Monitoring Spectral Anomalies in Cellular Wireless Networks and Training Spectral Anomaly Neural Networks
By using RF receivers and processing units to train spectrum anomaly neural networks in 5G dedicated cellular wireless networks, the spectrum anomaly recognition problem in the Industrial 4.0 environment is solved, and the spectrum anomaly recognition is quickly recognized and positioned, ensuring error-free communication and reducing cost.
Patent Information
- Application Number
- CN202011040173.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2019-10-31
- Filing Date
- 2020-09-28
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2040-09-28
AI Technical Summary
In the Industrial 4.0 environment, possible interference or spectrum anomalies in 5G dedicated cellular wireless networks are difficult to quickly identify, resulting in potential huge damage and high costs, especially in restricted areas of company factories.
A monitoring system and training system are designed to monitor the spectrum and derive the spectrum and physical measurements using RF receivers. Spectral anomaly neural network is trained through machine learning algorithms, detect and locate spectrum anomalies, including processing units in cloud services for data processing. and output exception information.
It realizes the rapid identification and location of spectrum anomalies in 5G dedicated networks, ensures error-free communication, reduces the damage and costs caused by interference, and improves the reliability and flexibility of the production environment.
Smart Images

Figure CN112752288B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a monitoring system and a monitoring method for detecting spectrum anomalies in a cellular wireless network, especially a 5G dedicated ultra-reliable and low-latency communication (ultra-reliable and low-latency communication, uRLLC) network, wherein the monitoring system and the monitoring method include a radio frequency (RF) receiver and a processing unit. Further, the present invention also relates to a training system and a training method for training a spectrum anomaly neural network, wherein the training system and the training method are used in a cellular wireless network, especially a 5G dedicated uRLLC network, and include an RF receiver and a processor. Background Art
[0002] Currently, a new cellular wireless network technology is being developed using 5G NR, which allows ultra-reliable and low-latency communications. These conditions are essential for automation in Industry 4.0 environments.
[0003] In addition to the overall development of 5G NR, there is also interest from different parties (e.g., some companies) in developing 5G private cellular / mobile wireless networks that can be used in constrained / restricted areas (e.g., a company's factory premises), where the 5G private network is logically and / or physically separated from the 5G public network.
[0004] In more detail, in Germany and some other countries it is / will be possible to acquire dedicated spectrum (e.g. 3.7GHz-3.8GHz) within restricted areas of a factory building to install and operate your own 5G private network.
[0005] However, the installation and operation of 5G private networks are associated with regulatory requirements such as maximum transmitter output power in compliance with property boundaries, advance planning and technical characterization of base stations and their antenna technologies, and capacity planning / calculation (spectrum).
[0006] However, own 5G private networks operating with dedicated spectrum offer the possibility of highly reliable / ultra-reliable data communications with low latency, which can be used to control machines and modular production cells, but also to control autonomously moving robots and / or vehicles. This allows highly flexible and modular production lines.
[0007] In general in the broad field of wireless technology and also in 5G networks, interference / interruptions / disturbances may occur, for example between base stations and terminal devices or any kind of general faults on the air path. In 5G public networks, there are different application areas that can tolerate a certain interference sensitivity.
[0008] However, when using 5G private networks in the application field of production environments in restricted areas of company plants, such interference could potentially cause huge damage and therefore lead to high costs, which must be avoided or at least identified very quickly in order to take appropriate countermeasures.
[0009] In the past and today, wireless communication in Industry 4.0 environments has been used essentially only in non-security-related situations and mostly via WIFI / WLAN. To date, mobile radio technologies or cellular wireless networks such as UMTS (Universal Mobile Telecommunications System) and LTE (Long Term Evolution) have not been used, as they generally cannot be operated privately, as there is no dedicated spectrum available for UMTS or LTE (so far, only a few examples are known in which public operators have implemented private networks for Industry 4.0 customers).
[0010] Therefore, the basic goal of installing and operating a 5G private network within a restricted area of a company's premises is to avoid interference and, when it cannot be completely avoided, to identify interference as quickly as possible so that appropriate countermeasures can be taken. Summary of the invention
[0011] Therefore, the purpose of the present invention is to very quickly identify any interference / spectral anomalies occurring in cellular wireless networks, especially in 5G dedicated networks.
