of RNTI.

[0003] The channel state of the NR network is complex and changeable. The channel is affected by various factors such as multipath fading, shadowing effect, and interference. In an environment with high-rise buildings in the city, the signal is easily reflected multiple times between buildings, resulting in multipath fading and causing the channel state parameters such as the channel quality indicator (CQI) and the reference signal received power (RSRP) to fluctuate violently. This makes it difficult for traditional fixed-parameter RNTI detection methods to adapt to the dynamic changes of the channel, unable to accurately detect RNTI, and thus affecting the accuracy of network scheduling and communication quality.

[0004] The signal strengths of user equipment vary greatly. Different user equipment has different positions, transmission powers, and hardware performances, resulting in uneven received signal strengths. In areas with weak signal strengths, the signals of user equipment are easily submerged by noise. If the signal detection threshold of traditional detection methods is set improperly, there will be a situation of missing the detection of RNTI; while in areas with strong signal strengths, if the threshold is too high, it may lead to false detection, increasing the ineffective detection overhead and reducing the network detection efficiency.

[0005] The network load shows dynamic changes. During the peak period of network usage, such as at the scene of a large event or during working hours in an office area, a large number of users access the network simultaneously, and the occupancy rate of physical layer resource blocks rises sharply, increasing the network load. At this time, if fixed resource block allocation weights and detection window lengths are still used, the detection accuracy will decrease, and it will be impossible to detect the RNTI of all user equipment in a timely and accurate manner, affecting the reasonable allocation of network resources and the communication experience of users.

[0006] Most traditional RNTI detection methods adopt fixed detection parameters and lack the ability to adapt to changes in the network environment. These methods cannot dynamically adjust the detection strategy according to real-time channel state, signal strength, and network load and other parameters, and it is difficult to meet the requirements of the NR network for efficient and accurate detection of RNTI, restricting the further improvement of network performance. Summary of the Invention

[0007] The purpose of the present invention is to provide a method and system for adaptively detecting RNTI in an NR network environment to solve the problems raised in the above background art.

[0008] To achieve the above object, the present invention provides the following technical solution: An RNTI adaptive detection system in an NR network environment, the system includes a processor, a data acquisition module, a preprocessing module, a feature extraction module, an adaptive adjustment module, a detection module, and a verification module;

[0009] The data acquisition module collects channel state parameters, user equipment signal strength, and network load data in the NR network environment in real time, and sends the collected data to the preprocessing module;

[0010] The preprocessing module performs noise reduction processing on the channel state parameters, normalization processing on the user equipment signal strength, and time series alignment processing on the network load data to obtain a preprocessed data set;

[0011] The feature extraction module extracts multi-dimensional feature vectors from the preprocessed data set, including time domain features, frequency domain features, and spatial domain features, and inputs the multi-dimensional feature vectors into the adaptive adjustment module;

[0012] The adaptive adjustment module dynamically adjusts RNTI detection parameters based on the multi-dimensional feature vectors, including the detection window length, signal detection threshold, and resource block allocation weight, and generates an adaptive parameter configuration instruction;

[0013] The detection module executes a blind detection algorithm according to the adaptive parameter configuration instruction, combines a pre-trained neural network model to identify the RNTI of the user equipment, and outputs a candidate RNTI set;

[0014] The verification module performs redundancy check and conflict detection on the candidate RNTI set, filters out a valid RNTI list, and sends the valid RNTI list to the network scheduling end through the processor.

[0015] Preferably, the specific operation process of the data acquisition module is as follows:

[0016] Collect the channel quality indicator CQI, reference signal received power RSRP, and physical layer resource block occupancy rate of the NR network;

[0017] Match the CQI with a preset standard channel quality mapping table to obtain a channel state grading value;

[0018] Calculate the difference between the RSRP and the dynamic signal strength baseline to obtain a signal strength offset value;

[0019] Compare the resource block occupancy rate with historical data of the same period to obtain a network load fluctuation value;

[0020] If the channel state grading value, signal strength offset value, or network load fluctuation value exceeds the corresponding preset threshold, generate a network environment anomaly mark.

