Low-delay industrial internet of things data transmission system based on 5G network

Abstract

The invention relates to the technical field of industrial Internet of Things, and discloses a low-delay industrial Internet of Things data transmission system based on a 5G network. The system comprises a plurality of functional modules such as a data acquisition module, a data preprocessing module, a network transmission optimization module and a congestion control module. The data acquisition module acquires a heterogeneous data set, the preprocessing module performs standardization processing, the network transmission optimization module selects an optimal channel and node, and the congestion control module predicts flow and adjusts a transmission strategy. In addition, the system also has the functions of security encryption, heterogeneous data synchronization, dynamic spectrum allocation and the like. Through cooperative work of the modules, the system realizes low-delay and high-reliability transmission of industrial Internet of Things data, effectively processes heterogeneous data, guarantees data security, improves spectrum utilization efficiency and energy efficiency, enhances fault tolerance and compatibility of the system, and is suitable for various industrial scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial Internet of Things, and specifically to a low-latency industrial Internet of Things data transmission system based on a 5G network. Background Art

[0002] Against the backdrop of Industry 4.0, the Industrial Internet of Things (IIoT) is developing rapidly, enabling the interconnection of an increasing number of industrial devices. This trend is generating vast amounts of industrial data, which is crucial for optimizing industrial production, real-time monitoring and maintenance of equipment, and refined management of production processes. However, traditional industrial networks have numerous limitations in data transmission, making them unable to meet the growing demands of modern industry.

[0003] Traditional industrial networks, such as wired-based Industrial Ethernet and some low-bandwidth wireless communication networks, often experience excessively high transmission delays when handling massive amounts of data due to network bandwidth limitations and inadequate data processing and forwarding mechanisms. For example, in automated production lines, the transmission of real-time monitoring data on equipment operating status and control instructions requires extremely high timeliness. Excessive transmission delays can prevent equipment failures from being detected and addressed promptly, or prevent control instructions from reaching equipment in a timely manner. This can impact the continuity and stability of the entire production process, reduce production efficiency, and even cause production accidents.

[0004] Heterogeneous data processing is also a major challenge for traditional industrial networks. Industrial production processes involve a wide variety of sensors and devices, each generating data in different formats and using different communication protocols, resulting in a massive amount of heterogeneous data. Sensors like temperature sensors, pressure sensors, and image sensors each output analog or digital signals, and may use different communication protocols. This creates significant challenges in data collection, transmission, and processing, making it difficult to effectively integrate and coordinate data across different devices, increasing system complexity and maintenance costs.

[0005] Network congestion is also a significant issue. With the increasing number of industrial IoT devices, network traffic is growing exponentially. With limited network resources, network congestion is a common occurrence. Once congestion occurs, data transmission delays are exacerbated and the probability of packet loss increases significantly, severely impacting the quality and reliability of data transmission. For example, in the network environment of a large factory campus, where multiple workshops are simultaneously transmitting data, network congestion often prevents the timely transmission of critical production data, impacting the timeliness and accuracy of production decisions.

[0006] Furthermore, ensuring data security and integrity is crucial in the industrial sector. Industrial data contains crucial information such as core production technologies, trade secrets, and production safety. Leakage or tampering can lead to significant economic losses and security risks. Traditional encryption and verification technologies are increasingly struggling to cope with increasingly sophisticated cyberattacks and are unable to meet the high security requirements of industrial data.

[0007] Traditional industrial networks also have shortcomings in spectrum utilization and energy efficiency. Spectrum resources are limited, and traditional spectrum allocation methods are often inflexible and unable to dynamically adjust to actual network needs, resulting in low spectrum utilization. Furthermore, energy consumption by industrial equipment is a significant issue, especially in distributed Industrial IoT scenarios where large numbers of devices operate for extended periods of time. Reducing energy consumption while ensuring data transmission performance is a pressing issue. Summary of the Invention

[0008] The purpose of the present invention is to provide a low-latency industrial Internet of Things data transmission system based on 5G network to solve the problems raised in the above background technology.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a low-latency industrial Internet of Things data transmission system based on a 5G network, the system comprising:

[0010] The data acquisition module is used to collect raw data streams in real time through distributed edge computing nodes to generate spatiotemporally correlated heterogeneous data sets;

[0011] A data preprocessing module, used to preprocess the heterogeneous data set to generate a standardized data packet;

[0012] The network transmission optimization module is used to build an end-to-end transmission link quality evaluation matrix based on a multi-path dynamic routing algorithm and an adaptive bandwidth allocation strategy, and select the optimal 5G channel and edge server node;

[0013] The congestion control module uses a deep reinforcement learning model to predict network traffic fluctuations, dynamically adjust the packet priority queue and retransmission timeout threshold, and generate a low-latency transmission strategy.