[0012] This object is achieved by the features of the independent claim. The dependent claims further develop the central idea of the invention.
[0013] The present invention relates to a monitoring system and a monitoring method for detecting spectrum anomalies in a cellular wireless network, especially a 5G dedicated uRLLC network, wherein an RF receiver of the monitoring system monitors the spectrum of the cellular wireless network and derives the spectrum of the cellular wireless network and / or physical measurement values of the spectrum, and a processing unit of the monitoring system executes a spectrum anomaly neural network trained in a training system by a machine learning algorithm, wherein the processing unit obtains the spectrum and / or the physical measurement values of the spectrum and processes them to detect spectrum anomaly information.
[0014] Advantageously, the monitoring system further comprises a memory for storing the spectrum and / or physical measurements of the spectrum derived by the RF receiver, and the processing unit obtains the spectrum and / or physical measurements of the spectrum from the memory.
[0015] In a preferred embodiment, the RF receiver monitors and derives a spectrum varying with time and frequency to produce a spectrogram, wherein the processing unit obtains and processes the spectrogram. Alternatively or additionally, the RF receiver derives a power value varying with frequency per time unit as a physical measurement of the spectrum, wherein the processing unit obtains and processes the power value varying with frequency per time unit.
[0016] In addition, the monitoring system can also derive additional parameters of active communication links in the cellular wireless network, especially the achieved data rate / bit rate, signal-to-noise ratio and / or delay time, and the processing unit further obtains and processes the additional parameters to detect spectrum anomaly information.
[0017] Advantageously, the processing unit is arranged remotely from the RF receiver and remotely from the cellular wireless network, in particular in a cloud service.
[0018] The monitoring system may further comprise an output unit, in particular a display or a monitor, for presenting the detected spectrum anomaly information.
[0019] In a preferred embodiment, the monitoring system comprises at least two RF receivers and the system triangulates the location of spectral anomalies detected by the processing unit, wherein the triangulation is based on correlation of spectral and / or physical measurements of the at least two RF receivers.
[0020] Furthermore, it is possible that the derived spectrum of the cellular radio network and / or physical measurements of the spectrum are used for further training of the spectrum anomaly neural network trained by a training method.
[0021] The present invention also relates to a training system and a training method for training a spectral anomaly neural network, wherein the training system / training method is used in a cellular wireless network, especially a 5G dedicated uRLLC network, and the RF receiver of the training system monitors the spectrum of the cellular wireless network and derives the spectrum of the cellular wireless network and / or physical measurement values of the spectrum, and the processor of the training system executes a machine learning algorithm based on the derived spectrum of the cellular wireless network and / or physical measurement values of the spectrum to train the spectral anomaly neural network.
[0022] In a preferred embodiment, the training system also includes: at least two test communication devices, wherein a first test communication device of the at least two test communication devices has an active communication link to a second test communication device of the at least two test communication devices; and a signal anomaly generator for generating a signal anomaly that distorts the communication link, in particular a broadband jammer or a continuous waveform (CW) generator, wherein the spectrum anomaly neural network is also trained based on additional parameters of the generated signal anomaly and / or the active communication link, in particular the achieved data rate / bit rate, signal-to-noise ratio and / or delay time.
[0023] Advantageously, the RF receiver monitors and derives a spectrum varying with time and frequency to produce a spectrograph, wherein the spectrum anomaly neural network is trained based on the spectrograph. Alternatively or additionally, the RF receiver derives a power value per time unit varying with frequency as a physical measurement of the spectrum, wherein the spectrum anomaly neural network is trained based on the power value per time unit varying with frequency.
[0024] It is also possible that the processor is arranged remotely from the RF receiver and remotely from the cellular wireless network, in particular in a cloud service.