[0021] Preferably, the preprocessing module includes:

[0022] Perform high-frequency noise filtering on the channel state parameters by using wavelet transform to generate denoised channel parameters;

[0023] Perform dynamic range compression on the signal strength of the user equipment through the range normalization algorithm to obtain a normalized signal strength sequence;

[0024] Use the sliding window mechanism to segment the time series of the network load data and synchronize it with the system clock to form time-series aligned load data.

[0025] Preferably, the specific implementation of the feature extraction module includes:

[0026] Calculate the signal periodicity of the time domain features by using the autocorrelation function to generate time domain period coefficients;

[0027] Extract the main frequency component energy ratio of the frequency domain features by using the fast Fourier transform to generate a frequency domain energy distribution vector;

[0028] Analyze the spatial correlation of multi-antenna signals by using the covariance matrix for the spatial domain features to generate a spatial correlation matrix.

[0029] Preferably, the operation process of the adaptive adjustment module includes:

[0030] Input the multi-dimensional feature vector into the reinforcement learning model and calculate the expected utility values of different detection parameter combinations through the Q-learning algorithm;

[0031] Select the optimal detection window length, signal detection threshold, and resource block allocation weight according to the principle of maximizing the expected utility value;

[0032] When the network load fluctuation value exceeds the preset threshold, enable the standby parameter configuration strategy to dynamically expand the detection window length and reduce the signal detection threshold.

[0033] Preferably, the process of the detection module executing the blind detection algorithm is as follows:

[0034] Preliminarily screen potential RNTI frequency points by using the energy detection method in the frequency domain;

[0035] Perform symbol-level alignment on the candidate frequency points by using a matched filter in the time domain to generate a time domain synchronization signal;

[0036] Input the time domain synchronization signal into the convolutional neural network CNN model to identify the RNTI coding pattern and output a candidate RNTI set.

[0037] Preferably, the verification module includes:

[0038] Perform cyclic redundancy check on each RNTI in the candidate RNTI set, and eliminate the items with failed checks;

[0039] Record the allocated RNTIs through a hash table, detect the conflict situation between the candidate RNTIs and the allocated RNTIs, and remove the conflicting items;

[0040] Sort the remaining RNTIs in descending order of signal strength, and select the top N as the valid RNTI list.

[0041] Preferably, it further includes a historical data analysis module, and its operation process is as follows:

[0042] Store the historical valid RNTI list and the corresponding network environment parameters;

[0043] Analyze the RNTI distribution pattern in the historical data through a clustering algorithm to generate a typical scenario template;

[0044] When the similarity between the current network environment parameters and any typical scenario template exceeds a preset threshold, directly call the detection parameter configuration corresponding to the template.

[0045] Preferably, it further includes a dynamic threshold setting module, and its operation process is as follows:

[0046] Adjust the sensitivity coefficient of the signal detection threshold according to the real-time network load fluctuation value, and the calculation formula is:

[0047]

[0048] where, L current is the current signal strength offset value, L base is the baseline signal strength, V load is the network load fluctuation value, V max is the maximum allowable fluctuation value, and α and β are weighting factors;

[0049] Dynamically update the signal detection threshold according to the sensitivity coefficient K.

[0050] Preferably, the present invention further includes an RNTI adaptive detection method in an NR network environment, including the following steps:

[0051] Step 1: Real-time collect the channel state parameters, user equipment signal strength, and network load data in the NR network environment through a data collection module, and send the collected data to a preprocessing module;

[0052] Step 2: Use the preprocessing module to perform noise reduction processing on the channel state parameters, normalization processing on the user equipment signal strength, and time series alignment processing on the network load data to obtain a preprocessed data set;

[0053] Step 3: Extract multi-dimensional feature vectors from the preprocessed dataset with the help of the feature extraction module. The multi-dimensional feature vectors include time-domain features, frequency-domain features, and spatial-domain features, and input the multi-dimensional feature vectors into the adaptive adjustment module;

[0054] Step 4: Dynamically adjust the RNTI detection parameters based on the multi-dimensional feature vectors by the adaptive adjustment module. The detection parameters include the detection window length, signal detection threshold, and resource block allocation weight, and generate an adaptive parameter configuration instruction;

[0055] Step 5: According to the adaptive parameter configuration instruction, use the detection module to execute the blind detection algorithm, and combine with the pre-trained neural network model to identify the RNTI of the user equipment, and output a candidate RNTI set;

[0056] Step 6: Use the verification module to perform redundancy check and conflict detection on the candidate RNTI set, screen out a list of valid RNTIs, and send the list of valid RNTIs to the network scheduling end through the processor.

