Method for optimizing performance of indoor industrial network, TSN bridge and industrial network
By using AI modules in the TSN bridge to analyze and optimize SINR, delay and obstacle data, the time synchronization problem of TSN network in dynamic environment and device mobility is solved, network performance and synchronization accuracy are improved, and low latency and high Reliable industrial communications.
Patent Information
- Application Number
- CN202410691690.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-06-06
- Filing Date
- 2024-05-30
- Publication Date
- 2024-12-06
AI Technical Summary
Existing TSN networks have difficulty maintaining time synchronization accuracy when facing dynamic environments and device mobility, especially when GPS signals are unavailable or disturbed, resulting in network performance degradation and increased latency.
Introducing artificial intelligence (AI) modules into the TSN bridge, using machine learning algorithms to analyze SINR, delay information, obstacle data, etc., adjust the packet size and time information between mobile devices and fixed devices in real time, ensure time synchronization accuracy, and Optimize network performance through AI modules interact with TSN controllers and user configuration hives (CUCs).
It improves the time synchronization accuracy and network performance in indoor industrial networks, can respond to environmental changes and equipment mobility in real time, reduce network latency and interference impacts, and ensure low-latency and high-reliability communication for industrial applications.
Smart Images

Figure CN119094347A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to industrial networks, and more particularly, to a computer-implemented method, a Time-Sensitive Netwoking (TSN) bridge, and a computer program for optimizing the performance of an indoor industrial network and predicting network problems. Background Art
[0002] In today’s fast-paced world, connected systems have become the backbone of most industries. These systems are particularly dependent on Time Sensitive Networking (TSN), a technology that enables real-time communication over a network. This technology is very important in industries that need to share data instantly and reliably. However, in such implementations, managing network traffic and improving network performance is a major challenge. When the network is too congested with data, it causes delays, which is a serious problem when data needs to be transmitted in a timely manner. Another challenge is how to ensure that all devices on the network are synchronized. TSN applications require accurate time synchronization to ensure smooth and coordinated communication.
[0003] However, several factors such as environmental conditions, network interference, and device clock offsets can affect this synchronization. Traditional TSN networks are usually designed for static conditions and have difficulty coping with mobile scenarios, such as mobile robots in industrial environments. That is, TSN technology struggles when network conditions change or devices on the network move. For example, in an industrial environment using mobile robots, factors such as noise, physical obstacles, or signal interference can lead to degraded communication quality and loss of time accuracy. There are also some problems in dealing with changing network conditions and device mobility. Factors such as signal interference, noise, and physical obstacles can lead to a decrease in the Signal-to-Interference-Plus-Noise ratio (SINR), which ultimately leads to degradation of time accuracy. Current solutions often have difficulty managing these factors effectively.
[0004] Therefore, in view of the foregoing discussion, there is a need to overcome the above-mentioned shortcomings. Summary of the invention
[0005] The object of the present disclosure is to provide a computer-implemented method, a TSN bridge, an indoor industrial network and a method for predicting network problems to improve the accuracy of time synchronization in an indoor industrial network (e.g., a dedicated mobile network) where a Global Navigation Satellite System (GNSS), such as a Global Positioning System (GPS), signal is not available and time information needs to be distributed wirelessly to a mobile device or router. The objects of the present disclosure are achieved by a computer-implemented method and a TSN bridge for optimizing the performance of an indoor industrial network, and an indoor industrial network thereby and a method for predicting network problems in an indoor industrial network as defined in the attached independent claims. Preferred features are listed in the attached dependent claims.
[0006] The embodiments of the present disclosure overcome the above-mentioned problems in the prior art, and can effectively predict network problems in industrial networks and improve the accuracy of time synchronization in indoor industrial networks. BRIEF DESCRIPTION OF THE DRAWINGS
[0007] The above summary and the following detailed description of illustrative embodiments will be more easily understood when read in conjunction with the accompanying drawings.For the purpose of illustrating the present disclosure, reference is made to the following schematic diagrams, each of which shows an exemplary structure of an embodiment of the present disclosure.
[0008] Figure 1 is an illustration of a schematic block diagram of an indoor industrial network according to an embodiment of the present disclosure;
[0009] Figure 2 is an illustration of a schematic block diagram of a TSN system according to an embodiment of the present disclosure;
[0010] Figure 3A and Figure 3B is a diagram of an exemplary time signal according to an embodiment of the present disclosure;
[0011] Figure 4 is an illustration of a schematic diagram of a time synchronization mechanism according to an embodiment of the present disclosure;
[0012] Figure 5 is an illustration of a schematic diagram of a transmission management mechanism according to an embodiment of the present disclosure;
[0013] Figure 6 is an illustration of a flow chart listing the steps involved in a method for optimizing indoor industrial network performance according to an embodiment of the present disclosure; and
[0014] Figure 7is an illustration of a flowchart listing the steps involved in a method for predicting network problems in an indoor industrial network according to an embodiment of the disclosure. DETAILED DESCRIPTION
[0015] The following detailed description illustrates embodiments of the present disclosure and the manner in which they may be implemented.
[0016] In a first aspect, the present disclosure provides a computer-implemented method for optimizing the performance of an indoor industrial network. The method includes: in an artificial intelligence (AI) module on the network side of a TSN bridge of an indoor industrial network, receiving a signal to interference and noise ratio (SINR) and delay information from a synchronization agent of a mobile router on the device side of the TSN bridge; receiving one or more first parameters of the indoor industrial network from one or more first sensors; receiving obstacle data from one or more cameras; determining the degradation of the time accuracy of a clock in one or more mobile devices based on the received SINR, the one or more first parameters, and at least one of the obstacle data; providing information of the determined degradation of the time accuracy to a central network controller (CNC) and a central user configuration (CUC) of a TSN controller of the indoor industrial network; instructing the CNC and the CUC to adjust the size of a data packet to be sent between a mobile device and one or more fixed devices; and instructing a user plane function (UPF) synchronization agent to send time information to one or more mobile devices to recalculate the time information and adjust the clock.
[0017] The above computer-implemented method addresses the key challenges faced by indoor industrial networks regarding time synchronization accuracy and network performance optimization. In this article, the AI module optimizes network performance by receiving and processing various types of data, including SINR, delay information, environmental parameters, and obstacle data. SINR and delay information from the synchronization agent of the mobile router enables understanding of the current state of the network, while data from sensors and cameras provides detailed information about environmental conditions and physical obstacles that may interfere with network operation. The use of the AI module enables intelligent and real-time analysis of this data to determine the accuracy of the clock in the mobile device, identify patterns, and make decisions to improve network efficiency. The AI module communicates the determined time accuracy information to the central units that maintain the network operation—the CNC and CUC. In particular, based on its observations, the AI module instructs the CNC and CUC to adjust the size of the data packets sent between mobile and fixed devices, which helps to manage network traffic more efficiently and reduce the possibility of network congestion. This continuous communication ensures that any problems are resolved in a timely manner and network management is optimally adjusted. Together, these features provide a comprehensive, adaptive, and intelligent solution for optimizing the performance of indoor industrial networks.