[0025] Through the monitoring system, monitoring method, training system and training method of the present invention, it is therefore possible to train a spectrum anomaly neural network based on error-free communication and possible interference / spectrum anomalies through a machine learning algorithm. The trained spectrum anomaly neural network can then be used to monitor the 5G private network, especially the spectrum of the network, to detect spectrum anomalies. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] These and other aspects and advantages of the present invention will become more apparent when the following detailed description is studied in conjunction with the accompanying drawings, in which:
[0027] Figure 1 The idea of the present invention is schematically shown;
[0028] Figure 2 A monitoring system according to the present invention is shown;
[0029] Figure 3 A training system according to the invention is shown. DETAILED DESCRIPTION
[0030] As mentioned above, in the technical field of 5G NR, 5G private networks can be installed and operated on restricted areas such as company plants. This is interesting for highly flexible and modular production lines, for example. However, in such a production environment, any interference / spectral anomalies in the 5G private network (especially in the air path of the 5G private network), for example, between the base station and the terminal device, must be avoided. However, it is not always possible to avoid any interference at all times. In this case, it is desirable to identify / detect such interference in the 5G private network as quickly as possible so that appropriate countermeasures can be taken.
[0031] exist Figure 1 The general idea of the invention is shown in FIG. 1 , wherein the spectral data (e.g., spectral vibration) monitored and derived are Figure 1 The spectral anomaly neural network 2 is then used by the machine learning algorithm to train the spectral anomaly neural network 2. The trained spectral anomaly neural network 2 is then used to detect spectral anomalies, wherein the monitored and derived spectral data (e.g., spectral vibration Figure 1 The detection result or the detected spectral anomaly can be presented / displayed on an output unit (such as a display or monitor 3).
[0032] In more detail, Figure 2 As shown, the present invention provides a monitoring system and a monitoring method for detecting spectrum anomalies in a 5G dedicated uRLLC network 8. For example, in a restricted area of a company's factory, such a 5G dedicated network 8 can be installed and operated. The 5G dedicated network 8 in the restricted area may include, for example, a base station 6 and a terminal device 7 communicating with the base station 6.
[0033] exist Figure 2 In the monitoring system and monitoring method shown, the RF receiver 4 monitors the spectrum of the 5G private network 8 and derives the spectrum and / or physical measurement values of the spectrum. Based on the derived spectrum and / or physical measurement values, the processing unit 5 executes the spectrum anomaly neural network that has been trained by the corresponding training system or training method to detect possible spectrum anomalies in the 5G private network 8.
[0034] This means that the radio communication in the cellular wireless network (i.e., the 5G dedicated uRLLC network 8) is captured by the corresponding RF capture device (i.e., the RF receiver 4 of the monitoring system of the present invention), and the captured radio communication data is sent to the corresponding processing device (i.e., the processing unit 5 of the monitoring system of the present invention) for analysis based on the trained spectrum anomaly neural network to detect anomalies / interference.
[0035] In order to train the spectrum anomaly neural network, the present invention also provides a training system and a training method, wherein the training system and the training method are also used in the 5G dedicated uRLLC network. Figure 3Such a training system is shown, where a similar Figure 2 , such 5G dedicated uRLLC networks8 can be installed and operated in restricted areas of company premises, for example. Figure 3 The 5G private network shown in 8 can be used with Figure 2 This means that the spectrum anomaly neural network is trained in the same 5G private network 8, and then the monitoring system and monitoring method monitor the spectrum in the 5G private network 8 and possibly detect spectrum anomalies.
[0036] In addition, it will also be possible that a spectrum anomaly neural network trained in one 5G private network at one company's plant can be used in another 5G private network at another plant of another company. For example, it will be possible that a trained spectrum anomaly neural network of one 5G private network is used as a basis for training another 5G private network. It will also be possible that a trained spectrum anomaly neural network of one 5G private network is directly used in another 5G private network for monitoring.
[0037] In the training system and training method, the RF receiver 4 monitors the spectrum of the 5G private network and derives the spectrum and / or the physical measurement of the spectrum. The derived spectrum and / or the physical measurement of the spectrum are then used to train the spectrum anomaly neural network, wherein the processor 9 executes a machine learning algorithm based on the derived spectrum of the cellular wireless network and / or the physical measurement of the spectrum to train the spectrum anomaly neural network.