[0014] Preferably, the data preprocessing module includes:

[0015] The heterogeneous data sets are segmented in the time domain and the improved discrete wavelet transform algorithm is used to separate the high-frequency transient noise and the low-frequency trend component.

[0016] Design a dynamic quantization encoder based on information entropy to adaptively allocate bits to low-frequency components according to signal energy distribution;

[0017] The high-frequency components are compressed and reconstructed through sparse matrix coding technology to generate data packets with reduced dimension.

[0018] Preferably, the network transmission optimization module includes:

[0019] Construct a multi-dimensional evaluation index including channel bandwidth, bit error rate and node load, and use fuzzy logic algorithm to calculate the comprehensive transmission weight of each link;

[0020] Generate a set of k shortest paths based on the improved Dijkstra dynamic routing algorithm, and update edge weight parameters based on real-time network topology;

[0021] The asymmetric Nash bargaining model is used to negotiate the bandwidth allocation ratio between edge servers to maximize channel utilization.

[0022] Preferably, the congestion control module includes:

[0023] A dual-time-scale LSTM network is used to model historical traffic sequences and burst traffic patterns, and predict network load status within future time windows.

[0024] Design a priority weight function to dynamically adjust queue scheduling order based on packet delay sensitivity and service type;

[0025] The joint parameters of the number of retransmissions and the timeout window are optimized based on the gradient descent method to minimize the end-to-end transmission delay.

[0026] Preferably, the system further comprises:

[0027] The secure encryption and integrity verification module is used to perform end-to-end encryption and integrity signing of data packets through the quantum key distribution protocol and lightweight hash tree structure to generate an attack-resistant transmission channel. The specific methods include:

[0028] Establish a dynamically updated shared key pool between edge nodes through the BB84 quantum key distribution protocol;

[0029] Lattice-based post-quantum encryption algorithm is used to encrypt data packets in blocks and generate an aggregate hash value based on the Merkle tree;

[0030] Utilize zero-knowledge proof protocol to verify data integrity.

[0031] Preferably, the system further comprises:

[0032] The heterogeneous data synchronization module detects clock offsets through a sliding window mechanism and uses a Kalman filter to compensate for timing errors caused by transmission jitter. It generates globally consistent timing tags and embeds them into data packet headers to construct a spatiotemporally aligned data stream.

[0033] Preferably, the system further comprises:

[0034] Dynamic spectrum allocation module, used to collaboratively train spectrum occupancy prediction models based on a federated learning framework;

[0035] Adopting the covariance matrix adaptive evolution strategy to search for the optimal subcarrier allocation scheme and avoid co-channel interference;

[0036] Generate a spectrum utilization efficiency evaluation report and feed it back to the base station controller to trigger dynamic frequency band switching instructions.

[0037] Preferably, the system further comprises:

[0038] Fault-tolerant and redundant transmission module, used to build a multi-copy sharding encoding mechanism based on algebraic geometry codes;

[0039] Mapping data shards to physically dispersed edge storage nodes through topology-aware algorithms;

[0040] Design a shard reorganization strategy based on a voting mechanism to restore complete data when some nodes fail.

[0041] Preferably, the system further comprises:

[0042] Energy efficiency optimization module, used to establish a nonlinear relationship model between device power consumption and transmission rate;

[0043] The Lagrange multiplier method is used to solve the maximum data transmission rate optimization problem under energy consumption constraints;

[0044] Dynamically adjust the CPU frequency and RF power of edge nodes to generate transmission configuration parameters with optimal energy efficiency.