[0057] Compared with the prior art, the beneficial effects of the present invention are:

[0058] The present invention can effectively remove noise and interference in the data and improve the data quality by the data acquisition module to collect channel state parameters, user equipment signal strength, and network load data in real time, and use the preprocessing module to perform noise reduction, normalization, and time series alignment processing on the data, providing an accurate data basis for subsequent feature extraction and detection. The feature extraction module extracts multi-dimensional feature vectors from the preprocessed dataset, including time-domain, frequency-domain, and spatial-domain features, comprehensively reflecting the characteristics of the signal. The adaptive adjustment module dynamically adjusts the RNTI detection parameters, such as the detection window length, signal detection threshold, and resource block allocation weight, based on these feature vectors, enabling the detection parameters to adapt to different network environments. When the channel fading is severe, the adaptive adjustment module can appropriately increase the detection window length to improve the signal capture ability; when the signal strength is weak, reduce the signal detection threshold to avoid missing the RNTI detection, thus significantly improving the accuracy of RNTI detection.

[0059] The adaptive adjustment module adopts a reinforcement learning model and Q-learning algorithm to select the optimal parameters according to the expected utility values of different combinations of detection parameters. When the network load fluctuation value exceeds the preset threshold, a backup parameter configuration strategy is enabled to dynamically expand the detection window length and reduce the signal detection threshold, ensuring stable detection of RNTI under changing network loads. The historical data analysis module stores the list of historical valid RNTIs and network environment parameters, and analyzes the RNTI distribution pattern through a clustering algorithm to generate a typical scenario template. When the current network environment matches the typical scenario, the corresponding detection parameter configuration is directly called, greatly improving the system's adaptation speed to different network scenarios and reducing the time overhead of parameter adjustment.

[0060] Accurately detecting RNTI is the basis for reasonable network resource scheduling. The verification module of the present invention performs redundancy verification and conflict detection on the candidate RNTI set, screens out the list of valid RNTIs and sends it to the network scheduling end, ensuring the accuracy of the RNTI information obtained by the network scheduling end, avoiding resource allocation errors caused by incorrect RNTIs, and improving the utilization efficiency of network resources. The adaptive adjustment module dynamically adjusts the resource block allocation weights, enabling network resources to be reasonably allocated according to the actual needs of user equipment. When the signal strength of the user equipment is good and the channel quality is excellent, more resource blocks are allocated to improve the communication rate of users; when the signal is weak or the network load is high, the resource allocation is reasonably adjusted to ensure the overall performance of the network, optimizing network resource scheduling and enhancing the communication experience of users.

[0061] The dynamic threshold setting module adjusts the sensitivity coefficient of the signal detection threshold according to the real-time network load fluctuation value, and dynamically updates the signal detection threshold. This avoids over-detection or missed detection problems caused by a fixed threshold in different network environments, reduces invalid detection operations, lowers the detection overhead, and improves the detection efficiency. The detection module uses a blind detection algorithm that combines energy detection and a matched filter. During the preliminary screening and symbol-level alignment processes, it can quickly exclude invalid signals, reducing the amount of data processed by the subsequent convolutional neural network (CNN) model, further reducing the computational complexity and time overhead of detection. Description of the Drawings

[0062] Figure 1 is the working principle diagram of the system of the present invention;

[0063] Figure 2 is the working flowchart of the data acquisition module;

[0064] Figure 3 is the working flowchart of the adaptive adjustment module. Detailed Embodiments

[0065] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with 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. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0066] Please refer to Figures 1-3 , the present invention provides a technical solution: an RNTI adaptive detection system in an NR network environment, and the system includes:

[0067] The system mainly consists of a processor, a data acquisition module, a preprocessing module, a feature extraction module, an adaptive adjustment module, a detection module, and a verification module. The data acquisition module collects channel state parameters, user equipment signal strength, and network load data in the NR network environment in real time, and transmits these data to the preprocessing module. The preprocessing module performs noise reduction processing on the channel state parameters, normalization processing on the user equipment signal strength, and time series alignment processing on the network load data to obtain a preprocessed data set. The feature extraction module extracts a multi-dimensional feature vector including time domain features, frequency domain features, and spatial domain features from the preprocessed data set and inputs it into the adaptive adjustment module. The adaptive adjustment module dynamically adjusts the RNTI detection parameters according to the multi-dimensional feature vector, including the detection window length, signal detection threshold, and resource block allocation weight, and then generates an adaptive parameter configuration instruction. The detection module executes a blind detection algorithm according to this instruction, combines a pre-trained neural network model to identify the RNTI of the user equipment, and outputs a candidate RNTI set. The verification module performs redundancy check and conflict detection on the candidate RNTI set, filters out a valid RNTI list, and finally sends the valid RNTI list to the network scheduling end through the processor.

[0068] The following further illustrates the present invention in conjunction with Embodiments 1 to 5:

[0069] Embodiment 1:

[0070] When the data acquisition module is running, it collects the channel quality indicator CQI, reference signal received power RSRP, and physical layer resource block occupancy rate of the NR network. After collecting the CQI, it matches the CQI with a preset standard channel quality mapping table to obtain a channel state grading value. In this process, the preset standard channel quality mapping table is established in advance, which maps different ranges of CQI values to corresponding channel state gradings, such as mapping CQI values within a certain interval to different gradings such as "good", "average", "poor", etc., to facilitate subsequent evaluation of the channel state.

[0071] For the RSRP, the data acquisition module calculates the difference between it and the dynamic signal strength baseline to obtain the signal strength offset value. The dynamic signal strength baseline is not a fixed value. It will be dynamically adjusted according to the historical data of the network and the real-time changes. For example, the system will adjust the baseline according to the average RSRP value over a period of time, combined with factors such as the current network load and user distribution, to more accurately reflect the changes in signal strength.

[0072] When processing the physical layer resource block occupancy rate, the data acquisition module compares it with the historical data of the same period to obtain the network load fluctuation value. The historical data of the same period is the resource block occupancy rate data collected and stored by the system in the same past time period. By comparison, the fluctuation degree of the current network load compared with the historical situation can be intuitively seen.

[0073] If the channel state grading value, signal strength offset value, or network load fluctuation value exceeds the corresponding preset threshold, the data acquisition module will generate a network environment anomaly mark. These preset thresholds are determined through a large amount of experimental data and actual network operation experience. For example, after analyzing a large amount of network environment data, it is determined that when the channel state grading value reaches a certain specific level, or the signal strength offset value exceeds a certain numerical range, or the network load fluctuation value exceeds a certain proportion, it is considered that the network environment is abnormal. At this time, the generated network environment anomaly mark will be an important basis for subsequent processing, reminding the system that special detection parameter adjustments or other countermeasures may be required.

[0074] Embodiment 2:

[0075] The preprocessing module includes different processing methods for channel state parameters, user equipment signal strength, and network load data. For channel state parameters, wavelet transform is used to filter out high-frequency noise and generate denoised channel parameters. Wavelet transform is a time-frequency analysis method that can effectively separate different frequency components in the signal. In the NR network environment, channel state parameters may be interfered by various high-frequency noises, which will affect the subsequent accurate judgment of the channel state. Through wavelet transform, the high-frequency noise can be extracted and removed from the channel state parameters, and the useful signal components can be retained, so as to obtain more accurate denoised channel parameters. For example, in practical applications, using the multi-resolution analysis feature of wavelet transform, the channel state parameters are decomposed into different frequency sub-bands, then the noise in the high-frequency sub-bands is processed by thresholding, and finally the denoised channel parameters are reconstructed through inverse wavelet transform.