[0038] With such a configuration of the training system, it is possible to learn error-free communication, wherein the detection of error-free communication is based, among other things, on the identification of signal components that are always available, for example in the case of 5G on so-called SSB blocks (synchronization and broadcast signals specified in 3GPP).
[0039] In order to learn and train interference / anomalies, the training system includes at least two test communication devices, such as a base station 6 and a terminal device 7, wherein a communication link is established between the at least two test communication devices 6 and 7 to have an active communication link. Further, the training system includes a signal anomaly generator 10 for generating a signal anomaly that distorts the communication link, wherein such a generator may be, for example, a broadband jammer or a continuous waveform (CW) generator for generating an interference signal.
[0040] Based on the already learned and trained error-free communication and the signal anomalies generated by the signal anomaly generator 10 that distort the communication link, the spectral anomalies are additionally learned, wherein in detail the spectral anomaly neural network is further trained based on the generated signal anomalies.
[0041] In the monitoring system, monitoring method, training system and training method, one possible implementation of the monitoring and derivation for the RF receiver 4 may be that the RF receiver 4 monitors and derives the spectrum varying with time and frequency to produce a spectrograph, wherein the spectrograph is then used in the training system and training method to train a spectral anomaly neural network based on images from the spectrograph. In the monitoring system and monitoring method, the spectrograph or images of the spectrograph are then used to detect spectral anomalies and may even classify them in the spectrograph, wherein, for example, a vertical line in the spectrograph would indicate CW interference. The potential classification may serve as a basis for proposing countermeasures.
[0042] Another possible implementation for monitoring and derivation of the RF receiver 4 may be that the RF receiver 4 derives / uses power values per time unit as a physical measurement of the spectrum, and uses these power values per time unit as a function of frequency in a training system and training method to train time series of these values (pattern recognition). In the monitoring system and monitoring method, these power values per time unit as a function of frequency are then used to detect spectrum anomalies.
[0043] In addition to spectral anomalies similar to those generated by the signal anomaly generator 10 that appear in the spectrum of the cellular wireless network, faulty electronic components may also cause interference and anomalies, which can also be detected by the monitoring system and monitoring method. In this regard, the monitoring system and monitoring method can also be used to monitor channels used for communication and frequency ranges outside the frequency band used. Additional observation of frequency bands outside the channel can allow detection of faults from faulty electronic components, which is independent of damage to the communication path of the material or other indicated component damage. For example, a robot's motor can have an increased power value in a dedicated frequency band before failing due to material fatigue.
[0044] Furthermore, in addition to detecting spectrum anomalies, monitoring systems and monitoring methods can also be developed to locate spectrum anomalies. Figure 2 At least two RF receivers 4 at different points in the 5G private network 8 shown may monitor and derive different spectral and / or physical measurements. Using these different spectral and / or physical measurements, it is possible to triangulate the location / location of a spectral anomaly detected by the processing unit 5, wherein the triangulation is based on the correlation of the spectral and / or physical measurements of the at least two RF receivers 4. Further, at least two RF receivers 4 may be connected to one processing unit 5.
[0045] Data from RF spectrum measurements (spectrum and / or physical measurements of the spectrum, respectively) can also be associated with data from active connections / communications. Performance parameters from active communications are for example the data rate achieved, the potential signal-to-noise ratio, or latency times on active communications. In addition to the RF spectrum, these parameters can also improve the training and identification of anomalies. The monitoring system also derives such additional parameters of active communication links in the cellular wireless network, and the processing unit 5 processes additional parameters in addition to the spectrum and / or physical measurements of the spectrum to detect spectral anomalies. In the training system, the spectral anomaly neural network can also be trained based on the additional parameters of the active communication links.
[0046] The RF receiver 4 of the monitoring system can be positioned / arranged together with the processing unit 5 in one housing or close to each other. However, it is also possible that the processing unit 5 is arranged away from the RF receiver 4 and / or away from the cellular wireless network, such as Figure 2 As shown, it is especially arranged in a cloud service or centrally in a company's computer center. Therefore, the processing unit 5 can also be implemented on a central computer or server.