[0045] Preferably, the system further comprises:

[0046] The protocol conversion and compatibility module is used to define semantic middleware for industrial Modbus and OPC UA protocols; it realizes the convergence and conversion of multi-protocol data streams through dynamic port mapping technology.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] To address data transmission latency, the system utilizes a multi-path dynamic routing algorithm and adaptive bandwidth allocation strategy, combined with an end-to-end transmission link quality assessment matrix, to precisely select the optimal 5G channel and edge server node. In complex industrial environments, where network conditions are constantly changing, the system monitors key metrics such as channel bandwidth, bit error rate, and node load in real time. Using fuzzy logic, it calculates the combined transmission weight of each link and rapidly generates and adjusts the shortest path based on a modified Dijkstra dynamic routing algorithm. Furthermore, the congestion control module employs a deep reinforcement learning model to predict network traffic fluctuations and dynamically adjust packet priority queues and retransmission timeout thresholds. This series of measures effectively mitigates the impact of network congestion on data transmission, significantly reducing end-to-end transmission latency and ensuring the real-time nature of industrial data. For example, in smart grid power dispatching systems, this system can minimize transmission latency for real-time monitoring data such as grid voltage and current, as well as control commands, ensuring stable operation and preventing power outages caused by data transmission delays.

[0049] In heterogeneous data processing, after the data acquisition module collects temporally and spatially correlated heterogeneous data sets, the data preprocessing module utilizes a series of advanced technologies. Through time-domain segmentation and an improved discrete wavelet transform algorithm, high-frequency transient noise and low-frequency trend components can be effectively separated, improving data accuracy and stability. A dynamic quantization encoder based on information entropy adaptively allocates bits to low-frequency components based on signal energy distribution, reducing data volume while ensuring accuracy. Sparse matrix coding technology performs compressed sensing reconstruction on high-frequency components, further reducing data dimensionality. This enables unified processing and transmission of different types of industrial data, facilitating subsequent data fusion and analysis, and improving the compatibility and scalability of the entire Industrial Internet of Things system. For example, in an automotive manufacturing plant, heterogeneous data from different sensors, after processing, can provide a more comprehensive and accurate basis for optimizing the production process.

[0050] The system's secure encryption and integrity verification modules significantly enhance data security. The quantum key distribution protocol establishes a dynamically updated shared key pool between edge nodes, the Lattice-based post-quantum encryption algorithm performs block-by-block encryption of data packets, and Merkle-tree-based aggregate hash values ​​and zero-knowledge proof protocols verify data integrity. This suite of technologies effectively protects against various cyberattacks, including the potential threat of quantum computing attacks. In industries with extremely high data security requirements, such as finance and the military, these security measures ensure confidentiality, integrity, and non-repudiation of data during transmission and storage, preventing data leakage and tampering and safeguarding core corporate interests and national security.

[0051] The dynamic spectrum allocation module uses a federated learning framework to collaboratively train a spectrum occupancy prediction model, enabling accurate predictions based on real-time spectrum usage. It employs a covariance matrix adaptive evolutionary strategy to search for the optimal subcarrier allocation scheme, effectively mitigating co-channel interference and improving spectrum efficiency. The generated spectrum efficiency assessment report is fed back to the base station controller, triggering dynamic frequency band switching instructions and achieving dynamic optimization of spectrum resources. This not only improves data transmission stability and reliability, but also supports the access and data transmission of more industrial devices within limited spectrum resources, providing strong support for the large-scale application of the Industrial Internet of Things.

[0052] Regarding energy efficiency, the energy efficiency optimization module models the nonlinear relationship between device power consumption and transmission rate. It uses the Lagrange multiplier method to solve the maximum data transmission rate optimization problem under energy constraints, dynamically adjusting the CPU frequency and RF power of edge nodes. This allows industrial equipment to rationally adjust power consumption based on actual needs while ensuring data transmission performance, reducing energy consumption. In large-scale industrial IoT deployments, where large numbers of devices operate continuously, this energy efficiency optimization measure can significantly reduce operating costs for enterprises while also aligning with the concept of sustainable development.

[0053] In addition, the fault-tolerant and redundant transmission module constructs a multi-copy shard encoding mechanism based on algebraic geometry codes, maps data shards to physically dispersed edge storage nodes, and designs a shard reorganization strategy based on a voting mechanism. Even in the event of partial node failure, complete data can still be restored, improving the system's fault tolerance and data reliability. The protocol conversion and compatibility module defines semantic middleware for multiple protocols such as industrial Modbus and OPC UA. It achieves the convergence and conversion of multi-protocol data streams through dynamic port mapping technology, solving the communication difficulties caused by different protocols between industrial devices and enhancing the system's compatibility and integration. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a working principle diagram of the low-latency industrial Internet of Things data transmission system based on 5G network according to the present invention;

[0055] Figure 2 This is the working principle diagram of the data preprocessing module;

[0056] Figure 3 Schematic diagram of the working principle of the module for multi-network transmission optimization;

[0057] Figure 4 This is a working principle diagram of the congestion control module. DETAILED DESCRIPTION