[0018] In a second aspect, the present disclosure provides a TSN bridge for optimizing the performance of an indoor industrial network. The device side of the TSN bridge includes a synchronization agent of a mobile router, and the network side of the TSN bridge includes an AI module. The AI module is configured to use a machine learning algorithm to perform the following operations: receiving SINR and delay information from the synchronization agent of the mobile router on the device side of the TSN bridge; receiving and processing parameters of the indoor industrial network, and receiving and processing obstacle data from one or more cameras; determining the degradation of the time accuracy of the clock in one or more mobile devices of the indoor industrial network; communicating with the CNC and CUC of the TSN controller of the indoor industrial network.
[0019] The TSN bridge in this article uses AI and machine learning technologies to solve challenges in network operation. The AI module integrated into the TSN bridge enables the system to use machine learning algorithms to achieve a higher degree of intelligence and adaptability in managing network operation. The TSN bridge can collect and process a variety of data, including SINR, latency information, environmental parameters, and obstacle data from cameras, which enables the AI module to understand the status and operating conditions of the network, which directly affects decisions related to network management and optimization. The TSN bridge also actively monitors and manages the synchronization of devices passing through the network, solving the key challenge of limited GPS access in indoor network operation. The AI module also communicates with the CNC and CUC to ensure that important information such as clock accuracy data is seamlessly transmitted to these central units. The TSN bridge can help maintain consistent coordination and respond quickly to potential problems by facilitating real-time communication.
[0020] In a third aspect, the present disclosure provides an indoor industrial network, comprising: a TSN bridge and an AI module integrated into the TSN bridge. The AI module is configured to optimize the performance of the indoor industrial network by performing the following operations using a machine learning algorithm: analyzing data from synchronization agents of mobile routers, sensors, and cameras; discovering and reporting additional processing delays in the network, uplink (UP) delays and downlink (DL) delays of asymmetric communications, SINR, and obstacles that block signals and increase delays; predicting potential problems in the network; taking proactive measures to resolve these problems; and interacting with other network elements such as UPF synchronization agents, CUCs, and CNCs to optimize the performance of time-sensitive applications.
[0021] Indoor industrial networks that combine a TSN bridge and an AI module embedded in the TSN bridge offer several unique advantages that stem from the synergy of features. Combining the AI module with the TSN bridge creates a highly dynamic, intelligent component in the network. This combination enables indoor industrial networks to respond to changes in the network environment in real time, improving the level of responsiveness that traditional network hardware lacks. The AI module uses machine learning algorithms to analyze data from synchronization agents, sensors, and cameras. This enables indoor industrial networks to identify potential problems such as processing delays, uplink and downlink delays for asymmetric communications, signal interference, and obstacles that increase signal latency. The AI module can also predict potential problems based on the data it analyzes, which enables the network to proactively resolve these problems. The AI module interacts with key network elements such as the UPF synchronization agent, CUC, and CNC to ensure optimal performance for time-sensitive applications required for the operation of indoor industrial networks.
[0022] In a fourth aspect, the present disclosure provides a computer-implemented method for predicting network problems in an indoor industrial network to optimize the performance of the indoor industrial network. The method includes: collecting network data, the network data including at least one of network performance data, sensor data, and device data; using the collected network data to train an AI model; deploying the trained model in a TSN bridge; receiving real-time data from network elements and sensors; using the trained model to predict network problems; adjusting time information sent to mobile devices to improve synchronization, and evaluating the performance of the model.
[0023] The proposed method for predicting potential network problems in indoor industrial networks starts with collecting network data, including network performance data, sensor data, and device data. The network data is used to train an AI model to understand the dynamics of the network. Once the AI model is fully trained, it is deployed in a TSN bridge to provide it with a real-time perspective of the network. When real-time data is received from network elements and sensors, the AI model uses its trained capabilities to predict potential network problems, even before they occur. At the same time, the AI model continuously adjusts the time information associated with the mobile device to ensure improved synchronization via the network. Finally, the performance of the AI model is evaluated to ensure its effectiveness and to determine whether further optimization and adjustment are needed (if necessary). Therefore, the method of the present application provides a proactive, AI-enabled approach to predict and prevent network problems, thereby significantly improving the overall reliability and performance of indoor industrial networks.
[0024] In a fifth aspect, the present disclosure provides a computer program for optimizing indoor industrial network performance, wherein the computer program includes instructions, and when the computer program is executed by a computer, the instructions cause the computer to execute the above method.
[0025] The computer program provides a set of instructions that, when executed on a computer, triggers a series of actions consistent with the aforementioned computer-implemented method for optimizing the performance of an indoor industrial network. Thus, the computer program acts as an enabler for converting the prescribed method into digital operations that are operable within the network.
[0026] As used herein, the term "indoor industrial network" (hereinafter, sometimes referred to as "network") refers to a digital communication infrastructure established in an indoor industrial environment to facilitate data exchange and various automated operations. These networks are designed to provide robust, secure and efficient connections for numerous devices and devices involved in industrial processes. These networks are the backbone that integrates and coordinates different systems and technologies to ensure seamless operation of industrial organizations. Examples of indoor industrial networks can be found in different industrial environments, such as manufacturing plants, warehouses, data centers, hospitals, etc. Indoor industrial networks such as dedicated mobile networks for industrial use include wireless devices such as mobile devices and fixed devices (base stations). Mobile devices are, for example, mobile robots and mobile industrial machines. These mobile devices are connected to the network via the fifth generation mobile communication technology (5th Generation Mobile Communication Technology, 5G) radio or WiFi. These mobile devices can integrate 5G or WiFi modems in the device, or establish connections through 5G routers. Here, mobile devices are also referred to as "end stations (End Station)" without any restrictions. Fixed devices are wired devices, such as non-mobile machines, non-mobile robots, computers in control centers, or computers running services.
[0027] Time synchronization is a critical requirement for many industrial applications. Time synchronization ensures coordinated operation across multiple devices and is required for automation, real-time data transmission, process control, and other time-sensitive operations. Maintaining accurate time synchronization in dedicated mobile networks, especially those deployed indoors, is a challenging task. In many networks, the primary method of time synchronization relies on GPS signals. However, a major problem often faced in indoor environments is that GPS signals are often weak or non-existent inside buildings because GPS-like signals cannot penetrate dense materials such as concrete and steel. The lack of GPS signals leads to inaccurate time synchronization, affecting the performance and reliability of the network and can cause a cascade of problems in industrial processes that rely on the network. In addition, in dedicated mobile networks used in industry, network devices such as mobile devices and routers are often moving around the network. The network needs to maintain time synchronization between these devices in real time to support seamless operation. Distributing time information over wireless networks faces a series of challenges because factors such as signal interference, network congestion, and latency can introduce additional delays or jitter that lead to inaccurate time synchronization.
[0028] The computer-implemented method of the present application is dedicated to enhancing the capabilities of dedicated mobile networks for industrial applications, with particular emphasis on the implementation of TSN. TSN uses mobile devices and fixed devices to facilitate precise synchronization, which enables the TSN controller to accurately schedule communication time slots. The purpose is to coordinate wired and wireless devices so that they can communicate in precisely allocated time slots, which can improve network efficiency. The method uses AI to improve the accuracy of time synchronization. The AI module processes various parameters before relaying time information to mobile devices. In order to enhance the reliability and predictability of the network, AI is trained based on optimal, conventional, and high-risk scenarios. This enables AI to predict network behavior and potential problems. In addition, reliability can be further improved by utilizing multiple radios including various 5G base stations and Wi-Fi connections. This may involve connecting to the same device from different angles, and even working in parallel to send repeated timing packets.