[0076] For the user equipment signal strength, the preprocessing module performs dynamic range compression through the range normalization algorithm to obtain the normalized signal strength sequence. The formula of the range normalization algorithm is: where X is the original user equipment signal strength value, X minand X max are the minimum and maximum values respectively in the collected signal strength data. Through this algorithm, the signal strengths of different user devices are uniformly mapped to the interval [0, 1], eliminating the dimensional differences in signal strengths of different devices caused by factors such as hardware differences, making the subsequent analysis and comparison of signal strengths more reasonable and accurate.

[0077] For the network load data, the preprocessing module uses a sliding window mechanism to segment the time series and synchronize it with the system clock to form time-series aligned load data. The size of the sliding window is set according to the characteristics and requirements of the actual network data. For example, a sliding window with a length of n is set, and the window slides sequentially on the time series, sliding one time unit each time, and the data within the window is processed as a segment. Then, by synchronizing with the system clock, the load data in different time periods is aligned in time to ensure the consistency of the time dimension in subsequent analysis, facilitating feature extraction and analysis based on the time series.

[0078] Embodiment 3:

[0079] In the implementation process of the feature extraction module, for time-domain features, the autocorrelation function is used to calculate the signal periodicity and generate the time-domain periodicity coefficient. The calculation formula of the autocorrelation function is: where x(t) is the representation of the signal in the time domain, T is the observation time length of the signal, and τ is the time delay. By calculating the autocorrelation function, the similarity degree of the signal at different time points can be measured, thereby obtaining the periodicity characteristics of the signal. For example, in the NR network, the signal of the user equipment may have a certain periodicity. By calculating the autocorrelation function, this periodicity characteristic can be accurately extracted to generate the time-domain periodicity coefficient, providing a basis for subsequent detection parameter adjustment.

[0080] For frequency-domain features, the feature extraction module uses the fast Fourier transform to extract the main frequency component energy ratio and generate the frequency-domain energy distribution vector. The fast Fourier transform (FFT) is an efficient algorithm for calculating the discrete Fourier transform (DFT), which can quickly transform the time-domain signal to the frequency domain. In the frequency domain, the proportion of the energy of the main frequency component in the total energy is calculated to obtain the frequency-domain energy distribution vector. This vector reflects the energy distribution of the signal at different frequencies and is of great significance for analyzing the frequency characteristics and interference conditions of the signal.

[0081] In terms of spatial domain features, the feature extraction module uses covariance matrix analysis to analyze the spatial correlation of multi-antenna signals and generate a spatial correlation matrix. The covariance matrix can measure the degree of linear correlation between multiple variables. In a multi-antenna system, there is a certain correlation between the signals received by different antennas. By calculating the covariance matrix, the spatial correlation information between these signals can be obtained, and a spatial correlation matrix is generated. This matrix plays an important role in optimizing detection parameters such as resource block allocation weights.

[0082] After receiving the multi-dimensional feature vector, the adaptive adjustment module inputs it into the reinforcement learning model and calculates the expected utility values of different combinations of detection parameters through the Q-learning algorithm. The Q-learning algorithm is a model-free reinforcement learning algorithm that finds the optimal strategy through continuous trial and learning. In the present invention, the expected utility values of different combinations of detection window lengths, signal detection thresholds, and resource block allocation weights are calculated through the Q-learning algorithm. According to the principle of maximizing the expected utility value, the optimal combination of detection parameters is selected. When the network load fluctuation value exceeds the preset threshold, the standby parameter configuration strategy is enabled to dynamically expand the detection window length and reduce the signal detection threshold. This is because when the network load is high, the possibility of signal interference and conflict increases. Expanding the detection window length can increase the detection time range and improve the probability of detecting the RNTI; reducing the signal detection threshold can make the detection more sensitive and avoid missing some weak but effective signals.

[0083] Embodiment 4:

[0084] This embodiment details the working processes of the detection module and the verification module. Its function is to accurately identify the RNTI of the user equipment and screen out a list of valid RNTIs through an efficient detection algorithm and a strict verification mechanism, ensuring the accuracy and reliability of network scheduling.