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0059] See also Figure 1-Figure 4 The present invention relates to a low-latency industrial Internet of Things data transmission system based on a 5G network, and its overall implementation scheme is as follows:

[0060] The data acquisition module collects raw data streams in real time through distributed edge computing nodes. In industrial production scenarios, various sensors are distributed in diverse locations, such as key components of production equipment and environmental monitoring points in production workshops. These sensors continuously generate large amounts of data, and the data acquisition module is responsible for collecting this data to form heterogeneous datasets with temporal and spatial correlations. For example, in an automobile manufacturing plant, the data collected by temperature sensors, pressure sensors, and position sensors on equipment not only contains different types of information but is also related to specific time and spatial locations.

[0061] The data preprocessing module preprocesses the heterogeneous dataset to generate standardized data packets. This module first segments the heterogeneous dataset in the time domain and uses a modified discrete wavelet transform algorithm to separate high-frequency transient noise from low-frequency trend components. Next, a dynamic quantization encoder based on information entropy is designed to adaptively allocate bits to low-frequency components based on the signal energy distribution. Finally, sparse matrix coding technology is used to perform compressed sensing reconstruction of the high-frequency components, generating dimensionality-reduced data packets, thereby reducing data transmission volume and processing burden.

[0062] The network transmission optimization module, based on a multi-path dynamic routing algorithm and an adaptive bandwidth allocation strategy, constructs an end-to-end transmission link quality evaluation matrix to select the optimal 5G channel and edge server nodes. It constructs multi-dimensional evaluation metrics, including channel bandwidth, bit error rate, and node load, and uses a fuzzy logic algorithm to calculate the comprehensive transmission weight of each link. It generates a set of k shortest paths based on an improved Dijkstra dynamic routing algorithm and updates edge weight parameters based on real-time network topology. It also utilizes an asymmetric Nash bargaining model to negotiate bandwidth allocation between edge servers, maximizing channel utilization and ensuring efficient and stable data transmission.

[0063] The congestion control module uses a deep reinforcement learning model to predict network traffic fluctuations, dynamically adjust packet priority queues and retransmission timeout thresholds, and generate a low-latency transmission strategy. This module employs a dual-time-scale LSTM network to model historical traffic sequences and burst traffic patterns, predicting network load conditions within future time windows. A priority weighting function is designed to dynamically adjust queue scheduling based on packet delay sensitivity and service type. A gradient descent algorithm is used to optimize the combined parameters of the retransmission count and timeout window to minimize end-to-end transmission latency and ensure timely data transmission.

[0064] The implementation of the present invention will be further described below with reference to Examples 1 to 6.

[0065] Example 1:

[0066] In the data preprocessing module, time-domain segmentation is performed based on the characteristics of industrial data, selecting appropriate time intervals to partition heterogeneous data sets. For example, power system monitoring data exhibits periodicity, so segmentation based on the duration of a single cycle can be used to better analyze data characteristics. An improved discrete wavelet transform (DWT) algorithm is employed. This algorithm, based on the traditional DWT, optimizes the selection of wavelet basis functions, making it more suitable for separating high-frequency transient noise from low-frequency trend components in industrial data.

[0067] When designing an entropy-based dynamic quantization encoder, the entropy of low-frequency components is first calculated. This entropy reflects signal uncertainty. Fewer bits are allocated to components with higher signal energy and lower entropy, while more bits are allocated to components with lower signal energy and higher entropy. This adaptive bit allocation reduces data transmission while ensuring accuracy. When using sparse matrix coding technology for compressed sensing reconstruction of high-frequency components, a sparse representation is first created. Based on compressed sensing theory, these components are sampled using an appropriate measurement matrix. Finally, an optimization algorithm is used to reconstruct the high-frequency components, generating a dimensionality-reduced data packet.

[0068] When constructing multi-dimensional evaluation indicators in the network transmission optimization module, channel bandwidth refers to the maximum data transmission rate of a 5G channel, measured in bits per second; bit error rate refers to the probability of an erroneous symbol occurring during transmission; and node load indicates how busy edge server nodes are in processing data, measured through indicators such as the node's CPU utilization and memory utilization. Fuzzy logic algorithms are used to fuzzify these indicators, setting different fuzzy sets and membership functions, and calculating the comprehensive transmission weight of each link through fuzzy reasoning.