[0029] The computer-implemented method for optimizing the performance of an indoor industrial network begins with an AI module located on the network side of a TSN bridge. In particular, the AI module on the network side of the TSN bridge of the indoor industrial network can be connected to the synchronization agent of the TSN bridge, the synchronization agent of the mobile device, and the CNC and CUC of the TSN controller. The AI module is configured to discover additional processing delays in the indoor industrial network (including the 5G network therein), as well as the UP and DL delays of asymmetric communications during training. When the AI module is integrated into the TSN bridge of the indoor industrial network, the AI module enables the network to adapt to changing conditions and ensure low-latency, high-reliability communication for time-sensitive applications, which is required for the normal operation of industrial devices in the network.
[0030] The AI module drives optimization of network performance, including features that enable time synchronization by receiving and processing specific types of data from various sources within the network. In this article, a synchronization agent residing in a mobile router on the device side of a TSN bridge helps maintain synchronization within the network by managing the transmission of time-sensitive data between the network side and the device side of the TSN bridge. The synchronization agent communicates with the AI module by sending two types of information: SINR and delay information. SINR is a measurement that observes the quality of a wireless signal. SINR takes into account not only the expected signal power, but also the level of interference (from other signals) and noise (random, unexpected variations) that can distort the signal. A high SINR value indicates a clear, high-quality signal, which is required for accurate time synchronization. Delay information, on the other hand, indicates the time it takes for a data packet to be transmitted from one point in the network to another. In the context of time synchronization, delay information can help determine if there are any delays in the data transmission that may affect the accuracy of synchronization.
[0031] In addition to receiving SINR and delay information from the synchronization agent, the AI module also collects information from the first sensor that can be distributed on the indoor industrial network, and monitors and records various parameters or characteristics of the network environment. The nature of these first parameters can vary according to the specific design and requirements of the network, and can include factors such as signal strength, data transmission speed, network traffic, and similar network-specific metrics that may affect the accuracy of time synchronization. In particular, the first parameters may include factors such as the mobility of devices within the network, the distance between the transmitter and the receiver, and any interference present in the network. In this article, specifically, interference can take various forms. For example, an industrial environment may include various machines and devices that emit electromagnetic interference. Such interference can affect the SINR of devices such as mobile robots within the network by destroying the wireless communication modules therein. Other sources of interference include external noise, such as electrical interference and background noise. In this article, the AI module can measure and report electromagnetic interference in the network, which helps to take appropriate measures to deal with the noise. In addition, other signals present in the same frequency band as the desired signal may also cause interference and reduce SINR. Such interference signals can be detected using a spectrum analyzer or a signal strength meter.
[0032] Information about these first parameters helps the AI module make more informed decisions about optimizing network performance. For example, if the AI module identifies that a machine is causing such interference, it may take necessary actions, which may involve adjusting the transmit power level of the robot's wireless communication module or changing the direction of the antenna to maintain a stable SINR level. The process of collecting these first parameters is a continuous real-time operation. That is, the first sensor can continuously measure these parameters and then transmit the collected data to the AI module. This continuous influx of data enables the AI module to maintain an up-to-date understanding of network conditions and make timely adjustments to maintain or improve the accuracy of time synchronization.
[0033] In addition, in addition to receiving data from the synchronization agent and the first sensor, the AI module also collects data from cameras installed in the indoor industrial network. These cameras are configured to capture visual data from the environment, which can be analyzed to identify potential obstacles that may interfere with network performance. These cameras can be strategically placed to obtain a comprehensive view of the indoor industrial network, so as to capture any changes in the environment that may affect signal transmission. In this article, obstacle data refers to information related to physical obstacles that may affect the quality of wireless signals in the network. These obstacles can include walls, buildings, trees, pillars, devices, and even people moving in the environment. The nature of these obstacles may vary from fixed infrastructure to temporary mobile obstacles. In the configuration of the present application, the AI module analyzes and identifies obstacles and may involve technologies such as computer vision and image processing, which can identify and classify different objects in visual data. It is understandable that physical obstacles may significantly reduce wireless signals, resulting in lower SINR values and increased latency. These negative effects may then have an adverse effect on the accuracy of time synchronization in the network. By identifying these obstacles, the AI module can predict and mitigate their impact on network performance. For example, the AI module can optimize the network's routing algorithms based on identified obstacles, directing signals along paths that avoid these obstacles to ensure smoother communications, as discussed later in the specification.
[0034] As mentioned earlier, time synchronization is critical in a TSN environment, and any degradation in time accuracy can significantly affect the overall performance of the network, especially for industrial networks, where processes and devices often rely on precise timing to operate. Once the AI module has received all the necessary data, including SINR, latency information, first parameters from the first sensor, and obstacle data from the camera, the AI module begins analyzing and processing this information to estimate the time accuracy of the clock in the mobile device. In this article, the AI module uses a machine learning algorithm to perform this complex task. By considering all of these factors, the AI module applies a machine learning algorithm to determine the degree of degradation of the time accuracy of the mobile device clock. For example, the AI module can identify a correlation between high network load and increased timing differences, or identify a correlation between certain types of obstacles and the degradation of specific forms of signals. Therefore, by using this diverse array of network and environmental data, the AI module can generate an accurate understanding of how various factors affect the time synchronization accuracy within the network, and can then determine measures that can help maintain accurate time synchronization based on this understanding.
[0035] Once the AI module determines that the time accuracy of the clock in the mobile device has degraded, information about the degradation of the time accuracy is transmitted to the components that control and manage the network, including the CNC and CUC of the TSN controller of the indoor industrial network. It can be imagined that in the network, the CNC is responsible for controlling the operation of the network and handling the configuration of the equipment by adjusting the network parameters to ensure efficient and reliable network performance. The CNC also coordinates the timing and scheduling of data transmission via the network to meet the real-time requirements of the industrial applications it serves. On the other hand, the CUC is generally responsible for setting and maintaining user preferences and configurations, and can also manage terminal devices within the network, adjusting their settings as needed to maintain optimal performance. By providing the CNC and CUC with information determined about the degradation of time accuracy, the AI module ensures that these components of the network have the most accurate and up-to-date information about the timing performance of the network. The CNC and CUC can then use this information to adjust network parameters, reconfigure equipment, or change data transmission scheduling as needed to correct any identified timing inaccuracies.
[0036] In an example, if the AI module detects a pattern in which the clocks of mobile devices lose synchronization during periods of high network load, that information can be shared with the CNC. The CNC can then use that information to adjust the data transmission schedule during those periods, ensuring that timing information can still be accurately delivered to all devices. Similarly, if the AI module identifies a certain type of obstruction that is causing signal degradation and resulting in timing inaccuracies, that information can be shared with the CUC. The CUC can then reconfigure the settings of the affected devices to mitigate the effects of those obstructions.