[0085] The process of the detection module executing the blind detection algorithm is as follows: In the frequency domain, the energy detection method is used to preliminarily screen potential RNTI frequency points. The energy detection method is a detection method based on the energy characteristics of signals. In the frequency domain of the NR network, different RNTI signals have different energy distributions at specific frequency points. By detecting the energy of the frequency-domain signal and setting an energy threshold, when the signal energy at a certain frequency point exceeds the threshold, it is preliminarily screened as a potential RNTI frequency point. This process can quickly screen out the frequency points where RNTI signals may exist from a large number of frequency points, reducing the workload of subsequent processing. In the time domain, a matched filter is used to perform symbol-level alignment on the candidate frequency points to generate a time-domain synchronization signal. The matched filter is a filter used at the receiving end to improve the signal-to-noise ratio, and its characteristics match those of the transmitted signal. For the candidate frequency points preliminarily screened, through the processing of the matched filter, the received signal can be synchronized with the symbols of the transmitted signal in the time domain to generate a time-domain synchronization signal. This step is crucial for accurately identifying RNTI because only when the signal is synchronized can the subsequent neural network model correctly identify the RNTI coding pattern.

[0086] The time-domain synchronization signal is input into the convolutional neural network (CNN) model to identify the RNTI coding pattern and output a candidate RNTI set. The convolutional neural network (CNN) has powerful feature extraction and pattern recognition capabilities. In the present invention, the CNN model is pre-trained with a large amount of RNTI sample data to learn the characteristics of different RNTI coding patterns. When the time-domain synchronization signal is input into the CNN model, the model can identify the RNTI coding pattern in the signal according to the characteristics it has learned, thereby outputting a candidate RNTI set.

[0087] The verification module processes the candidate RNTI set. First, it performs cyclic redundancy check (CRC) on each RNTI and eliminates the items that fail the check. Cyclic redundancy check (CRC) is a commonly used error detection method. It generates a check code by performing a specific polynomial operation on the data. When transmitting the RNTI, its CRC check code is sent simultaneously. At the receiving end, the CRC check code of the received RNTI is recalculated and compared with the received check code. If the two are inconsistent, it indicates that the RNTI may have an error during transmission, and it is eliminated.

[0088] Next, the verification module records the assigned RNTIs through a hash table, detects the conflict situation between the candidate RNTIs and the assigned RNTIs, and removes the conflicting items. The hash table is an efficient data structure used for quickly searching and inserting data. In the present invention, the hash table is used to record the RNTIs that have been assigned in the network. For each RNTI in the candidate RNTI set, it is searched in the hash table. If it is found that the RNTI has already been assigned, it indicates a conflict, and it is removed from the candidate set.

[0089] Finally, sort the remaining RNTIs in descending order of signal strength, and select the top N as the valid RNTI list. The signal strength to a certain extent reflects the communication quality between the user equipment corresponding to the RNTI and the base station. Sorting the remaining RNTIs in descending order of signal strength and selecting the top N as the valid RNTI list can preferentially select user equipment with better communication quality, improving the efficiency and quality of network scheduling. The value of N is set according to the actual requirements and performance requirements of the network. For example, when the network load is low, the value of N can be appropriately increased to access more user equipment; when the network load is high, the value of N can be decreased to ensure the communication quality of critical user equipment.

[0090] Example 5:

[0091] When the historical data analysis module is running, it stores the historical valid RNTI list and the corresponding network environment parameters. These historical data are accumulated during the long-term operation of the system, including the valid RNTI list at different time points and network environment parameters such as channel state parameters, user equipment signal strength, and network load data at that time. By storing these historical data, it provides rich materials for subsequent analysis.

[0092] Then, the historical data analysis module analyzes the RNTI distribution pattern in the historical data through a clustering algorithm to generate a typical scenario template. The clustering algorithm is an unsupervised learning algorithm that can divide data into different categories according to similarity. In the present invention, the clustering algorithm is used to analyze the RNTI distribution in the historical data, and network environment data with similar distribution patterns are grouped into one category to form a typical scenario template. For example, according to factors such as different network load conditions and channel quality, the historical data are divided into different typical scenario templates such as "high load - good channel" and "low load - poor channel".