[0069] Based on the improved Dijkstra dynamic routing algorithm, the algorithm introduces a dynamic update mechanism when generating a set of k shortest paths, building on the traditional Dijkstra algorithm. When the real-time network topology changes, such as when a link fails or is overloaded, edge weight parameters are promptly updated and the shortest path set is recalculated to ensure that data is transmitted along the optimal path. When negotiating bandwidth allocation between edge servers using an asymmetric Nash bargaining model, the model takes into account factors such as each edge server's processing power, current load, and bandwidth demand. Through iterative calculations, the optimal bandwidth allocation solution is found to maximize channel utilization.

[0070] In actual application scenarios, such as smart factories, a large number of industrial equipment generate data in the workshop. After the data preprocessing module processes these heterogeneous data, the network transmission optimization module selects the optimal 5G channel and edge server node based on multi-dimensional evaluation indicators and algorithms to ensure that the data is quickly and stably transmitted to the data center for analysis and processing.

[0071] Example 2:

[0072] The congestion control module uses a dual-time-scale LSTM network to model historical traffic sequences and burst traffic patterns. This dual-time-scale LSTM network consists of two LSTM units with different time scales: the short-time-scale LSTM unit is used to capture short-term fluctuations in network traffic, and the long-time-scale LSTM unit is used to learn long-term trends in network traffic. For example, in an Internet Data Center (IDC) network, traffic varies significantly between peak daytime hours and low nighttime hours. The long-time-scale LSTM unit can learn these daily traffic trends. However, during peak hours, traffic can also experience short-term, sudden fluctuations, which the short-time-scale LSTM unit can capture.

[0073] By learning from historical traffic data, the dual-time-scale LSTM network can predict network load status within a future time window. This time window is set based on actual application requirements. For example, in industrial control scenarios with high real-time requirements, the time window can be set to 1-5 minutes. The prediction results are used to subsequently adjust packet priority queues and retransmission timeout thresholds.

[0074] When designing the priority weight function, two factors are considered: packet delay sensitivity and service type. Highly delay-sensitive packets, such as industrial control instruction packets, are assigned a higher priority weight; low-delay-sensitive packets, such as non-real-time monitoring data, are assigned a lower priority weight. Furthermore, different service types have different priorities; for example, data services related to critical production processes have a higher priority than ordinary equipment maintenance data services. The priority weight function can be expressed as: W = α × S + β × T, where W represents the packet's priority weight, and α and β are weight coefficients that are adjusted based on actual conditions. S represents the packet's delay sensitivity, ranging from 0 to 1, with larger values ​​indicating higher delay sensitivity. T represents the service type priority, set to different integer values ​​depending on the service type, with larger values ​​indicating higher priority.

[0075] The combined parameters of the number of retransmissions and the timeout window are optimized based on the gradient descent method. During data transmission, if a data packet is sent but no confirmation information is received within the specified time, it needs to be retransmitted. Too many retransmissions will increase the network burden, while too few may cause data loss; setting the timeout window too large may increase data transmission delays, while setting it too small may cause unnecessary retransmissions. Using the gradient descent method, the number of retransmissions N and the timeout window T are continuously adjusted. w The goal is to minimize the end-to-end transmission delay D. Take D as the loss function and calculate its loss function with respect to N and T w The gradient of the data transmission is calculated and the parameter value is adjusted according to the gradient direction until the convergence condition is reached. In actual network environments, by continuously optimizing these two parameters, the efficiency and reliability of data transmission can be effectively improved.

[0076] Example 3:

[0077] The secure encryption and integrity verification module uses the quantum key distribution protocol and lightweight hash tree structure to perform end-to-end encryption and integrity signing on data packets, generating an attack-resistant transmission channel.

[0078] A dynamically updated shared key pool is established between edge nodes using the BB84 quantum key distribution protocol. Based on the fundamental principles of quantum mechanics, the BB84 protocol uses the polarization state of photons to encode information. At the transmitter, the sender randomly selects one of four different polarization states (horizontal, vertical, +45°, -45°) to encode a 0 or 1 bit of information and sends it to the receiver via a quantum channel. The receiver randomly selects one of two measurement bases (horizontal-vertical basis and +45°--45° basis) to measure the polarization state of the received photon. The measurement result is correct only when the sender and receiver select the same measurement basis. By publicly comparing the measurement basis selection information, both parties can filter out the correct measurement result and form a shared key. To ensure the security and freshness of the key, the shared key pool is regularly and dynamically updated.