[0037] In addition, the AI module instructs the CNC and CUC to adjust the size of the data packets sent between the mobile devices and the fixed devices (i.e., to accommodate the clock differences that will affect the transmit time (Tx) slot and the receive time (Rx) slot). As used in the context of indoor industrial networks, mobile devices and fixed devices refer to two types of devices used in a network environment. As previously discussed, mobile devices can be portable electronic devices that can be easily carried or moved in an industrial environment, such as smartphones, tablets, handheld scanners, etc. Fixed devices can be stationary devices that are permanently installed or assembled at specific locations within an indoor industrial network, such as base stations, industrial control systems, network switches and routers, servers, etc. In addition, as used in this article, the size of the data packet can greatly affect the performance of the network. If the data packets are too large, they may exceed the maximum transmission unit (MTU) of the network, resulting in fragmentation, which may result in increased overhead and latency, negatively affecting network performance. Conversely, if the data packets are too small, this may result in the transmission of too many data packets, thereby increasing network overhead and possibly causing congestion.
[0038] Therefore, the AI module instructs the CNC and CUC to adjust the packet size to accommodate the identified timing inaccuracy. For example, if the AI module determines that the clock in the mobile device is running slowly, causing the mobile device to receive data later than expected, the AI module can instruct the CNC and CUC to reduce the packet size. In this way, it will take less time to transmit each packet, helping to compensate for the slower clock in the mobile device. Alternatively, if the AI module determines that the clock in the mobile device is running faster, it can instruct the CNC and CUC to increase the packet size. This can help ensure that the faster clock in the mobile device does not cause them to receive data too quickly, which can potentially cause congestion or data loss.
[0039] Furthermore, after identifying the degradation of the time accuracy of the clock in the mobile device and transmitting this information to the CNC and CUC, the next step in the process is to instruct the UPF synchronization agent to send updated time information to the mobile device. Here, the UPF synchronization agent is primarily responsible for data traffic routing and management in the indoor industrial network. The UPF synchronization agent manages the propagation of timing information on the network, ensuring accurate time synchronization. Specifically, the AI module instructs the UPF synchronization agent to send updated time information to the mobile device. When new time information is received, the mobile devices recalculate their time information and adjust their clocks. This is done to accommodate any time differences or drifts that may occur due to factors such as signal interference, delays, or obstacles. For example, if the AI module has determined that the clock of a particular mobile device continues to lag behind due to signal interference from physical obstacles, the UPF synchronization agent can forward the timing correction to the device. This will enable the mobile device to recalculate the time information and adjust the corresponding clock, thereby improving time synchronization within the indoor industrial network.
[0040] Therefore, the method involves a time synchronization process of communication between a reference clock and an end device within an industrial indoor network, which can be a mobile device or router connected via 5G or Wi-Fi. Time synchronization in an industrial indoor network can be achieved using protocols such as the Network Time Protocol (NTP). A grand master clock, which is typically a highly accurate reference clock, can send time information to other devices in the network, including AI / Machine Learning (ML) enabled devices (such as AI modules) and end devices such as mobile phones and routers. These devices can then adjust their own clocks to match the time provided by the grand master clock. In a 5G or Wi-Fi network, this synchronization can occur on the control plane or management plane using NTP or other time synchronization protocols.
[0041] The process includes calculating offset and transmission delay, which play an important role in maintaining accurate time synchronization. The offset between the master clock (reference clock) and the slave clock (end device clock) is calculated. This calculation is represented as T2-T1, which measures the time difference between the two instances. A similar calculation T4-T3 is performed to define the offset of the slave-master delay. The average of these two values is then calculated as the total offset, represented as ((T2-T1)-(T4-T3)) / 2. In addition to the offset, the transmission delay is also critical to understanding the time relationship between the master clock and the slave clock. The value of this transmission delay is calculated by adding the two offset measurements (T2-T1) and (T4-T3) and dividing the sum by 2 to obtain ((T2-T1)+(T4-T3)) / 2.
[0042] The process also extends to TSN data transmission between fixed devices and mobile devices in indoor industrial networks. In this case, the packet size is determined by the difference between the transceiver (TRX) speed divided by (T2-T1) and the end-to-end transmission (TX) delay. The AI module is configured to monitor the clock of the mobile device and detect any differences or changes in the TX delay. If the AI module identifies that the clock of the mobile device is out of sync, or if the AI module detects a change in TX delay, the AI module will update the TX delay within the network accordingly. In addition, the AI module can identify if any mobile device link is unreliable. To counter this, the AI module is programmed to replicate data over different radios (such as 5G or WiFi), providing redundancy measures to ensure reliable data transmission.
[0043] In an embodiment, the method further includes: measuring one or more second parameters of the environment of the indoor industrial network by one or more second sensors, the one or more second parameters including temperature and humidity. That is, additional sensors are used to measure other important environmental parameters such as temperature and humidity, which provides a more comprehensive perspective of the conditions of the indoor industrial network environment. It is understandable that temperature and humidity can significantly affect the performance of network equipment and wireless signal propagation. High temperature may cause equipment to overheat, which may in turn cause equipment failure or failure. In addition, changes in temperature may affect the clock accuracy of the device, because the crystal oscillator in the electronic device used for timing may cause its frequency to change due to the influence of temperature changes. Similarly, higher humidity may cause condensation and may damage electronic components. In addition, humidity may affect the attenuation of wireless signals, especially higher frequencies, which may cause reduced signal strength and network performance. In another example, temperature and humidity information is also used to adjust the power level of devices in the network to improve their performance under different environmental conditions.
[0044] In the configuration of this article, the second sensor measures the temperature and humidity of the network environment and sends this data to the AI module. The AI module can then combine these second parameters with SINR, delay information, and obstacle data to determine the timing accuracy of the clock in the mobile device with higher accuracy. For example, if the AI module identifies a high temperature and high humidity environment, it can predict the potential decline in clock accuracy or signal strength and instruct the CNC and CUC to proactively adjust the packet size or transmission scheduling accordingly, thereby mitigating potential interference in indoor industrial networks.
[0045] In an embodiment, the method includes: configuring the CNC to set the TSN frame transmission schedule and preference, and configuring the CUC to set the TSN frame transmission schedule in the terminal device. That is, the CNC that monitors the entire network architecture can be programmed to set the schedule and preference for transmitting TSN frames. This is done to ensure that time-sensitive data packets are prioritized and routed efficiently through the network to maintain synchronization and high network performance. At the same time, the CUC that processes the terminal device configuration can be instructed to set the TSN frame transmission schedule in the terminal device. The method of the present application can enhance synchronization and reduce potential delays by controlling the timing of data transmission from these devices. For example, if the AI module detects a degradation of time accuracy in a mobile device due to high interference or the presence of physical obstacles, in this scenario, the AI module can instruct the CNC and CUC to adjust the TSN frame transmission schedule. In this case, for example, the CNC and CUC change the transmission timing to a less crowded period, or prioritize the transmission of certain types of data packets to ensure uninterrupted and synchronized data flow.
[0046] In an embodiment, in the method of the present application, the above adjustment includes: if the synchronization is inaccurate, then reduce the size of the data packet, or if the synchronization is accurate, then maximize the size of the data packet. As discussed above, the adjustment of the packet size directly affects the accuracy of time synchronization within the network. In an indoor industrial network setting, the AI module in the TSN bridge works with the CNC and CUC to facilitate this dynamic adjustment process. When the AI module detects that the time accuracy of the clock in the mobile device is degraded, for example, a signal indicating that the synchronization via the network is inaccurate is detected, the AI module instructs the CNC and CUC to reduce the size of the data packet to be transmitted between the mobile device and the fixed device, because smaller data packets can be transmitted faster and are less likely to be delayed or lost due to network congestion or signal interference. In contrast, when the synchronization via the network is accurate and has a high SINR, the AI module can increase the data packet size to more efficiently utilize the network bandwidth because the ratio of header information (not carrying application data) to payload data (carrying application data) is low. This method can also reduce packet conflicts and processing loads on network devices because it requires fewer packets to be processed overall.