[0093] When the similarity between the current network environment parameters and any typical scenario template exceeds a preset threshold, the historical data analysis module directly calls the detection parameter configuration corresponding to that template. The calculation of similarity can adopt various methods, such as Euclidean distance, cosine similarity, etc. The preset threshold is determined through experiments and experience. When the similarity between the current network environment parameters and a certain typical scenario template exceeds this threshold, it indicates that the current network environment is similar to this typical scenario. At this time, the detection parameter configuration corresponding to this template can be directly called without re-performing complex parameter adjustment, improving the response speed and efficiency of the detection system.

[0094] The dynamic threshold setting module adjusts the sensitivity coefficient of the signal detection threshold according to the real-time network load fluctuation value, and the calculation formula is: where, L current is the current signal strength offset value, Lbase is the baseline signal strength, V load is the network load fluctuation value, V max is the maximum allowable fluctuation value, and α and β are weighting factors. The values of the weighting factors α and β are adjusted according to the actual situation and performance requirements of the network. For example, in a scenario where the change in signal strength has a greater impact on the detection result, the value of α can be appropriately increased; in a scenario where the network load fluctuation has a greater impact on the detection result, the value of β can be appropriately increased.

[0095] The dynamic threshold setting module dynamically updates the signal detection threshold according to the sensitivity coefficient K. When the network load fluctuates greatly, the sensitivity coefficient K calculated by the above formula will change accordingly, and the signal detection threshold is adjusted according to the K value. For example, when the network load fluctuation value V load increases, the K value may increase. At this time, the signal detection threshold is correspondingly reduced, making the detection system more sensitive and capable of detecting weaker signals; conversely, when the network load fluctuation value is small, the signal detection threshold is appropriately increased to reduce the probability of false detection.

[0096] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0097] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An RNTI adaptive detection system in a NR network environment, characterized in that: It includes a processor, a data acquisition module, a preprocessing module, a feature extraction module, an adaptive adjustment module, a detection module and a verification module; The data acquisition module collects channel state parameters, user equipment signal strength and network load data in the NR network environment in real time, and sends the collected data to the preprocessing module; The preprocessing module performs noise reduction processing on the channel state parameters, normalizes the signal strength of the user equipment, and performs time alignment processing on the network load data to obtain a preprocessing data set; The feature extraction module extracts a multidimensional feature vector from the preprocessed data set, including time domain features, frequency domain features and space domain features, and inputs the multidimensional feature vector into the adaptive adjustment module; The adaptive adjustment module dynamically adjusts RNTI detection parameters based on the multidimensional feature vector, including detection window length, signal detection threshold and resource block allocation weight, and generates adaptive parameter configuration instructions; The detection module executes a blind detection algorithm according to the adaptive parameter configuration instruction, identifies the RNTI of the user equipment in combination with the pre-trained neural network model, and outputs a candidate RNTI set; The verification module performs redundancy check and conflict detection on the candidate RNTI set, selects a valid RNTI list, and sends the valid RNTI list to the network scheduling end via the processor.

2. The RNTI adaptive detection system in an NR network environment according to claim 1, characterized in that: The specific operation process of the data acquisition module is as follows: Collect the channel quality indicator CQI, reference signal received power RSRP and physical layer resource block occupancy of the NR network; Matching the CQI with a preset standard channel quality mapping table to obtain a channel state classification value; The difference between RSRP and the dynamic signal strength baseline is calculated to obtain a signal strength offset value; Compare the resource block occupancy rate with historical data for the same period to obtain the network load fluctuation value; If the channel status classification value, signal strength offset value or network load fluctuation value exceeds the corresponding preset threshold, a network environment abnormality mark is generated.

3. The RNTI adaptive detection system in an NR network environment according to claim 1, characterized in that: The preprocessing module comprises: Wavelet transform is used to filter high-frequency noise from channel state parameters to generate noise-reduced channel parameters; The dynamic range of the user equipment signal strength is compressed by a range normalization algorithm to obtain a normalized signal strength sequence; The sliding window mechanism is used to segment the network load data into time series and align them with the system clock to form time-aligned load data.