[0079] Lattice-based post-quantum encryption is used to encrypt data packets in blocks, generating an aggregate hash value based on a Merkle tree. Lattice-based post-quantum encryption is resistant to quantum computing attacks and is suitable for network environments that may face quantum computing threats in the future. Data packets are divided into several fixed-size blocks, each of which is encrypted using a Lattice-based post-quantum encryption algorithm, using a key from a shared key pool. To verify data integrity, an aggregate hash value is generated using a Merkle tree structure. A Merkle tree is a binary tree where each leaf node is the hash value of a data block, and non-leaf nodes are the hash values ​​of their two child nodes. Ultimately, the hash value of the root node is the aggregate hash value of the entire data packet.

[0080] Data integrity is verified using a zero-knowledge proof protocol. This protocol allows the prover to prove a proposition is correct without providing any additional information to the verifier. To verify data integrity, the sender uses the zero-knowledge proof protocol to prove to the receiver that the data packet has not been tampered with during transmission. The receiver verifies the data integrity based on the received data packet, the aggregate hash value, and the zero-knowledge proof information. If the verification passes, the data packet is secure and reliable during transmission; otherwise, the packet may have been tampered with and needs to be retransmitted. In scenarios with extremely high security requirements, such as financial transaction data transmission, this secure encryption and integrity verification mechanism can effectively ensure data security and integrity.

[0081] Example 4:

[0082] This embodiment is described in detail with reference to the heterogeneous data synchronization module and the spectrum dynamic allocation module.

[0083] The heterogeneous data synchronization module uses a sliding window mechanism to detect clock offsets. In an industrial IoT environment, different sensors and devices may have different clock sources, leading to clock asynchrony during data collection and transmission. The sliding window mechanism uses a certain time interval as a window and compares the timestamps of data sent by different devices within the window. Assume that device A and device B send data in the same time window, and the timestamp of device A is t A , the timestamp of device B is t B , clock offset Δt = t A -t B By continuously updating the sliding window, the changes in clock offset are monitored in real time.

[0084] Combined with the Kalman filter to compensate for the timing error caused by transmission jitter. Transmission jitter will lead to instability in the arrival time of data. The Kalman filter is an optimal linear estimator that can predict and correct the timing error based on the system's state equation and observation equation. Assume that the state equation of the system is x k =Ax k-1 +Bu k +w k , the observation equation is y k =Cx k +v k , where x k is the state of the system at time k, A is the state transfer matrix, B is the control matrix, u k is the control input, w k is the process noise, y k is the observation value, C is the observation matrix, v k Observation noise. During data transmission, clock offset and transmission jitter are used as inputs. The Kalman filter performs iterative calculations to obtain compensated timestamps. This generates globally consistent timing labels and embeds them into the packet header, constructing a spatiotemporally aligned data stream.

[0085] The dynamic spectrum allocation module collaboratively trains the spectrum occupancy prediction model based on a federated learning framework. This framework allows edge nodes to collaboratively train the model without sharing raw data. Each edge node trains its own spectrum occupancy prediction model based on locally collected spectrum usage data. Each edge node then uploads the model parameters to a central server, which aggregates and updates them and distributes the updated model parameters to each edge node. This approach continuously optimizes the performance of the spectrum occupancy prediction model.

[0086] The covariance matrix adaptive evolutionary strategy is used to search for the optimal subcarrier allocation scheme to avoid co-channel interference. The covariance matrix adaptive evolutionary strategy is an optimization algorithm that gradually finds the optimal solution by continuously adjusting the covariance matrix of the search space. In spectrum allocation, the subcarrier allocation problem is transformed into an optimization problem, aiming to maximize spectrum efficiency while considering constraints such as co-channel interference. Based on the current spectrum occupancy and interference information, the algorithm searches for the optimal subcarrier allocation scheme, minimizing mutual interference between different devices using the same frequency band.

[0087] Generates a spectrum utilization efficiency assessment report and feeds it back to the base station controller, triggering dynamic frequency band switching instructions. The dynamic spectrum allocation module regularly evaluates spectrum utilization efficiency, using metrics such as spectrum utilization and interference level. This assessment report is fed back to the base station controller. When spectrum utilization efficiency is low or interference is severe, the base station controller triggers dynamic frequency band switching instructions based on the report content, adjusting the frequency band used by the device to improve spectrum utilization efficiency and data transmission quality. In intelligent transportation systems, data transmission between vehicles and roadside equipment requires efficient spectrum allocation. The heterogeneous data synchronization module and the dynamic spectrum allocation module work together to ensure stable and accurate data transmission.