[0047] In an embodiment, the method includes: receiving a reference time from one or more mobile devices with GPS access rights, the reference time being used to calculate the distance between the one or more mobile devices with GPS access rights and one or more fixed devices. That is, the AI module in the TSN bridge is configured to receive the reference time from any mobile device in the network that can access GPS. The reference time is typically acquired from a GPS satellite and is globally synchronized, providing a high-precision timestamp that can serve as a universal reference for the network. The reference time is then used to calculate the distance between a mobile device (with access to GPS) and a fixed device within an indoor industrial network. Since the speed of signal transmission is approximately known (close to the speed of light), the AI module can determine how far the sending device is by comparing the reference time (when the signal is sent) and the local time (when the signal is received). This distance calculation can also be used to estimate the layout and / or scale of indoor industrial networks, helping to optimize packet routing and minimize latency.
[0048] In an embodiment, the method includes: receiving the highest residence time from the AI module in the de-jitter module. As used herein, "residence time" is the amount of time a data packet spends within a system or a specific part of the system. In the context of the present application, the residence time is the time it takes for a data packet to pass through the TSN bridge. Therefore, here, the "highest residence time" will be the longest time taken for the process. In addition, as used in the present application, "network jitter" refers to the variation in the delay of the received packet. High jitter may cause poor performance or even cause the failure of time-sensitive tasks. By monitoring the highest residence time, the de-jitter module can understand the behavior of the network and implement techniques to reduce jitter. These techniques may include caching data packets to balance delays or rerouting data to avoid congested or slow parts of the network. In addition, by monitoring the highest residence time, the de-jitter module can identify potential bottlenecks in the network and take proactive measures to resolve them. For example, if the highest residence time increases steadily, this may indicate that a specific part of the network is becoming congested. Then, the de-jitter module can instruct the CNC to redistribute network traffic, such as via the AI module, to alleviate congestion and reduce the residence time in the network.
[0049] The present disclosure also relates to a TSN bridge as described above.The various embodiments and variants disclosed with respect to the above-described computer-implemented method for optimizing the performance of an indoor industrial network apply mutatis mutandis to the TSN bridge.
[0050] It can be understood that the TSN bridge helps to enhance the performance of indoor industrial networks. The TSN bridge has a dual-sided architecture, with the device side containing a synchronization agent of a mobile router and the network side containing an AI module. In this article, on the device side, the synchronization agent of the mobile router handles time-sensitive data transmitted over the network and is responsible for maintaining the precise timing required for synchronous data transmission via the network. This synchronous data communication is particularly important in indoor industrial networks, where devices often need to perform time-sensitive tasks in a collaborative manner. On the network side, the TSN bridge includes an AI module that is specifically designed to optimize performance using machine learning algorithms. The AI module processes different parameters related to the indoor industrial network to determine the degradation of the time accuracy of the clock in one or more mobile devices within the indoor industrial network. The AI module also communicates with the CNC and CUC of the TSN controller within the indoor industrial network to provide information about the time accuracy of the clock, potential network interference, and other relevant parameters. Based on this data, the CNC and CUC can make necessary adjustments to optimize the performance of time-sensitive applications, thereby improving the overall efficiency of the indoor industrial network.
[0051] The present disclosure also relates to an indoor industrial network as described above. The various embodiments and variations disclosed with respect to the above-described computer-implemented method for optimizing the performance of an indoor industrial network are mutatis mutandis applicable to the indoor industrial network.
[0052] Indoor industrial networks use TSN bridges to maintain the time accuracy required for synchronized data exchanges within the network. To enhance the capabilities of the network, AI modules are integrated directly into the TSN bridge. The AI module utilizes machine learning algorithms, an advanced computing method that learns and improves from experience, to optimize the overall performance of the network. Specifically, the AI module is configured to detect additional processing delays within the network that can adversely affect the efficiency of network operations. The AI module can also identify UP delays and DL delays in asymmetric communications that can contribute to overall network latency. The AI module can also analyze network data to predict potential problems that may arise in the future. By doing so, the AI module enables indoor industrial networks to take preemptive action to resolve these problems before they become serious, helping to maintain optimal performance of the network. The AI module also interacts with other key network elements, such as the UPF synchronization agent, CUC, and CNC, to effectively communicate the necessary adjustments required to optimize the performance of time-sensitive applications in current indoor industrial networks.
[0053] The present disclosure also relates to a method for predicting network problems in an indoor industrial network as described above. The various embodiments and variations disclosed with respect to the above computer-implemented method for optimizing the performance of an indoor industrial network apply mutatis mutandis to the method for predicting network problems in an indoor industrial network.
[0054] The method of the present invention for predicting network problems in indoor industrial networks begins with the collection of various types of network data, including: network performance data, sensor data, and device data. The collected data is then used to train the AI model, and once the AI model is properly trained, the model (as described above, as an AI module) is deployed in the TSN bridge. The TSN bridge receives real-time data from network elements and sensors, and the AI model uses this information to predict potential network problems. In addition, the AI model plays a role in adjusting the time information communicated with mobile devices, thereby improving the synchronization of these mobile devices with the network. The performance of the AI model is continuously evaluated to ensure optimal effectiveness.
[0055] In some examples, the AI module (corresponding to the AI model) also estimates the accuracy of the clock in the mobile device. This estimate is required to maintain accurate time synchronization via the indoor industrial network. The AI module also adjusts the size of the data packet transmitted between the mobile device and the fixed device. For example, if the synchronization is inaccurate, the AI module can reduce the data packet size to avoid conflicts, or if the synchronization is accurate, the data packet size can be increased to optimize bandwidth usage. In addition, the AI module continuously monitors clock synchronization and transmission (TX) delay changes in the mobile device. If the AI module detects any differences, it updates the TX delay in the mobile network to improve time synchronization. Finally, if the link to the mobile device is detected to be unreliable, the AI module replicates the data over a different radio (e.g., 5G or Wi-Fi). This redundancy ensures reliable communication and accurate time synchronization even if the radio over one link is unstable.
[0056] According to an embodiment of the present disclosure, an indoor industrial network is equipped with two integrated software modules to ensure smooth and synchronized communication via the network. The first software module is embedded in the mobile device and is responsible for receiving time information from the network and applying logic to correct potential inaccuracies caused by the wireless connection. The mobile module communicates with the second software module located within the network. The network module can access the reference time and transmit it to the mobile device. The operation between these two modules ensures that the mobile device can synchronize its own clock or provide the necessary time information to other connected devices, enabling these other connected devices to synchronize their clocks. This feature is particularly beneficial when the first mobile device operates as a mobile router connected to other devices.
[0057] The mobile device module also manages latency on the wireless connection to the network. This is accomplished by configuring the mobile device to send a probe message to the base station. The message carries time information that helps calculate latency in both the uplink and downlink. The latency can vary with the distance of the device from the base station; higher latency generally indicates lower SINR, which indicates that the device is farther from the base station or that the device is operating in a poor radio channel.