[0047] Similarly, the RF receiver 4 of the training system may also be positioned / arranged together with the processor 9 in a housing or close to each other. However, it is also possible that the processor 9 is arranged away from the RF receiver and / or away from the cellular wireless network, such as Figure 3 As shown, it is especially arranged in a cloud service or centrally in a company's computer center. Therefore, the processor 9 can also be implemented on a central computer or server.
[0048] Furthermore, data (such as the spectrum and / or physical measurements of the spectrum and / or additional parameters) can also be continuously recorded by the monitoring system and the monitoring method during operation to improve the spectral anomaly neural network, wherein the derived data is used to further train the spectral anomaly neural network. Thus, the spectral anomaly neural network is continuously improved throughout the operation.
Claims
1. A monitoring system for detecting spectrum anomalies in a cellular wireless network, wherein: The cellular wireless network is a 5G ultra-reliable and low-latency communication network (8) using a dedicated spectrum, and the monitoring system includes: - at least two radio frequency receivers (4) for monitoring the frequency spectrum of the cellular radio network and for deriving the frequency spectrum of the cellular radio network and / or physical measurements of the frequency spectrum, and - a processing unit (5), the processing unit (5) being configured to execute a spectrum anomaly neural network trained in a training system by a machine learning algorithm, wherein the processing unit (5) is configured to obtain the spectrum and / or the physical measurement value of the spectrum and process the spectrum and / or the physical measurement value of the spectrum to detect spectrum anomaly information, wherein the at least two RF receivers (4) are configured to monitor and derive the spectrum as it varies with time and frequency to produce a spectrogram, wherein the processing unit (5) is configured to obtain and process the spectrogram, and wherein the monitoring system is configured to triangulate the location of a spectral anomaly detected by the processing unit (5), wherein the triangulation is based on a correlation between the spectra and / or the physical measurements of the at least two RF receivers (4).
2. The monitoring system according to claim 1 further comprises a memory for storing the spectrum and / or the physical measurement value of the spectrum derived by the RF receiver (4), and the processing unit (5) is configured to obtain the spectrum and / or the physical measurement value of the spectrum from the memory.
3. The monitoring system according to claim 1 or 2, in, The radio frequency receiver (4) is configured to derive a power value per time unit varying with frequency as the physical measurement value of the spectrum, wherein the processing unit (5) is configured to obtain and process the power value per time unit varying with frequency.
4. The monitoring system according to any one of claims 1 to 3, wherein: The monitoring system is further configured to derive additional parameters of active communication links in the cellular wireless network, and the processing unit (5) is further configured to obtain and process the additional parameters to detect spectrum anomaly information.
5. The monitoring system according to any one of claims 1 to 4, wherein: The processing unit (5) is arranged in a cloud service.
6. The monitoring system according to any one of claims 1 to 5, further comprising an output unit (2) for presenting the detected spectrum anomaly information.
7. The monitoring system according to claim 4, wherein: The additional parameters are the achieved data rate / bit rate, the signal-to-noise ratio and / or the delay time.
8. The monitoring system according to claim 6, wherein: The output unit (2) is a display or a monitor.
9. A monitoring method for detecting spectrum anomalies in a cellular wireless network, wherein: The cellular wireless network is a 5G ultra-reliable and low-latency communication network (8) using a dedicated spectrum, and the monitoring method comprises: - monitoring the spectrum of the cellular radio network by means of at least two radio frequency receivers (4) and deriving physical measurements of the spectrum of the cellular radio network and / or of the spectrum, and - executing, by the processing unit (5), a spectral anomaly neural network trained in a training method by a machine learning algorithm by processing the derived spectrum and / or the physical measurement value of the spectrum to detect spectral anomaly information, Wherein, the monitoring method further comprises: monitoring and deriving the frequency spectrum varying with time and frequency by means of the at least two radio frequency receivers (4) to generate a spectrogram, The processing unit (5) obtains and processes the spectrogram, and triangulate the positions of the spectral anomalies detected by the processing unit (5), Therein, the triangulation is based on a correlation of the frequency spectra and / or the physical measurement values of the at least two radio frequency receivers.
10. The monitoring method according to claim 9, wherein: The derived frequency spectrum of the cellular wireless network and / or physical measurements of the frequency spectrum are used for further training the spectral anomaly neural network trained by the training method.