[0088] Example 5:

[0089] This embodiment describes in detail the specific implementation of the fault-tolerant and redundant transmission module and the energy efficiency optimization module.

[0090] The fault-tolerance and redundant transmission module implements a multi-replica shard encoding mechanism based on algebraic geometric codes. Algebraic geometric codes are error-correcting codes with strong error correction capabilities. The original data is divided into multiple shards, each of which is encoded to generate multiple redundant copies. These copies and the original shards are mapped to physically dispersed edge storage nodes using a topology-aware algorithm. Based on the network topology, the topology-aware algorithm selects distant and highly reliable edge storage nodes to store data shards and replicas, thereby improving data fault tolerance.

[0091] A shard reorganization strategy based on a voting mechanism is designed to restore complete data when some nodes fail. When some edge storage nodes fail or data is lost, data shards and replicas are obtained from the remaining nodes. The voting mechanism votes on each data shard based on the number and credibility of data shards and replicas, and selects the shard with the most votes as the correct shard for reorganization. For example, if a data shard has five replicas, three of which are from different nodes and have consistent content, and the other two have different content, then the voting mechanism selects the shards corresponding to the three consistent replicas for reorganization, thereby restoring the complete data.

[0092] The energy efficiency optimization module models the nonlinear relationship between device power consumption and transmission rate. In industrial IoT devices, a complex nonlinear relationship exists between device power consumption and transmission rate. Through experimental testing and data analysis, a power consumption model P = f(R) was established, where P represents device power consumption, R represents transmission rate, and f is a nonlinear function. For example, for a certain model of wireless sensor, power consumption initially increases slowly as the transmission rate increases. Once the transmission rate exceeds a certain threshold, power consumption rises rapidly.

[0093] The Lagrange multiplier method is used to solve the maximum data transmission rate optimization problem under energy consumption constraints. Let the objective function be R (transmission rate) and the constraint condition be P≤P max (P max is the maximum power consumption allowed by the device), introduce the Lagrangian multiplier λ, and construct the Lagrangian function L=R+λ(P max -P). By taking the partial derivatives of the Lagrangian function and setting them to zero, the maximum transmission rate under energy constraints is determined. Based on the solution, the CPU frequency and RF power of the edge node are dynamically adjusted to generate the transmission configuration parameters with the best energy efficiency. In smart home systems, numerous smart devices need to ensure data transmission while reducing energy consumption. The combination of fault-tolerant and redundant transmission modules and energy efficiency optimization modules can effectively improve system reliability and energy efficiency.

[0094] Example 6:

[0095] The protocol conversion and compatibility module defines semantic middleware for the industrial Modbus and OPC UA protocols. Industrial Modbus and OPC UA are commonly used communication protocols in the industrial sector, but they differ in data formats and communication methods. The semantic middleware parses and converts these two protocols by defining a unified interface and data model. For example, the Modbus protocol uses a request-response model, with data read and written in registers; the OPC UA protocol, on the other hand, uses a publish-subscribe model, supporting more flexible data interaction. The semantic middleware parses Modbus request data and converts it into a data format that complies with the OPC UA protocol, and vice versa.

[0096] Dynamic port mapping technology enables the convergence and conversion of multi-protocol data streams. In industrial networks, different devices may communicate using different protocols. Dynamic port mapping dynamically assigns port numbers based on the device's protocol type and communication requirements. For example, when a device using the Modbus protocol needs to communicate with a server using the OPC UA protocol, dynamic port mapping maps the Modbus device's communication request to a specific port, which then exchanges data with the OPC UA server. During data transmission, dynamic port mapping performs protocol conversion to ensure correct data transmission between devices using different protocols.

[0097] In a production workshop containing a variety of industrial equipment, some use the Modbus protocol, while others use the OPC UA protocol. The Protocol Conversion and Compatibility Module, through semantic middleware and dynamic port mapping technology, enables data interconnection between devices using different protocols. This enables more efficient collaboration throughout the production process, improving the automation level and management efficiency of industrial production.

[0098] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0099] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A low-latency industrial IoT data transmission system based on 5G network, characterized by: The system comprises: The data acquisition module is used to collect raw data streams in real time through distributed edge computing nodes to generate spatiotemporally correlated heterogeneous data sets; A data preprocessing module, used to preprocess the heterogeneous data set to generate a standardized data packet; The network transmission optimization module is used to build an end-to-end transmission link quality evaluation matrix based on a multi-path dynamic routing algorithm and an adaptive bandwidth allocation strategy, and select the optimal 5G channel and edge server node; The congestion control module uses a deep reinforcement learning model to predict network traffic fluctuations, dynamically adjust the packet priority queue and retransmission timeout threshold, and generate a low-latency transmission strategy.