[0058] To predict the relationship between SINR levels and latency, AI / ML algorithms are deployed. When calculating latency, cameras are used to identify any obstructions in the path between the mobile device and the network that could cause increased latency. AI / ML, specifically Bayesian networks, are used to estimate the impact of these obstructions and, therefore, the resulting additional latency.
[0059] Mobile devices with GPS access use the GPS information as a reference time to help calculate latency on wireless connections based on the distance between the GPS-enabled mobile device and the base station. In addition, software in these mobile devices is adapted to request to join a network slice to establish an advanced radio link for sharing time information.
[0060] Alternatively, a secondary or backup master clock may be used in the event of a failure or interruption of the primary clock. The node may include a GPS signal and may be used as a reference for measuring delays, etc. The node may also be used as a redundant secondary grand master clock or backup clock for a time source.
[0061] The network module integrates an AI module that calculates the correction time to be added to the time information sent to the mobile device. The network module is configured to receive information from the mobile device to calculate the delay based on factors such as SINR and identified obstacles. The network module then applies the correction factor to the reference time sent to the mobile device.
[0062] Timestamps from the base station are also integrated into the network module to help calculate latency. The network module includes a TSNNC / CUC or OLE for Process Control Unified Architecture (OPC UA) interface (OPC Unified Architecture is a standardized interface or protocol used in industrial automation and control systems, where OPC stands for OLE for Process Control and OLE stands for Object Linking and Embedding, a technology developed by Microsoft for integrating and sharing data between different applications in the Windows operating system). This interface allows devices in the fixed network to adjust data transmission based on changes in the latency of the wireless connection, which means that the network module is configured to alert the fixed TSN controller or OPC UA about any changes in transmission latency.
[0063] The network module also provides information about latency bounds, i.e., the maximum-minimum values in the wireless network, to other devices in the fixed network using OPC UA or TSN CNC / CUC interfaces. These fixed devices then use this information to set their transmission schedules. To ensure that these bounds are consistent, the network module also implements latency mitigation schemes, such as buffering to reduce latency variations. In this way, fixed devices can rely on the received information to set their schedules.
[0064] In addition, the AI model embedded in the network also calculates the dwell time of each mobile device, taking into account factors such as distance to the base station, mobility, and the presence of obstacles. The AI algorithm estimates the dwell time as a random function, taking these variables into account. In addition, the de-jitter module obtains the maximum dwell time from the AI module to buffer the data from the device, aligning all the data for timed communication.
[0065] The entire system is encapsulated as an OPC UA module that presents the 5G system as a metadata structure, managed by a fixed OPC UA controller or a TSN switch supervised by a TSN controller.
[0066] The present disclosure may be implemented in an industrial factory environment operating a dedicated mobile network for industrial use. Such a network plays a vital role in controlling and coordinating the movement of robots and other industrial machines across the factory floor. The network includes a TSN bridge with an integrated AI module. The AI module is designed to receive various data from the network, including SINR information and latency information from the synchronization agent of the mobile router on the device side of the TSN bridge. The AI module also collects ambient temperature and humidity data from sensors and obtains obstacle data from cameras. The AI module uses machine learning algorithms to process these different data sets and make real-time adjustments to the network. As a result, the AI module optimizes performance and ensures low-latency, high-reliability communications, which are required for time-sensitive applications.
[0067] For example, suppose the AI module identifies an additional 5 milliseconds of processing delay introduced by a particular router in the network. The AI module can quickly adjust the routing algorithm to bypass that router. Similarly, if the AI module detects an obstacle blocking the signal path between the mobile device and the base station, it can adjust the mobile device's transmit power level to compensate for the obstacle. In addition, the AI module adaptively adjusts transmission parameters to offset the effects of temperature and humidity on signal quality.
[0068] In addition to making real-time adjustments, the AI module also helps with performance monitoring. The AI module provides measured parameters and obstacle data to the control center of the industrial network. This comprehensive data perspective enables operators to keep a close eye on network performance and troubleshoot problems when they arise. In addition, the AI module regularly measures the clock accuracy of mobile devices in the industrial network and adjusts the time information distributed to these devices to enhance synchronization. The AI module also takes into account any interference that may potentially compromise the accuracy of the clocks in mobile devices. The above-mentioned interference includes fluctuations in SINR, which can be affected by a variety of factors, such as the distance between the transmitter and receiver, interfering signals, obstacles, antenna direction, noise, channel conditions, and the mobility of the device.
[0069] In a more specific scenario, such as a factory using a fleet of mobile robots for material transportation, these robots are equipped with wireless communication modules that enable them to interact with the TSN network. The AI module in the TSN bridge receives SNR information and latency from the synchronization agent of the mobile router in addition to receiving temperature, humidity, and obstacle data from the factory floor. The AI module uses machine learning algorithms to process this data to identify potential network problems. For example, the AI module can determine that the SINR of the wireless communication modules of these robots drops when the robots approach specific machines in the factory. Using this information, adjustments can be made to maintain consistent network performance.
[0070] The AI module also monitors the clock accuracy of the robot's wireless communication module. The AI module does this by measuring the delay and comparing it to the expected delay. On this basis, the AI module optimizes and adjusts the time information distributed to the robot, improves synchronization, and ensures the accuracy of the clock. For optimal accuracy in time synchronization, the present disclosure can utilize a Precise Time Protocol (PTP) such as IEEE1588 to synchronize the clocks in mobile devices and network elements. PTP uses a master-slave architecture in which the master clock propagates time information to the slave clocks, ensuring highly accurate time coordination. Alternatively, a GPS receiver can provide an accurate time reference to help with synchronization.
[0071] In terms of latency prediction or identifying potential reliability issues, the AI module analyzes data from various sources such as the mobile router's synchronization agent, sensors, and cameras. This data is processed to detect patterns and anomalies that may help identify emerging problems, such as a sudden surge in latency or a drop in SINR. By using historical data, the AI module can even predict when and where problems may occur, allowing proactive measures to be implemented.
[0072] The AI module can also be equipped to detect and report additional processing delays in the network, uplink and downlink delays for asymmetric communications, SNR fluctuations, and obstacles that block signals and increase latency. In addition to this, the AI module can predict potential problems in the network and take proactive measures to resolve them while interacting with other network elements such as the UPF synchronization agent, CUC, and CNC to optimize the performance of time-sensitive applications.