4. The RNTI adaptive detection system in an NR network environment according to claim 1, characterized in that: The specific implementation of the feature extraction module includes: The autocorrelation function is used to calculate the signal periodicity of the time domain characteristics to generate the time domain periodicity coefficient; Fast Fourier transform is used to extract the energy ratio of the main frequency component from the frequency domain features to generate the frequency domain energy distribution vector; The covariance matrix is ​​used to analyze the spatial correlation of multi-antenna signals for spatial domain features and generate a spatial correlation matrix.

5. The RNTI adaptive detection system in an NR network environment according to claim 1, characterized in that: The operation process of the adaptive adjustment module includes: The multi-dimensional feature vector is input into the reinforcement learning model, and the expected utility value of different detection parameter combinations is calculated through the Q-learning algorithm; According to the principle of maximizing expected utility value, the optimal detection window length, signal detection threshold and resource block allocation weight are selected; When the network load fluctuation value exceeds the preset threshold, the backup parameter configuration strategy is enabled to dynamically expand the detection window length and lower the signal detection threshold.

6. The RNTI adaptive detection system in an NR network environment according to claim 1, characterized in that: The process of the detection module executing the blind detection algorithm is as follows: In the frequency domain, energy detection method is used to preliminarily screen potential RNTI frequencies; In the time domain, a matched filter is used to perform symbol-level alignment on candidate frequency points to generate a time domain synchronization signal; The time domain synchronization signal is input into the convolutional neural network (CNN) model, the RNTI encoding mode is identified, and a candidate RNTI set is output.

7. The RNTI adaptive detection system in a NR network environment according to claim 1, characterized in that: The verification module comprises: Perform a cyclic redundancy check on each RNTI in the candidate RNTI set and remove items that fail the check; The allocated RNTI is recorded in a hash table, conflicts between the candidate RNTI and the allocated RNTI are detected, and the conflicting items are removed; Arrange the remaining RNTIs in descending order of signal strength, and select the first N as the valid RNTI list.

8. The RNTI adaptive detection system in a NR network environment according to claim 1, characterized in that: It also includes a historical data analysis module, which operates as follows: Store the historical valid RNTI list and the corresponding network environment parameters; Analyze the RNTI distribution pattern in historical data through clustering algorithm and generate typical scenario templates; When the similarity between the current network environment parameters and any typical scenario template exceeds a preset threshold, the detection parameter configuration corresponding to the template is directly called.

9. The RNTI adaptive detection system in an NR network environment according to claim 1, characterized in that: It also includes a dynamic threshold setting module, which operates as follows: The sensitivity coefficient of the signal detection threshold is adjusted according to the real-time network load fluctuation value. The calculation formula is: Among them, L current is the current signal strength offset value, L base is the baseline signal intensity, V load is the network load fluctuation value, V max is the maximum allowable fluctuation value, α and β are weight factors; The signal detection threshold is dynamically updated according to the sensitivity coefficient K.

10. A RNTI adaptive detection method in a NR network environment, characterized in that: The following steps are involved: Step 1: The data acquisition module collects channel state parameters, user equipment signal strength and network load data in the NR network environment in real time, and sends the collected data to the preprocessing module; Step 2: Use the preprocessing module to perform noise reduction processing on the channel state parameters, normalize the user equipment signal strength, and perform time alignment processing on the network load data to obtain a preprocessed data set; Step 3: extracting a multidimensional feature vector from the preprocessed data set with the help of a feature extraction module, wherein the multidimensional feature vector includes time domain features, frequency domain features and space domain features, and inputting the multidimensional feature vector into an adaptive adjustment module; Step 4: The adaptive adjustment module dynamically adjusts the RNTI detection parameters based on the multi-dimensional feature vector, wherein the detection parameters include the detection window length, the signal detection threshold and the resource block allocation weight, and generates an adaptive parameter configuration instruction; Step 5: According to the adaptive parameter configuration instruction, the detection module executes the blind detection algorithm, combines the pre-trained neural network model to identify the RNTI of the user equipment, and outputs the candidate RNTI set; Step 6: Use the verification module to perform redundancy check and conflict detection on the candidate RNTI set, filter out the valid RNTI list, and send the valid RNTI list to the network scheduling end through the processor.

Images