2. The low-latency industrial Internet of Things data transmission system based on 5G network according to claim 1, characterized in that: The data preprocessing module includes: The heterogeneous data sets are segmented in the time domain and the improved discrete wavelet transform algorithm is used to separate the high-frequency transient noise and the low-frequency trend component. Design a dynamic quantization encoder based on information entropy to adaptively allocate bits to low-frequency components according to signal energy distribution; The high-frequency components are compressed and reconstructed through sparse matrix coding technology to generate data packets with reduced dimension.

3. The low-latency industrial Internet of Things data transmission system based on 5G network according to claim 2, characterized in that: The network transmission optimization module includes: Construct a multi-dimensional evaluation index including channel bandwidth, bit error rate and node load, and use fuzzy logic algorithm to calculate the comprehensive transmission weight of each link; Generate a set of k shortest paths based on the improved Dijkstra dynamic routing algorithm, and update edge weight parameters based on real-time network topology; The asymmetric Nash bargaining model is used to negotiate the bandwidth allocation ratio between edge servers to maximize channel utilization.

4. The low-latency industrial Internet of Things data transmission system based on 5G network according to claim 1, characterized in that: The congestion control module includes: A dual-time-scale LSTM network is used to model historical traffic sequences and burst traffic patterns, and predict network load status within future time windows. Design a priority weight function to dynamically adjust queue scheduling order based on packet delay sensitivity and service type; The joint parameters of the number of retransmissions and the timeout window are optimized based on the gradient descent method to minimize the end-to-end transmission delay.

5. The low-latency industrial Internet of Things data transmission system based on 5G network according to claim 1, characterized in that: The system further comprises: The secure encryption and integrity verification module is used to perform end-to-end encryption and integrity signing of data packets through the quantum key distribution protocol and lightweight hash tree structure to generate an attack-resistant transmission channel. The specific methods include: Establish a dynamically updated shared key pool between edge nodes through the BB84 quantum key distribution protocol; Lattice-based post-quantum encryption algorithm is used to encrypt data packets in blocks and generate an aggregate hash value based on the Merkle tree; Utilize zero-knowledge proof protocol to verify data integrity.

6. The low-latency industrial Internet of Things data transmission system based on 5G network according to claim 1, characterized in that: The system further comprises: The heterogeneous data synchronization module detects clock offsets through a sliding window mechanism and uses a Kalman filter to compensate for timing errors caused by transmission jitter. It generates globally consistent timing tags and embeds them into data packet headers to construct a spatiotemporally aligned data stream.

7. The low-latency industrial Internet of Things data transmission system based on 5G network according to claim 1, characterized in that: The system further comprises: Dynamic spectrum allocation module, used to collaboratively train spectrum occupancy prediction models based on a federated learning framework; Adopting the covariance matrix adaptive evolution strategy to search for the optimal subcarrier allocation scheme and avoid co-channel interference; Generate a spectrum utilization efficiency evaluation report and feed it back to the base station controller to trigger dynamic frequency band switching instructions.

8. The low-latency industrial Internet of Things data transmission system based on 5G network according to claim 1, characterized in that: The system further comprises: Fault-tolerant and redundant transmission module, used to build a multi-copy sharding encoding mechanism based on algebraic geometry codes; Mapping data shards to physically dispersed edge storage nodes through topology-aware algorithms; Design a shard reorganization strategy based on a voting mechanism to restore complete data when some nodes fail.

9. The low-latency industrial Internet of Things data transmission system based on 5G network according to claim 1, characterized in that: The system further comprises: Energy efficiency optimization module, used to establish a nonlinear relationship model between device power consumption and transmission rate; The Lagrange multiplier method is used to solve the maximum data transmission rate optimization problem under energy consumption constraints; Dynamically adjust the CPU frequency and RF power of edge nodes to generate transmission configuration parameters with optimal energy efficiency.

10. The low-latency industrial Internet of Things data transmission system based on 5G network according to claim 1, characterized in that: The system further comprises: The protocol conversion and compatibility module is used to define semantic middleware for industrial Modbus and OPC UA protocols; it realizes the convergence and conversion of multi-protocol data streams through dynamic port mapping technology.

Images