[0073] Detailed description of the drawings
[0074] refer to Figure 1, a schematic block diagram of an indoor industrial network 100 (e.g., a private network) according to an embodiment of the present disclosure is shown. As shown in the figure, the indoor industrial network 100 includes a terminal station (terminal device) 102, a mobile router 104 including a synchronization agent 106 acting as one or more first sensors, and one or more radio modems 108. The indoor industrial network 100 also includes a mobile network architecture 110. The mobile network architecture 110 includes a plurality of base stations (gNodeB, gNB) 112 (which are components of the fifth generation (Fifth Generation, 5G) wireless network architecture), a Wi-Fi access point (Access Point, AP) 114, a PTP-compatible transport layer 116, and a UPF 118 with a synchronization agent 120. The UPF 118 has a time source indicator (TimeSource Indicator, TSi), which is a mechanism for identifying and transmitting the accuracy and reliability of a reference clock for time synchronization in a TSN network. The indoor industrial network 100 also includes a fixed network architecture 122. The fixed network architecture 122 includes a TSN bridge 124, a time-sensitive network generalized multicast (TSN Generalized Multicast, TSN-GM) 126, and an end station (end device) 128, which together form a TSN working domain (indicated by reference numeral 130). The indoor industrial network 100 also includes a reference clock 132 for training a senior master clock 134. The indoor industrial network 100 also includes a camera 136, and a temperature sensor 138 and a humidity sensor 140 that act as one or more second sensors. In addition, as shown in the figure, each network component in the mobile network architecture 110 and the fixed network architecture 122 includes a TSN module similar to the reference clock (as shown in the figure). In addition, as shown in the figure, the synchronization agent 106 provides a synchronization signal "S" to the end station 102. The gNB 112 provides a radio signal "R" (similar to a 5G signal) to the mobile router 104, and the Wi-Fi AP 114 provides a Wi-Fi signal "W" to the mobile router 104. The PTP compatible transport layer 116 provides a signal "s" to the UPF 118, and the TSN working domain 130 provides a modulated signal "M" to the UPF 118. In addition, as shown in the figure, the camera 136 provides an image signal "S1", the temperature sensor 138 provides a temperature signal "S2", and the humidity sensor 140 provides a humidity signal "S3".
[0075] In the indoor industrial network 100, the end station 102 is a networked device such as a computer or a robot that communicates with other networked devices or the network itself and receives a synchronization signal "S" from a synchronization agent 106 to ensure time synchronization of the networked devices. The mobile router 104 including the synchronization agent 106 enables networked devices to communicate with each other. The mobile router 104 receives a radio signal "R" and a Wi-Fi signal "W" from the gNB 112 and the Wi-Fi AP 114, respectively, to achieve wireless connection. The synchronization agent 106 within the mobile router 104 ensures time synchronization of the end station 102 by providing the synchronization signal "S". The radio modem 108 is a device that modulates and demodulates radio signals for communication. The gNB 112, as a component of the 5G architecture, processes the radio signal "R" that provides 5G connection to the mobile router 104. The Wi-Fi AP 114 sends a Wi-Fi signal "W" to the mobile router 104, enabling Wi-Fi connection. The PTP-compatible transport layer 116 provides a signal “s” to the user plane function (UPF) 118, ensuring accurate time synchronization via the network. In addition, the UPF 118, equipped with a synchronization agent 120 (and Tsi), manages the routing and forwarding of data packets. The TSN bridge 124 acts as a bridge between the fixed network architecture 122 and other devices, facilitating data transmission. The TSN-GM 126 provides a protocol that ensures efficient multicast data distribution in a time-sensitive network. The end station 128 is similar to the end station 102 and communicates with the network or other networked devices within the TSN working domain 130. The reference clock 132 is used for training to ensure accurate timing via the network, and the senior master clock 134 provides a master time source for the network to ensure accurate time synchronization. The camera 136, temperature sensor 138, and humidity sensor 140 monitor the environment and provide corresponding image signals (“S1”), temperature signals (“S2”), and humidity signals (“S3”) for further processing. All of these components work together to maintain reliable, high-speed, time-synchronized communications within the indoor industrial network 100.
[0076] Reference Figure 2, a schematic block diagram of a TSN system 200 according to an embodiment of the present disclosure is shown. As shown in the figure, the TSN system 200 includes a TSN bridge 202 having a device side 204 and a network side 206. The device side 204 of the TSN bridge 202 includes a synchronization agent 208 and a modem 209 of a mobile router 210, which receive and process data for transmission via the TSN system 200. The network side 206 (as a 5G core) of the TSN bridge 202 includes a unified data management (UDM) module 208 that stores and manages user data, and the UDM module 208 is linked to a network exposure function (NEF) 210 and is linked to an AI module 212 through an N33 interface. The UDM module 208 is also connected to an access and mobility function (AMF) 214 through an N8 interface, and is connected to a session management function (SMF) 216 through an N10 interface. AMF 214 is responsible for access and mobility management of the device, and is also connected to SMF 216 through an N11 interface, to Next Generation-Radio Access Network (NG-RAN) 218 through an N2 interface, and to the modem 209 of the device side 204 of the TSN bridge 202 through an N1 interface. SMF 216 manages sessions in the network, and is connected to Policy Control Function (PCF) 220 through an N7 interface, and to the synchronization agent 222 of UPF 224 through an N4 interface. UPF 224 includes synchronization agent 222 and is responsible for routing and forwarding of data packets. PCF 220 establishes policies that control how the network operates. NEF 210 opens the services and capabilities of the network to AI module 212 through an N33 interface. NG-RAN 218 provides connectivity between the device and the core network, and is also connected to UPF 224 through an N3 interface. Here, N1, N2, N3, N7, N8, N10, N11, and N33 are interfaces that facilitate connection and communication between different network elements. The TSN system 200 also includes a user plane (User-Plane, U-Plane), which is responsible for transmitting user data, and a control plane (Control-Plane, C-Plane), which is responsible for controlling network operation.
[0077] The TSN system 200 also includes a TSN unit 230 having a talker 232. The TSN system 200 also includes a TSN controller 240, which includes a CNC 242 and a CUC 244. The TSN controller 240 helps manage network traffic and configuration to meet the strict timing requirements in industrial applications. The TSN controller 240 also includes a listener 246 and a bridge 248, which is connected to the fixed network architecture to facilitate seamless communication via the network. The AI module 212 receives data from the NEF 210 via the N33 interface. The AI module 212, which is combined with a de-jitter module, applies AI and machine learning algorithms to optimize the performance of the network. This complex network design ensures reliable, low-latency, high-throughput communication required for industrial applications. By continuously monitoring and adjusting network parameters based on real-time conditions, the AI module 212 provides robust and efficient communication to support the operation of time-sensitive applications in industrial environments.
[0078] refer to Figure 3A and Figure 3B , respectively, illustrate exemplary descriptions of time signals 300A and 300B according to embodiments of the present disclosure. As shown in time signal 300A, poor synchronization at T6 can cause the CNC to notify the corresponding device to reduce the packet size to avoid conflicts. In addition, as shown in the figure, accurate synchronization at T2 can cause the CNC to notify the corresponding device to use the maximum packet size. Similarly, as shown in time signal 300B, accurate synchronization at all time intervals can cause the CNC to notify the corresponding device to use a regular packet size.
[0079] refer to Figure 4 , illustrating a time synchronization scheme 400 utilizing artificial intelligence and machine learning (AI / ML) in the context of a network including an intermediate mobile network according to an embodiment of the present disclosure. The time synchronization scheme 400 applied in TSN 402 ensures improved synchronization accuracy and overall network performance. In the time synchronization scheme 400, there are two main nodes, a senior master node 404 and a slave node 406. The senior master node 404 is the authoritative time source for the network, while the slave node 406 aligns its clock with the senior master node 404. The communication and synchronization between node 404 and node 406 take into account the transmission delay represented by the formula ((T2-T1)+(T4-T3)) / 2. During training, AI / ML discovers the additional processing delays inherent in 5G networks and identifies asymmetries in UP and DL communication delays. AI / ML reports the SINR in the device, which is a factor that can increase transmission delays. TSN 402 uses Automatic Weight Calculation (AWC) in conjunction with sensors such as cameras to detect new obstacles that may block the signal, which can affect network performance and time synchronization.
[0080] The time synchronization scheme 400 also introduces the dwell time formula Ti=a.Di+b.Mi+c.Bi. In this formula, "Di" represents the distance of node i, which has a corresponding weight factor "a" calculated by the AI algorithm. "Mi" represents the mobility of node i, which has an associated weight factor "b". "Bi" represents signal blocking, which has an associated weight factor "c". The time synchronization scheme 400 uses a detector for training the AI supervised learning algorithm. The detector is equipped with a GPS as a reference time and changes the Di value by moving, so that the parameter "a" can be calculated. The detector maintains a fixed speed Mi to calculate the parameter "b", and positions itself behind an obstacle to measure Bi, thereby calculating the parameter "c".
[0081] refer to Figure 5 , shows a transmission management mechanism 500 in TSN according to an embodiment of the present disclosure, which focuses on TSN controllers including CNC 502 and CUC 504, and their interactions with various network components. In the transmission management mechanism 500, a mobile network 506 is used, and a TSN switch 508 in the form of a bridge or router is provided within the mobile network 506. The mobile device 510 acts as a "speaker" for sending data, and the fixed device 512 acts as a "listener" for receiving data. The mobile device 510 sends a signal (T3) to the TSN switch 508 within the mobile network 506. The signal then reaches the fixed network framework via the TSN switch 508 and finally reaches the fixed device 512. The TSN controller, in particular the CNC 502 and the CUC 504, supervises the communication process. The CNC 502 coordinates the mobile network 506 and coordinates the data flow, while the CUC 504 is responsible for user configuration settings. In the mobile device 510, a transmission (TX) time slot is specified for sending data packets. Similarly, in fixed device 512, receive (RX) time slots are allocated for incoming data.
[0082] refer to Figure 6, showing a flowchart listing the steps involved in a method 600 for optimizing the performance of an indoor industrial network according to an embodiment of the present disclosure. The method 600 includes, at step 602: in an AI module on the network side of a TSN bridge of an indoor industrial network, receiving SINR and delay information from a synchronization agent of a mobile router on the device side of the TSN bridge. The method 600 includes, at step 604: receiving one or more first parameters of an indoor industrial network from one or more first sensors. The method 600 includes, at step 606: receiving obstacle data from one or more cameras. The method 600 includes, at step 608: determining the degradation of the time accuracy of a clock in one or more mobile devices based on at least one of the received SINR, one or more first parameters, and obstacle data. The method 600 includes, at step 610: providing information of the determined degradation of time accuracy to a CNC and a CUC of a TSN controller of an indoor industrial network. The method 600 includes, at step 612: instructing the CNC and the CUC to adjust the size of a data packet to be sent between a mobile device and one or more fixed devices. The method 600 includes, at step 614, instructing the UPF synchronization agent to send the time information to one or more mobile devices to recalculate the time information and adjust the clock.
[0083] refer to Figure 7 , showing a flowchart listing the steps involved in a method 700 for predicting network problems in an indoor industrial network according to an embodiment of the present disclosure. Method 700 includes at step 702: collecting network data, the network data including: at least one of network performance data, sensor data, and device data. Method 700 includes at step 704: training an AI model based on the collected network data. Method 700 includes at step 706: deploying the trained model in a TSN bridge. Method 700 includes at step 708: receiving real-time data from network elements and sensors. Method 700 includes at step 710: using the trained model to predict network problems. Method 700 includes at step 712: adjusting time information sent to mobile devices to improve synchronization, and evaluating the performance of the model.
Claims
1. A computer-implemented method for optimizing indoor industrial network performance, the method comprising: In an artificial intelligence AI module on a network side of a time-sensitive network TSN bridge of the indoor industrial network, a signal to interference plus noise ratio SINR and delay information are received from a synchronization agent of a mobile router on a device side of the TSN bridge; receiving one or more first parameters of the indoor industrial network from one or more first sensors; receiving obstacle data from one or more cameras; determining a degradation of time accuracy of a clock in one or more mobile devices based on at least one of the received SINR, the one or more first parameters, and the obstacle data; Providing information on the determined time accuracy degradation to a central network controller CNC and a central user configuration CUC of a TSN controller of the indoor industrial network; instructing the CNC and the CUC to adjust the size of a data packet to be sent between a mobile device and one or more fixed devices; as well as Instruct the user plane function UPF synchronization agent to send time information to the one or more mobile devices to recalculate the time information and adjust the clock.
2. The method according to claim 1, wherein: The method further includes: measuring one or more second parameters of the environment of the indoor industrial network by one or more second sensors, the one or more second parameters including temperature and humidity.
3. The method according to claim 1 or 2, wherein: The method further includes configuring the CNC to set a TSN frame transmission schedule and preference, and configuring the CUC to set a TSN frame transmission schedule in a terminal device.
4. A method according to any one of the preceding claims, wherein: The adjustment includes: if the synchronization is not accurate, then reducing the size of the data packet; or if the synchronization is accurate, then maximizing the size of the data packet.
5. A method according to any one of the preceding claims, wherein: The method further includes receiving a reference time from the one or more mobile devices having GPS access permission to calculate a distance between the one or more mobile devices having GPS access permission and the one or more fixed devices.
6. A method according to any one of the preceding claims, wherein: The method further includes receiving, in a debounce module, a highest dwell time from the AI module.
7. A time-sensitive network (TSN) bridge for optimizing indoor industrial network performance, comprising a synchronization agent of a mobile router on the device side of the TSN bridge, and an artificial intelligence (AI) module on the network side of the TSN bridge, wherein: The AI module is configured to use machine learning algorithms to: receiving signal to interference plus noise ratio SINR and delay information from a synchronization agent of a mobile router on the device side of the TSN bridge; receiving and processing parameters of the indoor industrial network, and receiving and processing obstacle data from one or more cameras; determining a degradation in time accuracy of a clock in one or more mobile devices of the indoor industrial network; as well as The TSN controller of the indoor industrial network communicates with the central network controller CNC and the central user configuration CUC.
8. An indoor industrial network, comprising a time-sensitive network (TSN) bridge and an artificial intelligence (AI) module integrated into the TSN bridge, wherein: The AI module is configured to optimize the performance of the indoor industrial network by using machine learning algorithms to: analyze data from synchronization agents, sensors, and cameras of mobile routers; discover and report additional processing delays in the network, uplink delays and downlink delays of asymmetric communications, signal-to-interference-plus-noise ratios (SINRs), and obstacles that block signals and increase delays; predict potential problems in the network; and take proactive measures to resolve the problems; And interact with other network elements such as the user plane function UPF synchronization agent, central user configuration CUC and central network controller CNC to optimize the performance of time-sensitive applications.
9. A computer-implemented method for predicting network problems in an indoor industrial network to optimize the performance of the indoor industrial network, comprising: Collecting network data, wherein the network data includes at least one of network performance data, sensor data, and device data; Train artificial intelligence (AI) models based on collected network data; Deploy the trained model in a time-sensitive network (TSN) bridge; receiving real-time data from network elements and sensors; Use trained models to predict network issues; as well as The time information sent to the mobile devices is adjusted to improve synchronization, and the performance of the model is evaluated.
10. A computer program for optimizing the performance of an indoor industrial network, comprising instructions which, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 7.
Citation Information
Cited By
PTP network performance monitoring method and system based on big data analysis, and medium
CN120110951A
